{"meta":{"query_hash":"93e5f7b86fa9","filters":{"venue":"Scientific Data"},"cohort_total":591,"direct_labels_cover":48,"predictions_cover":591,"exported":591,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/93e5f7b86fa9","api":"https://metacan.xera.ac/api/v1/cohort?venue=Scientific+Data"},"results":[{"id":"W1124213845","doi":"10.1038/sdata.2015.42","title":"A spatially comprehensive, hydrometeorological data set for Mexico, the U.S., and Southern Canada 1950–2013","year":2015,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":424,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Bureau of Reclamation; Instituto Tecnológico de Sonora; National Science Foundation","keywords":"Hydrometeorology; Precipitation; Orographic lift; Orography; Evapotranspiration; Environmental science; Climatology; Meteorology; Geography; Geology; Ecology","score_opus":0.1784396269167633,"score_gpt":0.29465462329151954,"score_spread":0.11621499637475624,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1124213845","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.061059617,0.00035620728,0.00060235563,0.00012483961,0.000019116003,0.00008979118,0.9321843,0.00025565323,0.005308121],"genre_scores_gemma":[0.080952436,0.0006285561,0.0033497682,0.000055510816,0.000017300245,0.00023023564,0.91072786,0.0000562217,0.0039821784],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9997589,0.000007798423,0.000021156182,0.000055762623,0.000105373336,0.000050918883],"domain_scores_gemma":[0.9986331,0.000045910983,0.00023269317,0.000090892485,0.00085557863,0.00014171172],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00024615045,0.00049810886,0.00052550493,0.004855618,0.0013883659,0.00092997006,0.00087804784,0.00027259687,0.0051362794],"category_scores_gemma":[0.0012582246,0.00026366883,0.0002938417,0.010593793,0.00029821423,0.00033670905,0.0005394683,0.00047289746,0.0011075177],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025281182,0.00021518451,0.4257286,0.0009260533,0.00040387656,0.0004294777,0.0010454318,0.012988343,0.0019629805,0.004047103,0.48372728,0.06827284],"study_design_scores_gemma":[0.000044845736,0.000015797545,0.7390649,0.00016815429,0.000081109865,0.000118875,0.0006920279,0.0024382405,0.00092208554,0.0002589841,0.25614375,0.000051275216],"about_ca_topic_score_codex":0.97028995,"about_ca_topic_score_gemma":0.9795869,"teacher_disagreement_score":0.029710054,"about_ca_system_score_codex":0.008942974,"about_ca_system_score_gemma":0.016676662,"threshold_uncertainty_score":0.06488615},"labels":[],"label_agreement":null},{"id":"W1993975089","doi":"10.1038/sdata.2015.8","title":"A global database of lake surface temperatures collected by in situ and satellite methods from 1985–2009","year":2015,"lang":"en","type":"article","venue":"Scientific Data","topic":"Aquatic Ecosystems and Phytoplankton Dynamics","field":"Environmental Science","cited_by":240,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"International Institute for Sustainable Development; Ministry of the Environment, Conservation and Parks; York University","funders":"Division of Environmental Biology; Leibniz-Gemeinschaft; Natural Environment Research Council; U.S. Geological Survey; U.S. Fish and Wildlife Service; National Oceanic and Atmospheric Administration; Leibniz-Institut für Gewässerökologie und Binnenfischerei; National Aeronautics and Space Administration; Institut National de la Recherche Agronomique; Ministry of Business, Innovation and Employment; European Commission; Eesti Teadusfondi; U.S. Department of Agriculture; Regione del Veneto; Österreichischen Akademie der Wissenschaften; Vale Canada Limited; National Park Service; New York State Department of Environmental Conservation; National Science Foundation; University of Washington; Gordon and Betty Moore Foundation; Chinese Academy of Sciences; Russian Academy of Sciences; University of Nebraska-Lincoln; Sight Research UK; Government of Canada; Andrew W. Mellon Foundation; Irkutsk State University; Bundesministerium für Land- und Forstwirtschaft, Umwelt und Wasserwirtschaft; Cornell University Agricultural Experiment Station; Naturvårdsverket; York University; Waikato Regional Council","keywords":"Longitude; Environmental science; Latitude; Climate change; Surface water; Elevation (ballistics); Lake ecosystem; Satellite; Global change; Global warming; Database; Ecosystem; Climatology; Physical geography; Atmospheric sciences; Geography; Geology; Oceanography; Ecology","score_opus":0.033381128971521914,"score_gpt":0.30651883685899795,"score_spread":0.27313770788747604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993975089","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.109242395,0.000461355,0.0017458252,0.00006711137,0.000021745107,0.0001257232,0.8852081,0.0003063896,0.0028213754],"genre_scores_gemma":[0.06984906,0.00030270038,0.005307569,0.000050066625,0.00002147586,0.0003727451,0.9229976,0.00006168727,0.001037102],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99939466,0.000047279147,0.00015667405,0.00019425401,0.0001506724,0.000056398945],"domain_scores_gemma":[0.9982639,0.00014303123,0.0007097272,0.00025961926,0.0004929951,0.00013069836],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00048767834,0.00066676876,0.0006202656,0.004039263,0.0004581216,0.0006386845,0.000567535,0.00035103795,0.003341217],"category_scores_gemma":[0.0014266811,0.00036634397,0.00045553426,0.008868584,0.00017808645,0.00059545785,0.001131676,0.00044730926,0.0020824634],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006125461,0.00028660244,0.7110538,0.00279264,0.0009266237,0.000697411,0.0019182289,0.0076565896,0.01700916,0.0014899407,0.1603886,0.095167816],"study_design_scores_gemma":[0.00005260073,0.00005288049,0.88276196,0.00012689312,0.00013173348,0.00018956834,0.0003321575,0.0014522122,0.0019409013,0.00021307381,0.11270323,0.00004294035],"about_ca_topic_score_codex":0.027286254,"about_ca_topic_score_gemma":0.049075004,"teacher_disagreement_score":0.027286254,"about_ca_system_score_codex":0.0007121548,"about_ca_system_score_gemma":0.0018338365,"threshold_uncertainty_score":0.05425489},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2012105696","doi":"10.1038/sdata.2014.15","title":"Tree of Sex: A database of sexual systems","year":2014,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genetic and Clinical Aspects of Sex Determination and Chromosomal Abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":302,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary; University of British Columbia","funders":"National Institute of General Medical Sciences; Natural Environment Research Council; Sight Research UK; National Evolutionary Synthesis Center; National Science Foundation","keywords":"Biology; Phylogenetic tree; Context (archaeology); Evolutionary biology; Evolution of sexual reproduction; Ecology; Genetics","score_opus":0.034122245715847946,"score_gpt":0.28951405907009076,"score_spread":0.2553918133542428,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2012105696","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009511323,0.0022484346,0.025336163,0.00026452777,0.00006379881,0.00016890651,0.94256276,0.008640792,0.011203402],"genre_scores_gemma":[0.017816447,0.001656686,0.025984805,0.00010976702,0.000044586526,0.00024895513,0.9521786,0.00056388055,0.0013963019],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99918455,0.00012525219,0.00026867393,0.0001714213,0.00019236945,0.00005773445],"domain_scores_gemma":[0.99552983,0.0017907353,0.0007309729,0.0007859131,0.0007524086,0.0004101996],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012010438,0.0010589296,0.0012075055,0.010044614,0.0005696285,0.0026722397,0.0018865896,0.0010742132,0.022282623],"category_scores_gemma":[0.007556887,0.0005702594,0.0007992294,0.010828185,0.0003841106,0.0029625525,0.0017681202,0.0010538131,0.0129982475],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016085479,0.00028099172,0.049933616,0.009311665,0.0005375435,0.0018298557,0.0017187486,0.008120076,0.0153141245,0.039925277,0.49577612,0.37564346],"study_design_scores_gemma":[0.00019819316,0.0001280299,0.025901824,0.00058631226,0.00016121982,0.001743108,0.00036798944,0.005052651,0.0026788556,0.016992178,0.9460884,0.00010119213],"about_ca_topic_score_codex":0.0026609646,"about_ca_topic_score_gemma":0.0033125803,"teacher_disagreement_score":0.022282623,"about_ca_system_score_codex":0.0006074923,"about_ca_system_score_gemma":0.0021692747,"threshold_uncertainty_score":0.07454282},"labels":[],"label_agreement":null},{"id":"W2027222148","doi":"10.1038/sdata.2014.31","title":"A repository of assays to quantify 10,000 human proteins by SWATH-MS","year":2014,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":458,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Spinal Cord Injury BC","funders":"National Center for Research Resources; National Institute of General Medical Sciences; Suomen Kulttuurirahasto; Eidgenössische Technische Hochschule Zürich; National Human Genome Research Institute; SystemsX.ch; Kommission für Technologie und Innovation; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; European Molecular Biology Organization; National Institutes of Health; National Science Foundation","keywords":"UniProt; Computational biology; Proteome; Computer science; Human proteins; Mass spectrometry; Human proteome project; Quantitative proteomics; Tandem mass tag; Label-free quantification; Tandem mass spectrometry; Proteomics; Compendium; Data mining; Bioinformatics; Chemistry; Biology; Chromatography; Biochemistry","score_opus":0.0338103432655551,"score_gpt":0.31668508190209504,"score_spread":0.28287473863653995,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2027222148","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02280462,0.013465743,0.20777535,0.0007220551,0.00045345447,0.0015003751,0.6100251,0.11638885,0.026864482],"genre_scores_gemma":[0.014041622,0.004281967,0.11752588,0.00042247312,0.00013240616,0.0015834713,0.8481507,0.0055376873,0.008323795],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9961002,0.00051368785,0.00063161604,0.00079669384,0.0017235891,0.00023429266],"domain_scores_gemma":[0.99328554,0.0014107411,0.0010132408,0.002507195,0.0012910033,0.0004922888],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0048685884,0.004725159,0.0031410109,0.011575153,0.0012372861,0.0027932145,0.004644883,0.0017390215,0.020426279],"category_scores_gemma":[0.0071829082,0.0020227751,0.0012551892,0.008295653,0.0005537059,0.0017947936,0.0031001803,0.0023611428,0.03974079],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022679595,0.000790178,0.008157113,0.007869147,0.0010302996,0.0006502161,0.00036308015,0.0020103212,0.44274387,0.003910799,0.26611334,0.2640937],"study_design_scores_gemma":[0.00036022163,0.0007195925,0.017834416,0.00066764577,0.00065904076,0.002212237,0.000073810545,0.0048573976,0.35943747,0.0042606583,0.6086384,0.0002791362],"about_ca_topic_score_codex":0.00074083806,"about_ca_topic_score_gemma":0.0015625107,"teacher_disagreement_score":0.020426279,"about_ca_system_score_codex":0.00071881025,"about_ca_system_score_gemma":0.0019414326,"threshold_uncertainty_score":0.06833267},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"bench_or_experimental","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2093745477","doi":"10.1038/sdata.2014.49","title":"An open science resource for establishing reliability and reproducibility in functional connectomics","year":2014,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":497,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Douglas Mental Health University Institute; Université de Montréal; Institut Universitaire de Gériatrie de Montréal","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute on Drug Abuse; National Institute of Biomedical Imaging and Bioengineering; National Institute of Neurological Disorders and Stroke; National Institute of Mental Health; National Key Research and Development Program of China","keywords":"Connectomics; Reliability (semiconductor); Reproducibility; Computer science; Neuroimaging; Connectome; Open science; Artificial intelligence; Functional connectivity; Psychology; Neuroscience; Statistics; Mathematics","score_opus":0.14234638286896062,"score_gpt":0.3469814610188182,"score_spread":0.2046350781498576,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2093745477","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0068717157,0.0019194211,0.66883,0.008834189,0.0031275223,0.007889228,0.13096169,0.043989077,0.12757716],"genre_scores_gemma":[0.039016005,0.0022142269,0.7109549,0.002778745,0.0025581853,0.025411323,0.1668875,0.016321393,0.03385768],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9334393,0.020816386,0.011617647,0.005015661,0.027103765,0.002007339],"domain_scores_gemma":[0.43359545,0.17116196,0.02091087,0.23737553,0.12366255,0.013293593],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.10747018,0.0018853596,0.004302735,0.020782102,0.0043686703,0.00779435,0.00918653,0.005400045,0.13306266],"category_scores_gemma":[0.3429892,0.002544304,0.0027049913,0.02326208,0.0032302514,0.007877639,0.0134454295,0.0051832623,0.0850973],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011278404,0.0006635409,0.0037558936,0.0025066908,0.00031212653,0.0005521677,0.00074182166,0.002864054,0.009119995,0.04755579,0.46922266,0.46157733],"study_design_scores_gemma":[0.0008428746,0.00045894418,0.013275452,0.002117909,0.00029993505,0.00084043166,0.00029034476,0.011726297,0.012406238,0.10864294,0.84874785,0.00035069208],"about_ca_topic_score_codex":0.0035047624,"about_ca_topic_score_gemma":0.004478469,"teacher_disagreement_score":0.9908135,"about_ca_system_score_codex":0.0029430406,"about_ca_system_score_gemma":0.022810351,"threshold_uncertainty_score":0.5683634},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2190609062","doi":"10.1038/sdata.2015.43","title":"Test-retest resting-state fMRI in healthy elderly persons with a family history of Alzheimer’s disease","year":2015,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Montreal Neurological Institute and Hospital; McGill University Health Centre; McGill University; Douglas Mental Health University Institute; Institut Universitaire de Gériatrie de Montréal","funders":"Pfizer Canada; Government of Canada; McGill University; Pfizer","keywords":"Dementia; Neuroimaging; Medicine; Alzheimer's disease; Disease; Reliability (semiconductor); Clinical trial; Cohort; Psychology; Audiology; Physical medicine and rehabilitation; Psychiatry; Internal medicine","score_opus":0.19530099344636476,"score_gpt":0.3150118630483633,"score_spread":0.11971086960199856,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2190609062","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9932701,0.00038600733,0.0018990454,0.00005054216,0.000014230017,0.00010142503,0.0032891124,0.00007044628,0.0009191994],"genre_scores_gemma":[0.9860373,0.00022260161,0.0033343562,0.000082640814,0.000038699258,0.00024616663,0.009286079,0.00003770259,0.0007144379],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99960583,0.000107203785,0.000040649942,0.00015322324,0.0000593528,0.00003372789],"domain_scores_gemma":[0.9988294,0.0003969578,0.0001567734,0.00028311697,0.00027185894,0.00006181111],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009694554,0.00037611512,0.00038440348,0.0005801025,0.00048222326,0.0003374686,0.00037794354,0.00055344,0.00082430267],"category_scores_gemma":[0.0027707745,0.00019738292,0.00022475257,0.0002919476,0.00032397223,0.00022731499,0.00030346715,0.0003000388,0.0004369625],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0070798807,0.0019086286,0.7532213,0.0006421966,0.0018906854,0.002870249,0.003919867,0.0022839678,0.11947579,0.00040157966,0.009764972,0.096540995],"study_design_scores_gemma":[0.000074009404,0.00050247955,0.99059063,0.000010481571,0.00016809646,0.0021031806,0.00014700978,0.0005256736,0.003263034,0.0002565454,0.0023368865,0.00002188513],"about_ca_topic_score_codex":0.0032024027,"about_ca_topic_score_gemma":0.014275041,"teacher_disagreement_score":0.0032024027,"about_ca_system_score_codex":0.00015562432,"about_ca_system_score_gemma":0.00018949558,"threshold_uncertainty_score":0.006367564},"labels":[],"label_agreement":null},{"id":"W2313381310","doi":"10.1038/sdata.2014.26","title":"An extended Arctic proxy temperature database for the past 2,000 years","year":2014,"lang":"en","type":"article","venue":"Scientific Data","topic":"Tree-ring climate responses","field":"Earth and Planetary Sciences","cited_by":166,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Oceanic and Atmospheric Administration; National Science Foundation","keywords":"Proxy (statistics); Arctic; Database; Climatology; Seasonality; Climate change; The arctic; Context (archaeology); Physical geography; Environmental science; Geography; Computer science; Oceanography; Geology; Archaeology","score_opus":0.033771668608765314,"score_gpt":0.27553162902992934,"score_spread":0.24175996042116402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2313381310","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.053292267,0.00068296667,0.00800783,0.00010625965,0.00008485787,0.000060197348,0.93230695,0.00083651213,0.0046220697],"genre_scores_gemma":[0.06273542,0.00071295263,0.016113937,0.000061018374,0.000051210714,0.00015515793,0.91834384,0.00016083269,0.0016656435],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99943167,0.00007557945,0.00010089709,0.00018376959,0.00016288384,0.000045174645],"domain_scores_gemma":[0.9971897,0.00023573835,0.00043209078,0.00078406156,0.0011999488,0.00015852283],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011245763,0.00040035893,0.0005004778,0.0025326526,0.00049604307,0.0009219575,0.0007539608,0.00036193372,0.0032588558],"category_scores_gemma":[0.0026292794,0.00023199942,0.0004104154,0.0063046273,0.0001272552,0.00051787164,0.00067891314,0.00046824347,0.0025391455],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014334907,0.00023634851,0.4037363,0.0024078002,0.0014869818,0.0005614121,0.0015257879,0.036803495,0.021567032,0.0066800877,0.2467383,0.27682295],"study_design_scores_gemma":[0.00008319496,0.000069993665,0.4550415,0.0002574168,0.0002750102,0.00039403085,0.00027065777,0.0087977545,0.00844053,0.0010178721,0.5252513,0.00010064123],"about_ca_topic_score_codex":0.06721927,"about_ca_topic_score_gemma":0.097575136,"teacher_disagreement_score":0.06721927,"about_ca_system_score_codex":0.0007108312,"about_ca_system_score_gemma":0.0026152781,"threshold_uncertainty_score":0.13365602},"labels":[],"label_agreement":null},{"id":"W2318435655","doi":"10.1038/sdata.2014.2","title":"The systematic identification of cytoskeletal genes required for Drosophila melanogaster muscle maintenance","year":2014,"lang":"en","type":"article","venue":"Scientific Data","topic":"Muscle Physiology and Disorders","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Drosophila melanogaster; Biology; RNA interference; Computational biology; Model organism; Identification (biology); Gene; Function (biology); Metadata; Melanogaster; Drosophila (subgenus); Bioinformatics; Genetics; Computer science; RNA; World Wide Web; Ecology","score_opus":0.021877616705588094,"score_gpt":0.2732356002784352,"score_spread":0.2513579835728471,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2318435655","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8415338,0.0069278176,0.009460715,0.0005067282,0.00015819768,0.0001824954,0.13132034,0.0013311796,0.008578717],"genre_scores_gemma":[0.7180934,0.006706886,0.032079436,0.0012331106,0.000049718998,0.00032226718,0.22862326,0.00072965375,0.012162255],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99974483,0.000015864425,0.000029233235,0.00007055433,0.000112332935,0.000027150942],"domain_scores_gemma":[0.9995983,0.00010903884,0.00009529694,0.000078978854,0.00007523368,0.000043126456],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00021321622,0.0005711771,0.0003990474,0.0010814628,0.00043350874,0.00039081578,0.00025567517,0.00035652766,0.0023538552],"category_scores_gemma":[0.00039062492,0.0002292692,0.0005275199,0.001059636,0.00018100826,0.00016965905,0.00035631206,0.0004680148,0.0013604532],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024359286,0.000025479218,0.010944904,0.000539789,0.00007770541,0.00051875715,0.00007328053,0.00034683163,0.95854986,0.00035508457,0.004109413,0.024215348],"study_design_scores_gemma":[0.000119852804,0.0003940946,0.32985145,0.0001985901,0.0006306208,0.00335238,0.0002787404,0.0029826981,0.5392729,0.0007667631,0.12205855,0.00009336913],"about_ca_topic_score_codex":0.0020520769,"about_ca_topic_score_gemma":0.005985381,"teacher_disagreement_score":0.0023538552,"about_ca_system_score_codex":0.00033455214,"about_ca_system_score_gemma":0.00045156406,"threshold_uncertainty_score":0.007874429},"labels":[],"label_agreement":null},{"id":"W2324317819","doi":"10.1038/sdata.2016.6","title":"Female and male life tables for seven wild primate species","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Primate Behavior and Ecology","field":"Psychology","cited_by":94,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Evolutionary Synthesis Center","keywords":"Gorilla; Biology; Survivorship curve; Demography; Reproductive value; Primate; Population; Ecology; Zoology","score_opus":0.12397454317852258,"score_gpt":0.36303121266904764,"score_spread":0.23905666949052506,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2324317819","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.36554587,0.0027645843,0.03052705,0.0005880936,0.00019969663,0.00022910064,0.5591852,0.0023278734,0.03863246],"genre_scores_gemma":[0.68824816,0.0011816191,0.047954977,0.00024221315,0.00010672804,0.00063461997,0.25077522,0.00027378977,0.010582722],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9997079,0.00009947196,0.000054846947,0.000056299934,0.000064620996,0.000016872948],"domain_scores_gemma":[0.9971898,0.0013034829,0.00066628243,0.00026793845,0.00047745358,0.00009504893],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00054144487,0.00024783658,0.00017691443,0.0021938654,0.000190332,0.0003894869,0.0002639186,0.00012808475,0.018636987],"category_scores_gemma":[0.004202242,0.0001108571,0.00015913267,0.0021155595,0.00009581838,0.00036453307,0.00021524946,0.00021694248,0.0028729406],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00063981983,0.00012992743,0.39274243,0.00084805535,0.00035536973,0.00022171381,0.0022440702,0.0068672015,0.003312075,0.01619003,0.178466,0.3979833],"study_design_scores_gemma":[0.00007515141,0.00030987294,0.506996,0.00013763766,0.0000968958,0.001366474,0.0006645555,0.011596188,0.0014671412,0.00854602,0.46866783,0.00007624727],"about_ca_topic_score_codex":0.0027691596,"about_ca_topic_score_gemma":0.0054318635,"teacher_disagreement_score":0.018636987,"about_ca_system_score_codex":0.00030118437,"about_ca_system_score_gemma":0.00022170073,"threshold_uncertainty_score":0.062346876},"labels":[],"label_agreement":null},{"id":"W2335458838","doi":"10.1038/sdata.2014.12","title":"Direct infusion mass spectrometry metabolomics dataset: a benchmark for data processing and quality control","year":2014,"lang":"en","type":"article","venue":"Scientific Data","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":139,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Pfizer Canada; University of Birmingham; British Heart Foundation; Natural Environment Research Council; Advantage West Midlands; Pfizer","keywords":"Workflow; Benchmark (surveying); Computer science; Metabolomics; Data mining; Quality (philosophy); Set (abstract data type); Data quality; Quality assurance; Bioinformatics; Biology; Database; Medicine; Engineering","score_opus":0.047168274210509537,"score_gpt":0.3218757350716982,"score_spread":0.2747074608611887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2335458838","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12093969,0.0026734897,0.05191797,0.0020174508,0.00048117214,0.001328448,0.80206615,0.012864982,0.0057106395],"genre_scores_gemma":[0.040758725,0.00043151213,0.05643514,0.0003432924,0.00006974349,0.0009454638,0.8996536,0.00067040895,0.0006921199],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9944722,0.0011003084,0.0007226783,0.0013996959,0.0020215733,0.00028353278],"domain_scores_gemma":[0.98923945,0.002396666,0.0010030841,0.003027365,0.0037857515,0.00054768595],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0073437165,0.0017389768,0.0013617573,0.002837037,0.001269182,0.0024206836,0.0034993275,0.0023928839,0.0022016566],"category_scores_gemma":[0.014433561,0.00044731374,0.0014518137,0.004072123,0.0008612641,0.0011950077,0.0027173015,0.0017503359,0.0032012356],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006588144,0.002284706,0.06561458,0.0063131503,0.0020898986,0.0013940292,0.00048730816,0.052962516,0.06338112,0.0060548387,0.5819332,0.21089649],"study_design_scores_gemma":[0.0017837724,0.0017716903,0.1286825,0.0007086119,0.0006286683,0.002827784,0.00050342793,0.114700235,0.07089192,0.015983665,0.66099215,0.00052561023],"about_ca_topic_score_codex":0.005543026,"about_ca_topic_score_gemma":0.0067380257,"teacher_disagreement_score":0.9926563,"about_ca_system_score_codex":0.0015503943,"about_ca_system_score_gemma":0.0029402953,"threshold_uncertainty_score":0.03883773},"labels":[],"label_agreement":null},{"id":"W2357032010","doi":"10.1038/sdata.2016.32","title":"Digital elevation model and orthophotographs of Greenland based on aerial photographs from 1978–1987","year":2016,"lang":"en","type":"letter","venue":"Scientific Data","topic":"Cryospheric studies and observations","field":"Earth and Planetary Sciences","cited_by":124,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"Villum Fonden","keywords":"Digital elevation model; Glacier; Geology; Geodetic datum; Elevation (ballistics); Glacier mass balance; Greenland ice sheet; Geodesy; Remote sensing; Photogrammetry; Physical geography; Geomorphology; Geography","score_opus":0.041430564278527496,"score_gpt":0.2240699001270552,"score_spread":0.1826393358485277,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2357032010","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4064276,0.004556376,0.021072458,0.097657904,0.004043553,0.0012289516,0.168645,0.0010571738,0.29531103],"genre_scores_gemma":[0.68282527,0.008632731,0.03239016,0.019833552,0.0024012455,0.0006678894,0.08994631,0.00032517815,0.16297774],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99980074,0.000030457028,0.000029762217,0.00003264808,0.00008993747,0.000016392552],"domain_scores_gemma":[0.99933136,0.00014658122,0.00009628748,0.00011960519,0.00027317708,0.000033018427],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003559532,0.00017341925,0.000108355176,0.0006549082,0.00029425102,0.0002434542,0.00026699287,0.00038115543,0.00523444],"category_scores_gemma":[0.001388596,0.00011509843,0.00006113173,0.0010641514,0.00022277313,0.00025122342,0.00026932862,0.0005246869,0.0034071594],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037762657,0.00013442301,0.053796876,0.0002052157,0.000027942555,0.004994969,0.000808564,0.0017332002,0.0058486606,0.005756594,0.6777702,0.24854577],"study_design_scores_gemma":[0.00003897358,0.0000629888,0.12975587,0.00009054934,0.0000145529075,0.0020861903,0.00033313243,0.0025868604,0.0014602018,0.0018526359,0.8616968,0.000021263173],"about_ca_topic_score_codex":0.024069628,"about_ca_topic_score_gemma":0.0783302,"teacher_disagreement_score":0.024069628,"about_ca_system_score_codex":0.0012547224,"about_ca_system_score_gemma":0.0006175157,"threshold_uncertainty_score":0.047859073},"labels":[],"label_agreement":null},{"id":"W2395631589","doi":"10.1038/sdata.2015.59","title":"Multi-contrast submillimetric 3 Tesla hippocampal subfield segmentation protocol and dataset","year":2015,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neuroscience and Neuropharmacology Research","field":"Neuroscience","cited_by":113,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"Canadian Institutes of Health Research","keywords":"Segmentation; Hippocampal formation; Computer science; Neuroimaging; Dentate gyrus; Artificial intelligence; Pattern recognition (psychology); Temporal lobe; Neuroscience; Biology; Epilepsy","score_opus":0.3282191305563179,"score_gpt":0.4502045408260256,"score_spread":0.12198541026970772,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2395631589","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1818479,0.002322456,0.2693658,0.0014092744,0.0007130664,0.017214227,0.49646816,0.018614568,0.012044512],"genre_scores_gemma":[0.096205205,0.0007393658,0.25094354,0.00076261,0.00014205852,0.0237544,0.62019545,0.0024982921,0.0047590593],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99920446,0.00012782076,0.000157178,0.00029551596,0.0001377636,0.00007718366],"domain_scores_gemma":[0.99844676,0.00023682651,0.0000900713,0.0006373537,0.0004984188,0.000090568996],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019811606,0.0015202059,0.0013180396,0.0019394286,0.0015258089,0.0013684944,0.0030990914,0.0020755336,0.01073569],"category_scores_gemma":[0.0035780086,0.0007853792,0.0009864612,0.0016869659,0.00092710933,0.0005655923,0.001717952,0.0015179572,0.00791125],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0064175115,0.0023442246,0.030218687,0.0055336333,0.0010040783,0.0048078327,0.002083441,0.02725779,0.21593939,0.009011206,0.41155788,0.28382435],"study_design_scores_gemma":[0.0022422385,0.0020011812,0.100292996,0.0009674769,0.0011241251,0.01153228,0.0009579931,0.039674193,0.089884564,0.024685236,0.7259577,0.00067997834],"about_ca_topic_score_codex":0.005368335,"about_ca_topic_score_gemma":0.011773786,"teacher_disagreement_score":0.01073569,"about_ca_system_score_codex":0.0007437143,"about_ca_system_score_gemma":0.0024668716,"threshold_uncertainty_score":0.03591448},"labels":[],"label_agreement":null},{"id":"W2415986366","doi":"10.1038/sdata.2016.28","title":"MycoDB, a global database of plant response to mycorrhizal fungi","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Mycorrhizal Fungi and Plant Interactions","field":"Agricultural and Biological Sciences","cited_by":105,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; Okanagan University College; University of British Columbia, Okanagan Campus; University of British Columbia","funders":"University of Pennsylvania","keywords":"Biology; Symbiosis; Abiotic component; Context (archaeology); Ecology; Ecosystem; Mycorrhizal fungi; Mutualism (biology); Biomass (ecology); Phylogenetic tree; Inoculation","score_opus":0.03651476630220665,"score_gpt":0.25812701480581557,"score_spread":0.22161224850360892,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2415986366","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012693757,0.018510379,0.003934025,0.0001655619,0.00006348364,0.000059040794,0.95833236,0.0028709169,0.0033704336],"genre_scores_gemma":[0.03148414,0.01001573,0.018924369,0.00027820648,0.000079202226,0.0004792293,0.9367035,0.001102395,0.0009332356],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971432,0.00047890624,0.0008402678,0.00080052955,0.000578315,0.00015877416],"domain_scores_gemma":[0.98487407,0.006549785,0.0039005268,0.0018897319,0.0018669367,0.000918782],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031837514,0.0021305052,0.002722901,0.02365317,0.00072884175,0.002743957,0.002288305,0.0016049333,0.00913436],"category_scores_gemma":[0.013889135,0.0009937627,0.0020804142,0.024307946,0.00044241958,0.0020442195,0.0026786625,0.0011267896,0.0047950395],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004518732,0.00035484324,0.14847162,0.12840714,0.012566623,0.001672899,0.003847629,0.0077530616,0.08009358,0.010801986,0.38846585,0.21304601],"study_design_scores_gemma":[0.0002584633,0.00018872395,0.11658193,0.0029541473,0.0043092216,0.00080339314,0.0004954983,0.0011801819,0.0063640513,0.0040179677,0.8626779,0.00016858819],"about_ca_topic_score_codex":0.004318416,"about_ca_topic_score_gemma":0.0068969205,"teacher_disagreement_score":0.02365317,"about_ca_system_score_codex":0.0008753489,"about_ca_system_score_gemma":0.002691841,"threshold_uncertainty_score":0.030557454},"labels":[],"label_agreement":null},{"id":"W2418133265","doi":"10.1038/sdata.2016.37","title":"Electricity, water, and natural gas consumption of a residential house in Canada from 2012 to 2014","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Building Energy and Comfort Optimization","field":"Engineering","cited_by":288,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Simon Fraser University","funders":"British Columbia Institute of Technology","keywords":"Electricity; Consumption (sociology); Sustainability; Environmental economics; Power consumption; Computer science; Natural resource; Natural gas; Power (physics); Engineering; Economics; Ecology; Waste management","score_opus":0.01223721011083671,"score_gpt":0.20241991080083227,"score_spread":0.19018270068999557,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2418133265","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.38949132,0.00067214924,0.00056863634,0.0005007523,0.000034263474,0.00004271232,0.6019522,0.00029553712,0.0064424],"genre_scores_gemma":[0.5136465,0.00075178995,0.0014675341,0.00013568421,0.000016435735,0.000039577895,0.47746196,0.00005967677,0.0064209416],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996891,0.0000124049975,0.000017493518,0.00006382286,0.0001224521,0.00009464242],"domain_scores_gemma":[0.99920636,0.000031934844,0.000058481728,0.000031869124,0.00056744117,0.00010395631],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00021027871,0.0004993994,0.00041515587,0.002074168,0.00089151866,0.0008790793,0.0009991765,0.00036607566,0.0026026382],"category_scores_gemma":[0.00093435793,0.00023964574,0.00049218425,0.006053785,0.0004005068,0.0004913981,0.0006930656,0.000518646,0.0007486994],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00054729945,0.00019938701,0.67772585,0.00047503674,0.00033679898,0.0004370536,0.0009804938,0.01711399,0.0010346376,0.0022276505,0.24921043,0.049711388],"study_design_scores_gemma":[0.000021957609,0.000019113979,0.94562125,0.00012977241,0.000050230847,0.00012624807,0.0020588352,0.008929452,0.0007061814,0.00025526874,0.042022504,0.00005912412],"about_ca_topic_score_codex":0.99221367,"about_ca_topic_score_gemma":0.9970693,"teacher_disagreement_score":0.017170608,"about_ca_system_score_codex":0.017170608,"about_ca_system_score_gemma":0.01027827,"threshold_uncertainty_score":0.12458205},"labels":[],"label_agreement":null},{"id":"W2508921413","doi":"10.1038/sdata.2016.67","title":"Global terrestrial Human Footprint maps for 1993 and 2009","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine animal studies overview","field":"Environmental Science","cited_by":817,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Northern British Columbia","funders":"Australian Research Council; James Cook University; Wildlife Conservation Society","keywords":"Footprint; Ecological footprint; Environmental science; Environmental resource management; Land use; Physical geography; Geography; Remote sensing; Ecology; Sustainability; Archaeology","score_opus":0.06895880683861093,"score_gpt":0.3052505647130123,"score_spread":0.23629175787440138,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2508921413","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.042630207,0.0006515915,0.0012388072,0.00019644643,0.000067915396,0.00016163419,0.9400774,0.00027607728,0.014699974],"genre_scores_gemma":[0.13730584,0.001183321,0.011340241,0.00013162928,0.000046811518,0.00058169995,0.84362304,0.000063307656,0.005724129],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.999701,0.00004265809,0.00003949663,0.00005432864,0.00011056628,0.00005195979],"domain_scores_gemma":[0.999134,0.00006487824,0.00024835506,0.00008157283,0.0004061888,0.00006512807],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000523057,0.00054906955,0.00021508022,0.003796356,0.00024011642,0.00038816413,0.00044578387,0.00023170299,0.0055343327],"category_scores_gemma":[0.001026861,0.00014010369,0.000343567,0.007097063,0.00011239678,0.0004843851,0.00051699765,0.00032495282,0.0016888388],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005583578,0.00013638879,0.2665133,0.0020398847,0.0004447734,0.00041743577,0.0016271649,0.006903293,0.0035040232,0.00502737,0.5057684,0.2070596],"study_design_scores_gemma":[0.000027392007,0.000052091167,0.6914451,0.00014876275,0.000049120274,0.00021668315,0.0005393503,0.0012403602,0.0004603595,0.0003715972,0.30542132,0.000027901437],"about_ca_topic_score_codex":0.077768594,"about_ca_topic_score_gemma":0.12240449,"teacher_disagreement_score":0.077768594,"about_ca_system_score_codex":0.00081774575,"about_ca_system_score_gemma":0.0010502723,"threshold_uncertainty_score":0.15463185},"labels":[],"label_agreement":null},{"id":"W2511183000","doi":"10.1038/sdata.2016.70","title":"Mean hydrography on the continental shelf from 26 repeat glider deployments along Southeastern Australia","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and coastal ecosystems","field":"Earth and Planetary Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ocean Networks Canada Society","funders":"Australian Government","keywords":"Continental shelf; Oceanography; Hydrography; Glider; Geology; Boundary current; Stratification (seeds); Continental margin; Upwelling; Ocean current; Water mass; Current (fluid); Paleontology","score_opus":0.05478621495191094,"score_gpt":0.2405939781581103,"score_spread":0.18580776320619935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2511183000","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9955735,0.000041861847,0.00014223605,0.000016733988,0.0000017944192,0.0000062742697,0.0029955273,0.00001107305,0.0012110184],"genre_scores_gemma":[0.990576,0.00012889123,0.00085423974,0.000026059286,0.0000049419,0.0000203479,0.0064399536,0.000008774437,0.0019407022],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9999411,0.000005902671,0.000004779627,0.00002140424,0.000017462753,0.000009392146],"domain_scores_gemma":[0.9998337,0.000010549348,0.00003676681,0.000015446836,0.00007083339,0.00003278564],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00012707138,0.00016187917,0.00013694308,0.0005630095,0.00026324057,0.00021080686,0.00014730306,0.00012982392,0.00078580447],"category_scores_gemma":[0.00032675554,0.0001500174,0.00015256206,0.00062496297,0.00013454581,0.00015916394,0.00039463222,0.00011415351,0.00033194094],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000054895158,0.000031270207,0.9741,0.00003985212,0.00006393007,0.00020717534,0.0013630692,0.0016241138,0.0049600527,0.00007883372,0.0023344974,0.015142274],"study_design_scores_gemma":[0.0000010469895,0.000008747564,0.99840087,0.000003935382,0.0000048052098,0.000044623273,0.00009537105,0.00039520027,0.00008372459,0.000008050759,0.00095165445,0.000001928364],"about_ca_topic_score_codex":0.08695767,"about_ca_topic_score_gemma":0.24986143,"teacher_disagreement_score":0.08695767,"about_ca_system_score_codex":0.00034307424,"about_ca_system_score_gemma":0.0002934207,"threshold_uncertainty_score":0.172903},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2518992676","doi":"10.1038/sdata.2016.79","title":"Single-cell transcriptome and epigenomic reprogramming of cardiomyocyte-derived cardiac progenitor cells","year":2016,"lang":"en","type":"letter","venue":"Scientific Data","topic":"Congenital heart defects research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Cedars-Sinai Medical Center; Division of Chemistry; Massachusetts General Hospital; Heart and Stroke Foundation of Canada; National Cancer Institute; National Institutes of Health; National Science Foundation","keywords":"Epigenomics; Reprogramming; Transcriptome; Biology; DNA methylation; Cell biology; Differentially methylated regions; Myocyte; Regeneration (biology); Induced pluripotent stem cell; Cell; Genetics; Embryonic stem cell; Gene expression; Gene","score_opus":0.03993420189386562,"score_gpt":0.27213786819127384,"score_spread":0.2322036662974082,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2518992676","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24729438,0.027053447,0.015133716,0.6628867,0.015706794,0.00018324243,0.0017545208,0.0007813026,0.029205965],"genre_scores_gemma":[0.7290327,0.024593497,0.008729548,0.18399702,0.02220812,0.00043373866,0.0011478916,0.000099302284,0.029758204],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9995048,0.0001342502,0.000053726948,0.00008371826,0.00016677876,0.000056690064],"domain_scores_gemma":[0.9993129,0.00032824092,0.0000775221,0.000085960826,0.00012604744,0.00006939486],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00073711836,0.00018451801,0.00026858857,0.00022761471,0.00034068924,0.00058945455,0.00024187099,0.0030164332,0.0014081192],"category_scores_gemma":[0.0015058094,0.00014558675,0.00022402193,0.00022574177,0.00047253608,0.00036057475,0.00023284438,0.0024865817,0.0011611349],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025645106,0.00036789628,0.030423129,0.0006226492,0.00011186921,0.0443547,0.0007041672,0.0013575754,0.28524393,0.009796269,0.27010673,0.35434663],"study_design_scores_gemma":[0.00025822158,0.001099072,0.042633325,0.00014323156,0.00010939298,0.07636299,0.00065194885,0.0069559747,0.15196979,0.01842623,0.70129997,0.00008989388],"about_ca_topic_score_codex":0.0002717973,"about_ca_topic_score_gemma":0.0004998942,"teacher_disagreement_score":0.0030164332,"about_ca_system_score_codex":0.0005450813,"about_ca_system_score_gemma":0.00023803303,"threshold_uncertainty_score":0.0047106743},"labels":[],"label_agreement":null},{"id":"W2530539846","doi":"10.1038/sdata.2016.82","title":"Data from a pre-publication independent replication initiative examining ten moral judgement effects","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Psychology of Moral and Emotional Judgment","field":"Neuroscience","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; University of Manitoba","funders":"","keywords":"Replication (statistics); Judgement; Psychology; Political science; Biology; Law","score_opus":0.40607369125410125,"score_gpt":0.36454883497262675,"score_spread":0.04152485628147451,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2530539846","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5843761,0.004224259,0.14889269,0.0070733475,0.003602769,0.06211752,0.08369136,0.0027646713,0.10325723],"genre_scores_gemma":[0.65055144,0.0014730804,0.15269014,0.0076380363,0.0007076935,0.107523926,0.050672166,0.0024901868,0.026253348],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.8865513,0.050368097,0.014166953,0.009565594,0.037034642,0.0023133378],"domain_scores_gemma":[0.37111038,0.24251829,0.026220849,0.22047384,0.13520199,0.0044746483],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.12515019,0.0013435885,0.0015492071,0.0056334026,0.0049707703,0.0039570346,0.003363547,0.00271177,0.011139049],"category_scores_gemma":[0.43680438,0.0013064444,0.0028785327,0.006124064,0.004924896,0.0026165186,0.005927913,0.0038832517,0.006689669],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0113131525,0.009725872,0.12680039,0.01751126,0.0044664787,0.004430941,0.11219291,0.0015580878,0.06691139,0.024103379,0.19702995,0.42395613],"study_design_scores_gemma":[0.0043106526,0.0074484204,0.39129916,0.0046380046,0.004304655,0.003288598,0.024289034,0.001703894,0.07307298,0.02421476,0.46049,0.00093987107],"about_ca_topic_score_codex":0.0052727866,"about_ca_topic_score_gemma":0.011190271,"teacher_disagreement_score":0.8748498,"about_ca_system_score_codex":0.0025102692,"about_ca_system_score_gemma":0.008101367,"threshold_uncertainty_score":0.66186535},"labels":[],"label_agreement":null},{"id":"W2530955115","doi":"10.1038/sdata.2016.87","title":"Long-term, high frequency in situ measurements of intertidal mussel bed temperatures using biomimetic sensors","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Physiological and biochemical adaptations","field":"Environmental Science","cited_by":98,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Environment Research Council; National Research Foundation; National Oceanic and Atmospheric Administration; Sight Research UK; National Aeronautics and Space Administration; David and Lucile Packard Foundation; National Geographic Society; National Science Foundation","keywords":"Intertidal zone; Ectotherm; Mussel; Environmental science; Habitat; Data logger; Intertidal ecology; Oceanography; Atmospheric sciences; Ecology; Biology; Geology","score_opus":0.08470878739218475,"score_gpt":0.28391623266699606,"score_spread":0.19920744527481132,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2530955115","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9913701,0.00006936954,0.006713487,0.00001910293,0.000010218638,0.000014538303,0.00083720585,0.0000737212,0.00089237717],"genre_scores_gemma":[0.9910972,0.00007936424,0.0077495542,0.000021564047,0.000012879232,0.000034042005,0.0006607988,0.000009884912,0.0003346802],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99982435,0.000022300032,0.000009416259,0.00008071696,0.000046149704,0.000017083928],"domain_scores_gemma":[0.99974364,0.00004099692,0.00007967245,0.00004002745,0.00006703503,0.000028598037],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019308776,0.00025518323,0.00018916055,0.00054490776,0.00030528227,0.0003504675,0.00026349997,0.0002336334,0.0005119313],"category_scores_gemma":[0.00030875843,0.00015350443,0.000113680115,0.0005949144,0.00016773812,0.00029742037,0.00032557207,0.0002840374,0.00021144885],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025302442,0.00021593421,0.47959337,0.00015278159,0.00012750256,0.00010867281,0.00055264134,0.004644148,0.46085182,0.00009273258,0.00045508705,0.05295225],"study_design_scores_gemma":[0.000014820771,0.00026574335,0.95024306,0.000011043186,0.00005402272,0.00023370565,0.00021256771,0.0076923077,0.039785825,0.0001243539,0.0013410039,0.000021582282],"about_ca_topic_score_codex":0.0019864107,"about_ca_topic_score_gemma":0.009497499,"teacher_disagreement_score":0.0019864107,"about_ca_system_score_codex":0.00016067906,"about_ca_system_score_gemma":0.0001641316,"threshold_uncertainty_score":0.003949642},"labels":[],"label_agreement":null},{"id":"W2550462889","doi":"10.1038/sdata.2016.97","title":"A variant reference data set for the Africanized honeybee, Apis mellifera","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant and animal studies","field":"Agricultural and Biological Sciences","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"Ontario Ministry of Research and Innovation; Natural Sciences and Engineering Research Council of Canada; Fundação de Amparo à Pesquisa do Estado de São Paulo; York University","keywords":"Biology; SNP; Beekeeping; Population; Single-nucleotide polymorphism; Honey bee; Evolutionary biology; Population genetics; Genetics; Zoology; Ecology; Gene; Genotype; Demography","score_opus":0.4006907821873715,"score_gpt":0.29471166432926116,"score_spread":0.10597911785811032,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2550462889","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06556269,0.0021339615,0.0231006,0.00020885878,0.00027308468,0.0002114661,0.9012751,0.0014661275,0.005768091],"genre_scores_gemma":[0.026898198,0.00034812273,0.014884936,0.00008765491,0.000017686516,0.00038473855,0.95534766,0.00033122883,0.0016997354],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99840456,0.00021780447,0.00027907503,0.0005659616,0.00039935872,0.00013330295],"domain_scores_gemma":[0.9972589,0.0005276182,0.0003898227,0.00094376516,0.00072918256,0.00015063185],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020252457,0.0009335521,0.0013018028,0.0038649621,0.0014190902,0.0011018262,0.0017707868,0.0010370723,0.009129674],"category_scores_gemma":[0.0051868036,0.00040484854,0.0009776833,0.0062638093,0.0003462868,0.00044138439,0.0010561117,0.0012365651,0.007465281],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0038557881,0.0006565966,0.1282793,0.007656065,0.0027779406,0.0037383735,0.002015075,0.010365326,0.27332836,0.012844014,0.30871826,0.24576491],"study_design_scores_gemma":[0.0002848078,0.0002548267,0.18694094,0.00046356014,0.0007142625,0.0020368171,0.00031643847,0.0037365356,0.015532209,0.004317888,0.78525895,0.00014281974],"about_ca_topic_score_codex":0.017249452,"about_ca_topic_score_gemma":0.039442945,"teacher_disagreement_score":0.017249452,"about_ca_system_score_codex":0.000635154,"about_ca_system_score_gemma":0.0017297784,"threshold_uncertainty_score":0.034298122},"labels":[],"label_agreement":null},{"id":"W2551102403","doi":"10.1038/sdata.2016.95","title":"Systematic chemical-genetic and chemical-chemical interaction datasets for prediction of compound synergism","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute for Research in Immunology and Cancer; Université de Montréal; Lunenfeld-Tanenbaum Research Institute; Mount Sinai Hospital; McMaster University","funders":"National Institutes of Health; Génome Québec; National Center for Research Resources; Canada Research Chairs; Wellcome Trust","keywords":"Benchmarking; Pairwise comparison; Benchmark (surveying); Chemical similarity; Computer science; Computational biology; Drug discovery; Data mining; Artificial intelligence; Bioinformatics; Biology; Cluster analysis","score_opus":0.05635813174344787,"score_gpt":0.3235587407957939,"score_spread":0.267200609052346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2551102403","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71778554,0.0044897995,0.032081287,0.001071044,0.000110350964,0.00072957814,0.23448518,0.0028711676,0.0063760015],"genre_scores_gemma":[0.6405189,0.0014022241,0.07684973,0.0002764277,0.000050386156,0.000885831,0.2792584,0.00013445434,0.0006235954],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9983949,0.00055468094,0.00014458336,0.0003509814,0.00047974157,0.00007508422],"domain_scores_gemma":[0.99592435,0.0022887464,0.0005769078,0.0006287145,0.00033405048,0.00024725308],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002394421,0.0010979077,0.0011453463,0.0034722858,0.000661817,0.0009591025,0.0015303552,0.0013285271,0.0022173054],"category_scores_gemma":[0.006259797,0.00042956765,0.0016502199,0.0042023156,0.0006094406,0.0010648003,0.0012812726,0.0011457981,0.00053606124],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0032427679,0.0031376001,0.14422219,0.006942076,0.002906431,0.0009514433,0.00015166895,0.6051703,0.058184918,0.010215535,0.048562344,0.11631274],"study_design_scores_gemma":[0.001099796,0.0022137794,0.084699504,0.00017447563,0.0010480479,0.00063901383,0.00023747496,0.7933688,0.045973018,0.01858594,0.051769573,0.00019064841],"about_ca_topic_score_codex":0.002989622,"about_ca_topic_score_gemma":0.0069722226,"teacher_disagreement_score":0.0034722858,"about_ca_system_score_codex":0.0008992952,"about_ca_system_score_gemma":0.0019847164,"threshold_uncertainty_score":0.012663066},"labels":[],"label_agreement":null},{"id":"W2552656952","doi":"10.1038/sdata.2016.99","title":"Cross-cultural dataset for the evolution of religion and morality project","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cultural Differences and Values","field":"Psychology","cited_by":85,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Social Sciences and Humanities Research Council of Canada; Aarhus Universitet; Canadian Institute for Advanced Research; Masarykova Univerzita; Canada Excellence Research Chairs, Government of Canada; John Templeton Foundation","keywords":"Morality; Epistemology; Sociology; Philosophy","score_opus":0.24840185657579653,"score_gpt":0.47470366138294146,"score_spread":0.22630180480714493,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2552656952","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0052158623,0.000084732594,0.0010626916,0.00022846407,0.00008510764,0.00030042636,0.98986536,0.00027975912,0.0028775183],"genre_scores_gemma":[0.008970524,0.00006486591,0.0040705074,0.00013437223,0.00003453674,0.002480401,0.9814276,0.00014864164,0.0026686082],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981754,0.0005220842,0.00030163588,0.00036214004,0.0004606729,0.0001781766],"domain_scores_gemma":[0.99094534,0.0021307236,0.001142799,0.0022731978,0.0026229103,0.0008849739],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033575755,0.00069465593,0.0006732441,0.0036589736,0.0010399375,0.0015121159,0.0016752685,0.0010284072,0.038462047],"category_scores_gemma":[0.017399486,0.00041272552,0.0007018379,0.0052238014,0.00031111817,0.00082747097,0.002417796,0.0019599344,0.022920635],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002561362,0.00018576975,0.022075111,0.0005974761,0.00009213827,0.00012018872,0.00067799183,0.00029664472,0.00027777295,0.0030627234,0.9520161,0.020342],"study_design_scores_gemma":[0.00044598395,0.00007265597,0.0963836,0.0003163411,0.00006947853,0.00020672784,0.0012460855,0.0008966157,0.00078395,0.0031182128,0.8963914,0.00006895441],"about_ca_topic_score_codex":0.02108263,"about_ca_topic_score_gemma":0.026564572,"teacher_disagreement_score":0.038462047,"about_ca_system_score_codex":0.0012689003,"about_ca_system_score_gemma":0.0019557227,"threshold_uncertainty_score":0.12866837},"labels":[],"label_agreement":null},{"id":"W2560741446","doi":"10.1038/sdata.2016.102","title":"Sharing brain mapping statistical results with the neuroimaging data model","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Neurological Institute and Hospital","funders":"National Center for Research Resources; Laura and John Arnold Foundation; National Institute on Drug Abuse; National Institute of Neurological Disorders and Stroke; National Institute of Mental Health; Medical Research Council; U.S. National Library of Medicine; National Institute on Alcohol Abuse and Alcoholism; Ludmer Centre for Neuroinformatics and Mental Health; National Institute of Biomedical Imaging and Bioengineering; Wellcome Trust; National Institutes of Health","keywords":"Metadata; Computer science; Neuroimaging; Software; Representation (politics); Univariate; Data sharing; Data science; External Data Representation; Information retrieval; Data mining; Artificial intelligence; Machine learning; World Wide Web; Programming language; Psychology","score_opus":0.24801594761508433,"score_gpt":0.3328507910988038,"score_spread":0.08483484348371945,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2560741446","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021888346,0.0002974661,0.85860825,0.0019540694,0.00040965728,0.0004721696,0.06638692,0.064300664,0.0053818403],"genre_scores_gemma":[0.052930288,0.0011957706,0.75090605,0.0016235097,0.0005121273,0.0050586094,0.14538671,0.03625969,0.0061272318],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9852014,0.0060963253,0.0027731615,0.002263533,0.0033131922,0.0003523899],"domain_scores_gemma":[0.89552623,0.05163696,0.003309717,0.038810745,0.009907382,0.000808991],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.052943368,0.0030226158,0.002172246,0.006020185,0.0010392556,0.009602589,0.0051458566,0.0027121066,0.033777304],"category_scores_gemma":[0.15023735,0.002348508,0.006496248,0.0047302092,0.0015790699,0.0070283557,0.006724075,0.00440875,0.025749182],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013543069,0.00031673684,0.012365963,0.0046678274,0.001809952,0.0014383036,0.0018938816,0.039286528,0.006580805,0.18103759,0.51747835,0.23176974],"study_design_scores_gemma":[0.0004999365,0.00016149966,0.0036456482,0.001140221,0.00071960175,0.00093119737,0.00036232674,0.104889326,0.018010212,0.35881874,0.5103844,0.00043696197],"about_ca_topic_score_codex":0.0048915874,"about_ca_topic_score_gemma":0.0038398746,"teacher_disagreement_score":0.052943368,"about_ca_system_score_codex":0.002117999,"about_ca_system_score_gemma":0.0075602992,"threshold_uncertainty_score":0.2799946},"labels":[],"label_agreement":null},{"id":"W2572195345","doi":"10.1038/sdata.2016.127","title":"A dataset of human decision-making in teamwork management","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cognitive Science and Mapping","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Natural Science Foundation of China; Nanyang Technological University; National Research Foundation","keywords":"Teamwork; Task (project management); Computer science; Bridge (graph theory); Baseline (sea); Knowledge management; Scale (ratio); Business decision mapping; Resource (disambiguation); Data science; Decision support system; Management science; Artificial intelligence; Engineering; Systems engineering","score_opus":0.09217427782541991,"score_gpt":0.39287430398431183,"score_spread":0.3007000261588919,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2572195345","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20074275,0.0017566726,0.008114039,0.0013961109,0.0002744109,0.0007795938,0.7710543,0.0010358737,0.014846164],"genre_scores_gemma":[0.17336673,0.00050876173,0.01922672,0.00035410287,0.00008579109,0.0018193466,0.7999632,0.00010833082,0.0045670723],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99785644,0.0009033663,0.0002129375,0.00044173904,0.00041476317,0.00017066432],"domain_scores_gemma":[0.9924118,0.004041632,0.0006543392,0.0011087407,0.00097293453,0.0008105638],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001389258,0.00083536806,0.00054015085,0.0018510997,0.00088414876,0.001156256,0.0015073803,0.0019471684,0.006333422],"category_scores_gemma":[0.008693531,0.00024011986,0.0008264816,0.0026035355,0.00039384505,0.00070663146,0.0013941831,0.0011674978,0.004743104],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001934153,0.002785879,0.15860315,0.0048801345,0.00056449085,0.0012885976,0.0052723,0.016674178,0.0040944535,0.0070808586,0.6677868,0.12903504],"study_design_scores_gemma":[0.00039689709,0.00061693834,0.373479,0.00085422985,0.00012468368,0.00080915476,0.0054709357,0.024160769,0.0036107383,0.011038293,0.5792097,0.0002285594],"about_ca_topic_score_codex":0.007368874,"about_ca_topic_score_gemma":0.018480144,"teacher_disagreement_score":0.007368874,"about_ca_system_score_codex":0.0010484263,"about_ca_system_score_gemma":0.0011535034,"threshold_uncertainty_score":0.021187365},"labels":[],"label_agreement":null},{"id":"W2574380552","doi":"10.1038/sdata.2016.129","title":"Two replications of an investigation on empathy and utilitarian judgement across socioeconomic status","year":2017,"lang":"en","type":"review","venue":"Scientific Data","topic":"Psychological Well-being and Life Satisfaction","field":"Psychology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Empathy; Socioeconomic status; Judgement; Psychology; Social psychology; Sociology; Demography; Political science","score_opus":0.3656238084735312,"score_gpt":0.5101224505794041,"score_spread":0.14449864210587288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2574380552","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18238346,0.7565219,0.011714993,0.001704215,0.003265552,0.034346025,0.002018262,0.00016693909,0.00787874],"genre_scores_gemma":[0.55923873,0.3429768,0.03239555,0.0023934105,0.000757175,0.05398176,0.002655802,0.000082399216,0.005518335],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99217963,0.003952333,0.0015478391,0.00080927793,0.0013445219,0.00016632621],"domain_scores_gemma":[0.9632471,0.019010963,0.0037524991,0.006293118,0.007342662,0.00035369312],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.025964193,0.0014129599,0.0032288881,0.0021770606,0.0010754649,0.0014030982,0.0018483625,0.0015169017,0.0025131197],"category_scores_gemma":[0.060051896,0.000806256,0.0065249098,0.003355392,0.0012754902,0.0012000016,0.0011134741,0.0023356858,0.00064795814],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.013279949,0.0071008643,0.014571897,0.37064558,0.061195362,0.0009228343,0.009163294,0.001140917,0.01135307,0.0059910263,0.009068637,0.49556646],"study_design_scores_gemma":[0.02768928,0.069390275,0.19763824,0.12882279,0.2713405,0.0030601919,0.010028038,0.0019010409,0.02457111,0.016235331,0.24847583,0.00084732904],"about_ca_topic_score_codex":0.0036701371,"about_ca_topic_score_gemma":0.008888163,"teacher_disagreement_score":0.9740358,"about_ca_system_score_codex":0.0013331668,"about_ca_system_score_gemma":0.0038036213,"threshold_uncertainty_score":0.13731337},"labels":[],"label_agreement":null},{"id":"W2577333108","doi":"10.1038/sdata.2016.132","title":"Whole genome sequencing data and de novo draft assemblies for 66 teleost species","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":85,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Science Council; Chiba University; Universitetet i Oslo; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; McGill University; Norges Forskningsråd; Norwegian Sequencing Centre; Thünen-Institut; Pinngortitaleriffik; Norges Teknisk-Naturvitenskapelige Universitet; National Science Foundation","keywords":"Phylogenetic tree; Biology; Genome; Evolutionary biology; Computational biology; Phylogenetics; Genomics; Whole genome sequencing; Comparative genomics; Genetics; Gene","score_opus":0.10669312967114036,"score_gpt":0.3204797343100458,"score_spread":0.21378660463890542,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2577333108","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5208985,0.002062024,0.08867914,0.0002980551,0.00022900655,0.0006514482,0.3761312,0.0027608462,0.0082896985],"genre_scores_gemma":[0.14375679,0.0015420296,0.23881865,0.00015471356,0.000050898838,0.0010473123,0.60955256,0.0014213729,0.003655709],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99911815,0.000103368926,0.0001808522,0.0003048011,0.00021035678,0.00008249648],"domain_scores_gemma":[0.99845123,0.00043265373,0.0003247485,0.00030309838,0.00035369006,0.00013452525],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001010969,0.000948207,0.0006943171,0.0028781223,0.001352103,0.0010867071,0.00083180005,0.0008291747,0.006138761],"category_scores_gemma":[0.0038789636,0.0006993188,0.0012999722,0.0039187768,0.0004188931,0.00089688034,0.0009324312,0.0013617099,0.004168716],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022089365,0.00037807765,0.048999473,0.0034358012,0.0006664335,0.0019763669,0.0033838265,0.009875258,0.7555993,0.0038390558,0.014109069,0.15552837],"study_design_scores_gemma":[0.00030629808,0.0008741024,0.4219586,0.00069922896,0.001746376,0.0027773897,0.0018538681,0.020663463,0.14993966,0.0062128375,0.39257216,0.0003960665],"about_ca_topic_score_codex":0.0022273446,"about_ca_topic_score_gemma":0.0065985587,"teacher_disagreement_score":0.006138761,"about_ca_system_score_codex":0.00062265113,"about_ca_system_score_gemma":0.0013879397,"threshold_uncertainty_score":0.020536244},"labels":[],"label_agreement":null},{"id":"W2580202562","doi":"10.1038/sdata.2016.120","title":"A 4-study replication of the moderating effects of greed on socioeconomic status and unethical behaviour","year":2017,"lang":"en","type":"review","venue":"Scientific Data","topic":"Experimental Behavioral Economics Studies","field":"Social Sciences","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Socioeconomic status; Replication (statistics); Psychology; Social psychology; Sociology; Biology; Demography; Virology","score_opus":0.27686203510584495,"score_gpt":0.48794559857871705,"score_spread":0.2110835634728721,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2580202562","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20019245,0.70262706,0.01215682,0.0030617611,0.0038122917,0.06274046,0.00495989,0.00042358713,0.010025749],"genre_scores_gemma":[0.48943624,0.36653072,0.042939544,0.003571607,0.0010768371,0.08525923,0.0049157063,0.00016130743,0.006108956],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9934743,0.0032822925,0.0013189646,0.00064938434,0.0010920584,0.00018297583],"domain_scores_gemma":[0.96333295,0.017532462,0.0037168348,0.0056552943,0.009308874,0.00045352054],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.024299579,0.0016885631,0.0040733474,0.002758883,0.0010660519,0.001576053,0.0020687722,0.0015289248,0.0034801287],"category_scores_gemma":[0.07142268,0.0011806773,0.010082469,0.0032341564,0.0009647374,0.001533449,0.0015070208,0.0027834838,0.0010791919],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.009322101,0.005400932,0.025326919,0.31271777,0.1018286,0.0011251009,0.0055990084,0.0009406744,0.005395326,0.0031240496,0.011381769,0.51783776],"study_design_scores_gemma":[0.030482994,0.034617506,0.18387206,0.10592184,0.45336574,0.0022190802,0.0037904151,0.0014232036,0.011857011,0.008984934,0.1627004,0.00076491595],"about_ca_topic_score_codex":0.008246999,"about_ca_topic_score_gemma":0.02218077,"teacher_disagreement_score":0.97570044,"about_ca_system_score_codex":0.0017638374,"about_ca_system_score_gemma":0.007930859,"threshold_uncertainty_score":0.12851},"labels":[],"label_agreement":null},{"id":"W2607313420","doi":"10.1038/sdata.2017.35","title":"A geographically-diverse collection of 418 human gut microbiome pathway genome databases","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; Genome Alberta; King Abdullah University of Science and Technology; Genome British Columbia; Genome Canada; Tula Foundation; School of Medicine, Stanford University; Compute Canada; National Institute of General Medical Sciences; Canadian Institute for Advanced Research","keywords":"Metagenomics; Microbiome; Human Microbiome Project; Compendium; Human microbiome; Computational biology; Annotation; Biology; Data science; Genome; Bioinformatics; Computer science; Genetics; Gene; Geography","score_opus":0.07441962437904774,"score_gpt":0.336488425230079,"score_spread":0.26206880085103124,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2607313420","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01915799,0.002561446,0.0042674877,0.00020268989,0.000034411943,0.0001686486,0.9686321,0.0015989174,0.0033762294],"genre_scores_gemma":[0.013536251,0.0010338599,0.013729591,0.00011480378,0.000010270497,0.0003073422,0.97060364,0.0001802677,0.00048402374],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983601,0.00028714194,0.00029741434,0.000509182,0.0004184985,0.00012771385],"domain_scores_gemma":[0.99759406,0.0007344794,0.00036167534,0.0004438387,0.0005154893,0.0003504849],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020606637,0.0012401561,0.0010225986,0.006693901,0.001077985,0.0016855275,0.0012434538,0.000981992,0.012734572],"category_scores_gemma":[0.006158739,0.00064742786,0.0010998469,0.012191703,0.00033355466,0.00090775895,0.0027518966,0.0010870291,0.007285304],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004273266,0.00050683605,0.08869828,0.02119255,0.0021816154,0.0021864614,0.001737622,0.00854882,0.07040643,0.0077343844,0.4603176,0.3322161],"study_design_scores_gemma":[0.00058767025,0.00026941756,0.12781544,0.0018340795,0.0009922696,0.0018599994,0.00082228915,0.004562765,0.022093205,0.0061307116,0.8328216,0.00021052979],"about_ca_topic_score_codex":0.00724403,"about_ca_topic_score_gemma":0.014352895,"teacher_disagreement_score":0.012734572,"about_ca_system_score_codex":0.00068596384,"about_ca_system_score_gemma":0.0046023186,"threshold_uncertainty_score":0.042601347},"labels":[],"label_agreement":null},{"id":"W2612115609","doi":"10.1038/sdata.2017.65","title":"Modern pollen data from the Canadian Arctic, 1972–1973","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"University of Colorado Boulder; National Science Foundation","keywords":"Pollen; Arctic; Taiga; Tundra; Physical geography; Transect; Environmental science; Vegetation (pathology); Moss; Sampling (signal processing); Boreal; Geography; Ecology; Geology; Oceanography; Forestry; Biology; Archaeology","score_opus":0.17918023584812012,"score_gpt":0.3210715420865105,"score_spread":0.14189130623839039,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2612115609","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.29621342,0.002364135,0.0009652185,0.0002755599,0.00009604194,0.00014201675,0.66338354,0.00031645477,0.03624364],"genre_scores_gemma":[0.47380996,0.0021113295,0.0035168128,0.00019278446,0.00006370585,0.00022924428,0.49878043,0.0001359867,0.021159725],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992654,0.000011632857,0.00003310453,0.0000706395,0.0004782555,0.00014100125],"domain_scores_gemma":[0.9976069,0.000059425576,0.00015156658,0.0000781434,0.0019023276,0.00020168589],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045060416,0.00053123164,0.00033721456,0.007145485,0.0029360068,0.0011377952,0.00078624295,0.00019962004,0.0034739494],"category_scores_gemma":[0.0012389006,0.0002800833,0.00040024956,0.013823736,0.00038737603,0.00029063207,0.0006399302,0.00042087902,0.000800495],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014846828,0.00016769223,0.56711733,0.0013842362,0.0006864226,0.00071977964,0.0030530957,0.0050162263,0.008366584,0.0033808472,0.23784752,0.17077558],"study_design_scores_gemma":[0.000020614018,0.00001986045,0.858058,0.000057933652,0.00003842782,0.00006989806,0.00030420773,0.00037948173,0.0009944123,0.00006988539,0.13995877,0.000028546294],"about_ca_topic_score_codex":0.9925895,"about_ca_topic_score_gemma":0.99695146,"teacher_disagreement_score":0.018551081,"about_ca_system_score_codex":0.018551081,"about_ca_system_score_gemma":0.025160104,"threshold_uncertainty_score":0.13459814},"labels":[],"label_agreement":null},{"id":"W2624462411","doi":"10.1038/sdata.2017.88","title":"A global multiproxy database for temperature reconstructions of the Common Era","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Tree-ring climate responses","field":"Earth and Planetary Sciences","cited_by":532,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; University of Regina; University of Ottawa; University of Toronto; Institut National de la Recherche Scientifique","funders":"U.S. Geological Survey; National Centers for Environmental Information; National Oceanic and Atmospheric Administration; Natural Environment Research Council; Sight Research UK","keywords":"Proxy (statistics); Database; Temperature record; Climate change; Sea surface temperature; Climatology; Geology; Physical geography; Oceanography; Geography; Computer science","score_opus":0.059064136772069696,"score_gpt":0.30584740734923593,"score_spread":0.24678327057716623,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2624462411","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008047542,0.00026815635,0.016123928,0.000099169614,0.00006386138,0.00007628016,0.9557281,0.015644554,0.003948334],"genre_scores_gemma":[0.019171473,0.00021031001,0.020068062,0.000055195367,0.000028890434,0.00025801433,0.9572671,0.0020654902,0.0008754602],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9993637,0.000073424126,0.00012804223,0.00020663835,0.00016341075,0.00006470023],"domain_scores_gemma":[0.9978229,0.00029086036,0.0003710647,0.00087295606,0.00045917116,0.00018302377],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011725583,0.0009879693,0.0008173325,0.0037487638,0.00043075046,0.0014293485,0.001746979,0.0008271844,0.019586494],"category_scores_gemma":[0.003820068,0.0005504975,0.00088207,0.007016654,0.00029002837,0.0018618548,0.002009271,0.0012363219,0.016232824],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008619294,0.00023560782,0.066409945,0.0030870738,0.00066788757,0.00052129256,0.0008085519,0.021850064,0.0112254545,0.012107863,0.74582016,0.13640405],"study_design_scores_gemma":[0.00026875696,0.00007137995,0.099663034,0.0004104962,0.00019503076,0.00043549735,0.00038039437,0.02130467,0.012183491,0.012025512,0.85286224,0.00019951399],"about_ca_topic_score_codex":0.006604872,"about_ca_topic_score_gemma":0.008274305,"teacher_disagreement_score":0.019586494,"about_ca_system_score_codex":0.00050812913,"about_ca_system_score_gemma":0.0016610752,"threshold_uncertainty_score":0.065523386},"labels":[],"label_agreement":null},{"id":"W2724559516","doi":"10.1038/sdata.2017.77","title":"Matched computed tomography segmentation and demographic data for oropharyngeal cancer radiomics challenges","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Radiomics and Machine Learning in Medical Imaging","field":"Medicine","cited_by":82,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Columbia College","funders":"Division of Mathematical Sciences; National Cancer Institute; National Science Foundation; National Institutes of Health; Andrew Sabin Family Foundation; National Institute of Dental and Craniofacial Research; University of Texas MD Anderson Cancer Center; Radiological Society of North America; Varian Medical Systems","keywords":"Radiomics; Head and neck cancer; Medicine; Data extraction; Workflow; Segmentation; Leverage (statistics); Radiation therapy; Medical physics; Computer science; Artificial intelligence; Radiology; MEDLINE; Database; Biology","score_opus":0.09448940482926785,"score_gpt":0.37674825807600376,"score_spread":0.28225885324673594,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2724559516","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5870688,0.0023144402,0.095286235,0.00810946,0.00067704485,0.00093004986,0.2819577,0.00896666,0.014689619],"genre_scores_gemma":[0.6811271,0.00074275356,0.06587657,0.0005916591,0.00023872335,0.00047840227,0.24785481,0.00049529487,0.0025947115],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99864095,0.00024409682,0.00015631778,0.00041977616,0.00042321402,0.00011564202],"domain_scores_gemma":[0.9961442,0.0009436246,0.0006438475,0.0011820307,0.00081401435,0.0002722871],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017681775,0.00040869793,0.0004731211,0.002466273,0.0004482675,0.0011592403,0.00087765726,0.0010777033,0.002351241],"category_scores_gemma":[0.009565775,0.00023298763,0.0006039433,0.0023671878,0.00048285385,0.0007477282,0.0013405107,0.0006054917,0.0016638562],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016201133,0.00055562804,0.47338957,0.000971965,0.00031473863,0.0020780687,0.0011155127,0.042047605,0.021594105,0.014921802,0.16244999,0.27894095],"study_design_scores_gemma":[0.00014305848,0.00040789187,0.38914528,0.00037322473,0.00020582834,0.0046598176,0.0021642118,0.23886971,0.05116423,0.025051113,0.2876309,0.00018468335],"about_ca_topic_score_codex":0.010969009,"about_ca_topic_score_gemma":0.014991191,"teacher_disagreement_score":0.010969009,"about_ca_system_score_codex":0.0013108248,"about_ca_system_score_gemma":0.0021025913,"threshold_uncertainty_score":0.021810353},"labels":[],"label_agreement":null},{"id":"W2734439118","doi":"10.1038/sdata.2017.90","title":"A multi-protease, multi-dissociation, bottom-up-to-top-down proteomic view of the Loxosceles intermedia venom","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Venomous Animal Envenomation and Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"IONICS Mass Spectrometry (Canada)","funders":"National Institutes of Health; Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro; Conselho Nacional de Desenvolvimento Científico e Tecnológico; National Institute of General Medical Sciences; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior","keywords":"Orbitrap; Venom; Trypsin; Proteome; Proteomics; Mass spectrometry; Protease; Chemistry; Computational biology; Tandem mass spectrometry; Chromatography; Shotgun proteomics; Biochemistry; Biology; Enzyme","score_opus":0.05228829954185441,"score_gpt":0.31943914345594576,"score_spread":0.26715084391409133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2734439118","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.53511053,0.05171868,0.3657125,0.006651528,0.00076773536,0.00029588153,0.014833705,0.006756564,0.018152943],"genre_scores_gemma":[0.55051327,0.025110284,0.3976161,0.0024533938,0.00031120403,0.00019504361,0.008819087,0.00058462244,0.014396957],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99986327,0.000010944936,0.000008522073,0.00005036217,0.0000516287,0.000015310165],"domain_scores_gemma":[0.9999212,0.000012913255,0.000016305059,0.000013902818,0.000014690129,0.00002099455],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00020296258,0.0006931495,0.0003888067,0.0008037542,0.0005505922,0.0009762546,0.00040931065,0.00051461655,0.0015008549],"category_scores_gemma":[0.00013083166,0.00039781828,0.0005711428,0.00042094023,0.00031660742,0.0011028028,0.00052957085,0.0013701173,0.00086888135],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000077589946,0.00002024827,0.00077484804,0.00022106525,0.000030334748,0.00026139052,0.000046349945,0.0002614217,0.9866137,0.00060786714,0.00076886686,0.010316332],"study_design_scores_gemma":[0.000032504387,0.00032173484,0.046667233,0.0001640494,0.00020852474,0.0055959327,0.0002788723,0.015490822,0.8565172,0.0062056244,0.068427205,0.00009037706],"about_ca_topic_score_codex":0.0004881744,"about_ca_topic_score_gemma":0.0010204293,"teacher_disagreement_score":0.0015008549,"about_ca_system_score_codex":0.00024106562,"about_ca_system_score_gemma":0.00034543444,"threshold_uncertainty_score":0.0050207973},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"bench_or_experimental","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2739179469","doi":"10.1038/sdata.2017.92","title":"A metagenomic survey of forest soil microbial communities more than a decade after timber harvesting","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Forest Ecology and Biodiversity Studies","field":"Agricultural and Biological Sciences","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Genome British Columbia; Genome Canada","keywords":"Ecology; Metagenomics; Forest ecology; Environmental science; Agroforestry; Amplicon sequencing; Ecosystem; Logging; Geography; Biology","score_opus":0.1239401510660578,"score_gpt":0.2728545170211218,"score_spread":0.148914365955064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2739179469","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99725395,0.00038871248,0.00050786196,0.000021555445,0.000004314157,0.00002331762,0.0013759502,0.000009568939,0.00041483107],"genre_scores_gemma":[0.99169403,0.0006426094,0.0018496243,0.00008306008,0.000011334331,0.00006123023,0.004442925,0.000007908746,0.0012072568],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9998203,0.000016951639,0.000008799259,0.00006728087,0.000045186094,0.000041499785],"domain_scores_gemma":[0.9997564,0.000020008132,0.00006999727,0.000016109892,0.000090401794,0.000047103873],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00025737204,0.00022962884,0.00025736567,0.0005938428,0.0006384967,0.00050099636,0.00017983875,0.00030099202,0.0003368359],"category_scores_gemma":[0.00024891124,0.00013366909,0.00020819591,0.00088258827,0.00020692381,0.00033725667,0.00026738134,0.00034442308,0.00015620264],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044148383,0.00014278569,0.35996735,0.00011969451,0.00008433858,0.00022640213,0.0007432947,0.00021252019,0.6154957,0.000047079346,0.00017003228,0.02234933],"study_design_scores_gemma":[0.000002564486,0.00013391576,0.9839166,0.000009884037,0.00002757582,0.00009042574,0.00035089595,0.00014050739,0.014327755,0.00002477743,0.000969481,0.0000054460465],"about_ca_topic_score_codex":0.0062382994,"about_ca_topic_score_gemma":0.019036295,"teacher_disagreement_score":0.0062382994,"about_ca_system_score_codex":0.00030670853,"about_ca_system_score_gemma":0.00055445294,"threshold_uncertainty_score":0.012404025},"labels":[],"label_agreement":null},{"id":"W2739974442","doi":"10.1038/sdata.2017.97","title":"Morphology of Dbx1 respiratory neurons in the preBötzinger complex and reticular formation of neonatal mice","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neuroscience of respiration and sleep","field":"Neuroscience","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Women and Children’s Health Research Institute; University of Alberta","funders":"National Institute of Neurological Disorders and Stroke; National Heart, Lung, and Blood Institute; National Institutes of Health","keywords":"Reticular connective tissue; Brainstem; Neuroscience; Reticular formation; Biology; Respiratory system; Biological neural network; Medulla; Medulla oblongata; Neuron; Anatomy; Central nervous system","score_opus":0.18026719664256857,"score_gpt":0.34651966745746965,"score_spread":0.16625247081490108,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2739974442","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97818357,0.0015291545,0.0128311645,0.0002136663,0.000041770185,0.0000487896,0.0025212914,0.00031744383,0.0043131555],"genre_scores_gemma":[0.95434135,0.0016412552,0.023302292,0.0001774894,0.000012588818,0.00018797263,0.002606763,0.0005822407,0.01714795],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99972934,0.00001894454,0.000028721031,0.00010996948,0.00007488411,0.000038115173],"domain_scores_gemma":[0.9994437,0.00011218564,0.00018784494,0.000047266138,0.000049490576,0.00015952854],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002504543,0.00030809772,0.00025446698,0.0011949121,0.00028390638,0.0005069117,0.0005026473,0.0005424719,0.002379647],"category_scores_gemma":[0.00030734693,0.0005017125,0.0003034446,0.00021169413,0.00067485846,0.00040140987,0.00039736458,0.0011688654,0.0005677352],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007994307,0.000021223876,0.00088395254,0.0000293094,0.000009108978,0.00011088163,0.00008632324,0.00019199178,0.99552286,0.00047740195,0.00006138415,0.002525608],"study_design_scores_gemma":[0.000040969084,0.00031879547,0.07169423,0.00005720263,0.00005147297,0.0014040016,0.0002185815,0.0032172478,0.9180637,0.00051531784,0.004388079,0.000030406172],"about_ca_topic_score_codex":0.0019200048,"about_ca_topic_score_gemma":0.0037202607,"teacher_disagreement_score":0.002379647,"about_ca_system_score_codex":0.00065319677,"about_ca_system_score_gemma":0.00026103322,"threshold_uncertainty_score":0.007960737},"labels":[],"label_agreement":null},{"id":"W2744041574","doi":"10.1038/sdata.2017.101","title":"Long-term chloride concentrations in North American and European freshwater lakes","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Smart Materials for Construction","field":"Environmental Science","cited_by":64,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Alberta Biodiversity Monitoring Institute; University of Waterloo; Ministry of the Environment, Conservation and Parks; Queen's University","funders":"U.S. Geological Survey; Ministry of Environment; Analyses et Expérimentations pour les Ecosystèmes; Global Lake Ecological Observatory Network; Suomen Ympäristökeskus; Institut National de la Recherche Agronomique; Alberta Environment and Parks; U.S. Environmental Protection Agency; Watershed Watch Salmon Society; Wisconsin Department of Natural Resources; National Science Foundation","keywords":"Impervious surface; Environmental science; Water quality; Chloride; Limnology; Precipitation; Drainage basin; Hydrology (agriculture); Freshwater ecosystem; Ecosystem; Watershed; Ecology; Lake ecosystem; Physical geography; Geography; Geology; Chemistry; Biology","score_opus":0.028046134789227534,"score_gpt":0.2566401330719734,"score_spread":0.22859399828274587,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2744041574","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99745065,0.00018232253,0.00011723308,0.00003190324,0.0000016435785,0.000003869446,0.0017153419,0.000007743237,0.00048932276],"genre_scores_gemma":[0.99393976,0.00017934029,0.0002877096,0.00004343323,0.000004785829,0.000019566678,0.005120451,0.0000052278997,0.00039974006],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99937636,0.0001262104,0.00008253086,0.00019390548,0.000148195,0.000072760406],"domain_scores_gemma":[0.99883574,0.00016232723,0.00043623892,0.00007907863,0.00038313534,0.000103477476],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008590549,0.00018519202,0.0002846718,0.001051341,0.00032794083,0.00066720543,0.0003664896,0.00042299993,0.00047736923],"category_scores_gemma":[0.0012973634,0.00015260892,0.00024413038,0.0021350353,0.0002891887,0.0006464244,0.0006723557,0.0001372245,0.0001568863],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017019967,0.00004785937,0.9910993,0.0000306582,0.00010962518,0.00010408107,0.00029511508,0.00082990597,0.0011862277,0.000037554717,0.00056747184,0.005521987],"study_design_scores_gemma":[0.0000048658203,0.000019761243,0.99836415,0.000005101973,0.000016268494,0.000034724293,0.00018659099,0.00041304034,0.0002620956,0.000012905965,0.00067473453,0.000005794186],"about_ca_topic_score_codex":0.047565185,"about_ca_topic_score_gemma":0.09646471,"teacher_disagreement_score":0.95243484,"about_ca_system_score_codex":0.00069722725,"about_ca_system_score_gemma":0.00040442578,"threshold_uncertainty_score":0.09457666},"labels":[],"label_agreement":null},{"id":"W2750304043","doi":"10.1038/sdata.2017.112","title":"FANTOM5 CAGE profiles of human and mouse samples","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Immunotherapy and Immune Responses","field":"Immunology and Microbiology","cited_by":362,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Child and Family Research Institute; University of British Columbia","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; Diamantina Institute, University of Queensland; RIKEN; Tokyo Medical and Dental University; Biotechnology and Biological Sciences Research Council; Universitätsklinikum Regensburg; Deutsches Zentrum für Neurodegenerative Erkrankungen; Karolinska Institutet; Universiteit van Amsterdam; International Centre for Genetic Engineering and Biotechnology; University of Queensland; University of Cape Town; Ministry of Education, Culture, Sports, Science and Technology; South African Medical Research Council; Medical Research Council; University of Texas Health Science Center at Houston; Wellcome Trust; National Institute on Deafness and Other Communication Disorders; Child and Family Research Institute; National Cancer Institute; Eidgenössische Technische Hochschule Zürich; Melanoma Research Alliance","keywords":"Biology; Computational biology; Transcription (linguistics); Cage; Genome; Promoter; Gene; Pipeline (software); Workflow; Gene expression; Cell biology; Genetics; Computer science; Database","score_opus":0.08039608286432703,"score_gpt":0.3263989378373338,"score_spread":0.24600285497300678,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2750304043","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21744947,0.004085904,0.0600165,0.00042508848,0.00030442496,0.00034206008,0.6946157,0.007804733,0.014956105],"genre_scores_gemma":[0.11628291,0.0015386379,0.08645102,0.000565216,0.00006211186,0.0007113107,0.78613424,0.002588292,0.0056662955],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9989888,0.00012441177,0.000067962486,0.00042952813,0.00028816413,0.00010102593],"domain_scores_gemma":[0.99912745,0.00024822296,0.00011594832,0.00017459337,0.00022210367,0.0001117632],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013331986,0.0012833696,0.0007930192,0.0039784117,0.0006365879,0.0010220585,0.000861171,0.00077158137,0.008259258],"category_scores_gemma":[0.0033414902,0.0005132449,0.0011934646,0.0025141514,0.00021573355,0.00038399218,0.001199104,0.0006062545,0.004478367],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025970973,0.0001038343,0.03511606,0.00493632,0.0010195517,0.0016112254,0.0010570484,0.0021255522,0.8009842,0.0031202147,0.06351403,0.08381491],"study_design_scores_gemma":[0.00016520369,0.0005514339,0.30163902,0.0004359182,0.0012067901,0.008093179,0.0006387193,0.009206178,0.24365021,0.0039427266,0.43010584,0.00036479754],"about_ca_topic_score_codex":0.0029823186,"about_ca_topic_score_gemma":0.0067168814,"teacher_disagreement_score":0.008259258,"about_ca_system_score_codex":0.00051590847,"about_ca_system_score_gemma":0.0007602061,"threshold_uncertainty_score":0.027629972},"labels":[],"label_agreement":null},{"id":"W2751862616","doi":"10.1038/sdata.2017.105","title":"RefEx, a reference gene expression dataset as a web tool for the functional analysis of genes","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":73,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Bioscience Database Center; Institute of Genetics; Japan Science and Technology Agency; Ministry of Education, Culture, Sports, Science and Technology","keywords":"Annotation; Gene; Computational biology; Metadata; Gene ontology; Gene Annotation; Gene nomenclature; Expression (computer science); Biology; Gene expression; Information retrieval; Gene expression profiling; Computer science; Database; Bioinformatics; Genetics; World Wide Web; Genome","score_opus":0.12038063034953955,"score_gpt":0.36283514841942693,"score_spread":0.24245451806988738,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2751862616","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004915937,0.0006285298,0.023851689,0.00012209475,0.00014348187,0.0001547192,0.94765043,0.02004986,0.0024832725],"genre_scores_gemma":[0.0026243834,0.00019394675,0.018459737,0.000071157876,0.00001277992,0.00048841914,0.9767289,0.0007909319,0.00062965596],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980696,0.00024965932,0.00033026232,0.0006376834,0.0005499993,0.00016278676],"domain_scores_gemma":[0.99820983,0.000460927,0.00022120307,0.0005518039,0.0003850899,0.00017118125],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021064568,0.0026088455,0.0020937806,0.007135678,0.0012357385,0.001718559,0.0030244282,0.001746575,0.013668936],"category_scores_gemma":[0.003504599,0.0009936899,0.001975087,0.0071711624,0.00032397162,0.0013679353,0.0023005162,0.0020590012,0.012278664],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016154881,0.00042684143,0.014565575,0.00664614,0.0008053745,0.0010540198,0.00045726696,0.006227189,0.10414249,0.0076555056,0.76523906,0.091165036],"study_design_scores_gemma":[0.00059122726,0.0002591289,0.041647177,0.0004937489,0.0003606453,0.0012895869,0.00022329339,0.012584427,0.03594288,0.0076865572,0.89863896,0.00028232505],"about_ca_topic_score_codex":0.005088672,"about_ca_topic_score_gemma":0.0076838396,"teacher_disagreement_score":0.013668936,"about_ca_system_score_codex":0.0009087936,"about_ca_system_score_gemma":0.002314484,"threshold_uncertainty_score":0.045727134},"labels":[],"label_agreement":null},{"id":"W2755202310","doi":"10.1038/sdata.2017.127","title":"Machine-learned and codified synthesis parameters of oxide materials","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":175,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Office of Naval Research; Natural Sciences and Engineering Research Council of Canada; National Science Foundation","keywords":"Computer science; Scale (ratio); Artificial intelligence; Computation; Machine learning; Programming language","score_opus":0.08278359054886814,"score_gpt":0.3268148658449304,"score_spread":0.24403127529606228,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2755202310","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.66243744,0.004603498,0.11484763,0.0006397953,0.00022574326,0.00025393817,0.18820645,0.018878035,0.009907501],"genre_scores_gemma":[0.63303083,0.0016851698,0.1411334,0.000088203495,0.000087468856,0.0004413564,0.22071905,0.0010282543,0.0017863994],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995229,0.000063816515,0.000052238433,0.00018122818,0.00015102248,0.000028883793],"domain_scores_gemma":[0.9982185,0.00092132273,0.00020408386,0.00031935124,0.00029676533,0.000039981405],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006689761,0.00091053697,0.00053475046,0.003067032,0.0003278783,0.0010608692,0.00073977135,0.0006815179,0.0014878474],"category_scores_gemma":[0.0039175837,0.00035360965,0.0007759199,0.0024176892,0.00044869783,0.001440161,0.0004326488,0.00070262526,0.0010053536],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000674745,0.00034327558,0.058957234,0.0035446652,0.00046210783,0.0005493119,0.00042702397,0.59937376,0.05947927,0.0069645587,0.03846785,0.23075616],"study_design_scores_gemma":[0.00009933604,0.00016058059,0.026637837,0.00010166516,0.00016352787,0.00020744142,0.00020538749,0.8312251,0.07198117,0.019174112,0.04996026,0.000083593906],"about_ca_topic_score_codex":0.0030018275,"about_ca_topic_score_gemma":0.0070958915,"teacher_disagreement_score":0.003067032,"about_ca_system_score_codex":0.00079456944,"about_ca_system_score_gemma":0.0009143806,"threshold_uncertainty_score":0.0059687495},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2765275410","doi":"10.1038/sdata.2017.159","title":"A compendium of geochemical information from the Saanich Inlet water column","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and coastal ecosystems","field":"Earth and Planetary Sciences","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Genome British Columbia; Fisheries and Oceans Canada; Concordia University; University of British Columbia","funders":"University of Victoria","keywords":"Biogeochemical cycle; Water column; Oceanography; Anoxic waters; Biogeochemistry; Fjord; Environmental science; Compendium; Oxygen minimum zone; Hydrology (agriculture); Geology; Environmental chemistry; Chemistry; Geography","score_opus":0.031415893320049106,"score_gpt":0.22344968093053444,"score_spread":0.19203378761048534,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2765275410","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0046394733,0.00066363555,0.00084412843,0.00006239082,0.00004020704,0.000036002304,0.9899366,0.000427076,0.00335052],"genre_scores_gemma":[0.007963006,0.0008664544,0.004525366,0.00006273388,0.000017048582,0.000096788375,0.98479617,0.00011785172,0.0015545164],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99919564,0.000043969314,0.00015937643,0.00016432788,0.0003754124,0.00006126067],"domain_scores_gemma":[0.99481356,0.0007788322,0.0005321073,0.0006203277,0.0029035495,0.00035158003],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00080067886,0.0011788252,0.0010576703,0.010565792,0.0007218162,0.0014359634,0.0010799803,0.0005216298,0.012734884],"category_scores_gemma":[0.0030984625,0.00052692916,0.00046456142,0.019227676,0.0004627864,0.00058660907,0.001253612,0.0006996485,0.0092233345],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025273435,0.00008402939,0.10329994,0.0066568954,0.0004646289,0.0009032296,0.0010793013,0.004586005,0.014065143,0.0018025733,0.6424694,0.22433606],"study_design_scores_gemma":[0.000031399108,0.00002297081,0.12767383,0.00057530054,0.000104075356,0.00020384662,0.00035005048,0.0010747418,0.0026005977,0.0007169074,0.8665722,0.00007389851],"about_ca_topic_score_codex":0.27779406,"about_ca_topic_score_gemma":0.46395925,"teacher_disagreement_score":0.72220594,"about_ca_system_score_codex":0.002066183,"about_ca_system_score_gemma":0.01017529,"threshold_uncertainty_score":0.5523542},"labels":[],"label_agreement":null},{"id":"W2765768951","doi":"10.1038/sdata.2017.157","title":"Small non-coding RNA transcriptome of the NCI-60 cell line panel","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Occupational Cancer Research Centre; BC Cancer Agency","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; Canadian Institutes of Health Research; National Institutes of Health; University of British Columbia","keywords":"Small nucleolar RNA; Biology; microRNA; Transcriptome; Computational biology; Gene; RNA; Genetics; Untranslated region; Non-coding RNA; Gene expression","score_opus":0.07912659260418488,"score_gpt":0.2810783434949748,"score_spread":0.20195175089078993,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2765768951","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7417935,0.007419901,0.019316368,0.0004981547,0.00025537625,0.00043480916,0.1951662,0.001304111,0.03381156],"genre_scores_gemma":[0.59030974,0.0057561374,0.021100003,0.00055446604,0.00006190216,0.0006619167,0.36460212,0.00030937255,0.016644336],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9997845,0.000010309204,0.000016471831,0.00006208749,0.00009257399,0.000034090514],"domain_scores_gemma":[0.9999021,0.000016545124,0.000006982301,0.000015527563,0.000041336567,0.000017356511],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00012342965,0.00030720985,0.00060224935,0.0007813158,0.00051474926,0.00042778984,0.00018465916,0.00016583546,0.0032567035],"category_scores_gemma":[0.0002238076,0.00009432679,0.00036537103,0.00089964,0.00013205054,0.00012167559,0.00030676264,0.0004115773,0.001959799],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023547337,0.000033077245,0.0023221546,0.00007858096,0.000019222929,0.00021787532,0.00006820139,0.0001683679,0.9865197,0.00016382553,0.001977595,0.008195951],"study_design_scores_gemma":[0.000076143486,0.00055058626,0.18699466,0.00006119083,0.00025237477,0.0020566047,0.0003299536,0.0047344854,0.71005845,0.0004688533,0.09434094,0.000075611555],"about_ca_topic_score_codex":0.008852746,"about_ca_topic_score_gemma":0.009084247,"teacher_disagreement_score":0.008852746,"about_ca_system_score_codex":0.00024823743,"about_ca_system_score_gemma":0.00064835965,"threshold_uncertainty_score":0.017602444},"labels":[],"label_agreement":null},{"id":"W2765918031","doi":"10.1038/sdata.2017.158","title":"Monitoring microbial responses to ocean deoxygenation in a model oxygen minimum zone","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Microbial Community Ecology and Physiology","field":"Environmental Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia; University of British Columbia; Canadian Institute for Advanced Research","funders":"","keywords":"Biogeochemistry; Anoxic waters; Climate change; Deoxygenation; Ecology; Marine ecosystem; Ecosystem; Global change; Environmental science; Oceanography; Environmental resource management; Biology; Geology","score_opus":0.07998527387381889,"score_gpt":0.3227257224829786,"score_spread":0.24274044860915972,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2765918031","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9989146,0.000020331623,0.00057118305,0.000029408387,0.0000025302568,0.000003981621,0.00030106233,0.0000152125995,0.0001416922],"genre_scores_gemma":[0.99765575,0.00005526654,0.001446508,0.000015272793,0.0000020070815,0.000017320557,0.00064614887,0.000005806996,0.00015577624],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99990225,0.000017175487,0.0000050435533,0.000041663756,0.000016971371,0.000017004097],"domain_scores_gemma":[0.9998323,0.000056335408,0.00003683003,0.000022711387,0.000017831371,0.000033907967],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00020477305,0.0003010592,0.00032517192,0.00022641235,0.00028093965,0.00043496356,0.0004825049,0.00044178552,0.00041534784],"category_scores_gemma":[0.0003610603,0.00011813191,0.00032032363,0.0003095321,0.00025641487,0.00039719968,0.0005299536,0.00040567084,0.000073747506],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027986043,0.00087670743,0.10412816,0.00014744971,0.00014372422,0.00020739695,0.00025204837,0.1446109,0.73867494,0.0011610755,0.00044755742,0.0065514585],"study_design_scores_gemma":[0.0004123391,0.0033460092,0.11868645,0.000015744603,0.00016511748,0.00008980716,0.00045621392,0.67051506,0.20267181,0.0020342995,0.0015146175,0.00009258693],"about_ca_topic_score_codex":0.0059611653,"about_ca_topic_score_gemma":0.004468486,"teacher_disagreement_score":0.0059611653,"about_ca_system_score_codex":0.00047603785,"about_ca_system_score_gemma":0.0002617597,"threshold_uncertainty_score":0.01185298},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"bench_or_experimental","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2766255725","doi":"10.1038/sdata.2017.160","title":"A compendium of multi-omic sequence information from the Saanich Inlet water column","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Microbial Community Ecology and Physiology","field":"Environmental Science","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Genome British Columbia; Concordia University; University of British Columbia","funders":"Pacific Northwest National Laboratory; Biological and Environmental Research; Natural Sciences and Engineering Research Council of Canada; Office of Science; Consejo Nacional de Ciencia y Tecnología; Genome British Columbia; Joint Genome Institute; Battelle; Canadian Institute for Advanced Research; Tula Foundation; U.S. Department of Energy","keywords":"Biogeochemical cycle; Water column; Anoxic waters; Oceanography; Fjord; Compendium; Climate change; Microbial population biology; Environmental science; Biogeochemistry; Ecology; Inlet; Metagenomics; Biology; Geology; Geography","score_opus":0.08973939690447069,"score_gpt":0.2984716612417852,"score_spread":0.2087322643373145,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2766255725","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03742701,0.0052290717,0.010756063,0.00026328262,0.00019851037,0.00021804075,0.93648946,0.0020249623,0.007393718],"genre_scores_gemma":[0.021065757,0.002189553,0.036413085,0.00021142056,0.000038709713,0.00027285537,0.936214,0.0006068037,0.002987768],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989127,0.000061950814,0.00014076759,0.00025794067,0.00052506,0.00010164193],"domain_scores_gemma":[0.9964216,0.0005949816,0.00045207085,0.00036736872,0.0018298328,0.0003341847],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012529686,0.0016609771,0.0012406146,0.008454837,0.0017430203,0.0022962398,0.0010492937,0.00078583695,0.00675802],"category_scores_gemma":[0.0031686225,0.00062874355,0.0008840629,0.010971446,0.0005483478,0.0005910607,0.0013050193,0.0014126962,0.005396942],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011383778,0.00016496805,0.06224927,0.010840804,0.0012147833,0.00322218,0.0032818017,0.003891891,0.3582341,0.0014570854,0.16363256,0.3906722],"study_design_scores_gemma":[0.000071349736,0.00012203317,0.21689197,0.0016452482,0.00068591704,0.0010602531,0.00090009125,0.0029394855,0.022087773,0.0010939285,0.75225717,0.00024474613],"about_ca_topic_score_codex":0.12618788,"about_ca_topic_score_gemma":0.32004827,"teacher_disagreement_score":0.87381214,"about_ca_system_score_codex":0.001994914,"about_ca_system_score_gemma":0.010995106,"threshold_uncertainty_score":0.2509067},"labels":[],"label_agreement":null},{"id":"W2766580836","doi":"10.1038/sdata.2017.140","title":"Targeted metabolomics and medication classification data from participants in the ADNI1 cohort","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; National Institutes of Health; Genentech; National Institute of Neurological Disorders and Stroke; IXICO; H. Lundbeck A/S; Servier; Eisai; Northern California Institute for Research and Education; University of California, San Diego; Pfizer; Biogen; BioClinica; F. Hoffmann-La Roche; University of Southern California; Eli Lilly and Company; U.S. Department of Defense; Meso Scale Diagnostics; Alzheimer's Disease Neuroimaging Initiative; Novartis Pharmaceuticals Corporation; Bristol-Myers Squibb; National Institute on Aging; Alzheimer's Association; Foundation for the National Institutes of Health","keywords":"Cohort; Metabolomics; Cohort study; Medicine; Computational biology; Data science; Bioinformatics; Computer science; Internal medicine; Biology","score_opus":0.1522656983292253,"score_gpt":0.3638493541484873,"score_spread":0.211583655819262,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2766580836","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82749635,0.00028652145,0.0019638485,0.00025999726,0.000024215151,0.00012228818,0.16836171,0.00015204374,0.0013330502],"genre_scores_gemma":[0.64811426,0.00032501094,0.005960214,0.00019329747,0.000049071983,0.00045742063,0.34331226,0.00007141055,0.001517184],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9995604,0.00007045794,0.000056338034,0.00015551009,0.000109668974,0.000047567915],"domain_scores_gemma":[0.9990959,0.00013480688,0.00021909484,0.00030486262,0.00015405842,0.00009118141],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008067875,0.00057370437,0.0005535318,0.001257316,0.00041674293,0.0006418925,0.0004864719,0.0005757509,0.0017026359],"category_scores_gemma":[0.0027041426,0.00021738872,0.00051842775,0.0017687161,0.00019829931,0.00018733619,0.00068790524,0.00052448636,0.0007268591],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0039725723,0.00041843322,0.9300265,0.00024591043,0.0009030735,0.0007414396,0.00032711928,0.0021413341,0.0120358765,0.00049512857,0.020187499,0.028505055],"study_design_scores_gemma":[0.00012489448,0.00018063957,0.98564583,0.000023057433,0.00020873861,0.0007514547,0.0001349481,0.0012658723,0.0024829754,0.0005838073,0.008556359,0.00004132258],"about_ca_topic_score_codex":0.010758939,"about_ca_topic_score_gemma":0.013125951,"teacher_disagreement_score":0.010758939,"about_ca_system_score_codex":0.0004199567,"about_ca_system_score_gemma":0.00056921085,"threshold_uncertainty_score":0.021392643},"labels":[],"label_agreement":null},{"id":"W2772764433","doi":"10.1038/sdata.2017.187","title":"Temporally inter-comparable maps of terrestrial wilderness and the Last of the Wild","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":144,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Northern British Columbia","funders":"Australian Research Council; Wildlife Conservation Society","keywords":"Wilderness; Wilderness area; Natural (archaeology); Geography; Biodiversity; Habitat; Disturbance (geology); Ecology; Underpinning; Environmental resource management; Environmental science; Archaeology; Biology; Geology","score_opus":0.06972847364218235,"score_gpt":0.28492402393358307,"score_spread":0.21519555029140072,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2772764433","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9165569,0.0007279577,0.008738346,0.00014884023,0.000042162792,0.00010050713,0.047433462,0.000403927,0.025847914],"genre_scores_gemma":[0.9680321,0.00029548953,0.013542509,0.000017812772,0.000026267562,0.00008624284,0.01613199,0.00005160297,0.0018160157],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998845,0.000023090537,0.000007804143,0.000035606645,0.00003140846,0.000017555036],"domain_scores_gemma":[0.9995603,0.000091061345,0.000121379395,0.000042345637,0.00012739937,0.00005753171],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00018665471,0.0001290412,0.00010799126,0.002746398,0.00020726264,0.00061730156,0.00017082553,0.00014629743,0.003511774],"category_scores_gemma":[0.0007954725,0.00007446649,0.00017117936,0.0036244276,0.00022610968,0.00032531298,0.0004977038,0.0001589695,0.00034949806],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008297707,0.00017459349,0.69538236,0.0007056317,0.00032650348,0.000945199,0.005480154,0.038731094,0.02065112,0.009772694,0.021281984,0.2057189],"study_design_scores_gemma":[0.000017804803,0.000037420916,0.96210945,0.00006420047,0.000027998869,0.00019629985,0.0010485847,0.005562717,0.0006885941,0.0007402375,0.029486049,0.000020655843],"about_ca_topic_score_codex":0.018573139,"about_ca_topic_score_gemma":0.041147415,"teacher_disagreement_score":0.018573139,"about_ca_system_score_codex":0.0002415861,"about_ca_system_score_gemma":0.00026895708,"threshold_uncertainty_score":0.036930084},"labels":[],"label_agreement":null},{"id":"W2780601052","doi":"10.1038/sdata.2017.181","title":"An open resource for transdiagnostic research in pediatric mental health and learning disorders","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":723,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Baycrest Hospital; University of Toronto","funders":"Stavros Niarchos Foundation; National Institute of Environmental Health Sciences; Child Mind Institute; Strong; Hearst Foundations","keywords":"Biobank; Neuroimaging; Open science; Mental health; Psychology; Psychopathology; Data sharing; Data science; Computer science; Medicine; Psychiatry; Bioinformatics; Alternative medicine","score_opus":0.30330965426452183,"score_gpt":0.46182131394146975,"score_spread":0.1585116596769479,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2780601052","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041812316,0.013362657,0.16449139,0.0706041,0.009403907,0.010063277,0.39051685,0.008643362,0.2911022],"genre_scores_gemma":[0.17143886,0.011271203,0.32889056,0.017842382,0.005587476,0.03258043,0.2575156,0.0070008077,0.16787276],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98643506,0.0066280384,0.0021224795,0.0017353258,0.0022065646,0.00087250845],"domain_scores_gemma":[0.85181737,0.054491702,0.008815278,0.042732324,0.023223836,0.018919518],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.025281476,0.00059575506,0.0013448833,0.008232099,0.002653033,0.0039368933,0.004024622,0.002173034,0.17046379],"category_scores_gemma":[0.087873116,0.00087764923,0.0006796968,0.0090905335,0.0014648262,0.0053967093,0.016484981,0.002365838,0.06440637],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009111835,0.00040496537,0.014403847,0.00085010397,0.0000686861,0.0011778923,0.0014716109,0.00026233337,0.002697251,0.016392313,0.48625314,0.47510663],"study_design_scores_gemma":[0.00024346946,0.00016935957,0.02453783,0.001393936,0.00006431991,0.0014098601,0.0016051466,0.0005280818,0.0017837198,0.018091107,0.95009744,0.000075794385],"about_ca_topic_score_codex":0.0044301003,"about_ca_topic_score_gemma":0.0071398006,"teacher_disagreement_score":0.9959754,"about_ca_system_score_codex":0.0014222981,"about_ca_system_score_gemma":0.012838461,"threshold_uncertainty_score":0.57025814},"labels":[],"label_agreement":null},{"id":"W2780974917","doi":"10.1038/sdata.2017.179","title":"Sequence data and association statistics from 12,940 type 2 diabetes cases and controls","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University Health Centre; McGill University; McGill University and Génome Québec Innovation Centre","funders":"National Institute of Diabetes and Digestive and Kidney Diseases; Wellcome Trust; Steno Diabetes Center Copenhagen; National Human Genome Research Institute; National Heart, Lung, and Blood Institute; National Institute on Aging; Novo Nordisk Fonden; National Institute of General Medical Sciences; National Institute for Health and Care Research; Medical Research Council; National Institutes of Health; Lundbeckfonden; British Heart Foundation","keywords":"Exome; Imputation (statistics); Minor allele frequency; Biology; Genetics; Genetic association; Allele frequency; Indel; Single-nucleotide polymorphism; 1000 Genomes Project; Genome-wide association study; Genotype; Genetic variation; Exome sequencing; Whole genome sequencing; Gene; Computational biology; Genome; Missing data; Phenotype; Statistics","score_opus":0.09003116363464679,"score_gpt":0.3437257837775208,"score_spread":0.25369462014287397,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2780974917","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.80828446,0.0018490816,0.009953091,0.00040370843,0.00016630309,0.0005300072,0.17251961,0.00023109034,0.006062633],"genre_scores_gemma":[0.6948293,0.0014449258,0.010274287,0.0005494143,0.00015641157,0.002305661,0.28196725,0.00021331373,0.008259386],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9966515,0.0006988194,0.00063759374,0.000979605,0.0008268248,0.00020577195],"domain_scores_gemma":[0.9939984,0.0026380818,0.0007568924,0.0013025097,0.0009900111,0.00031405425],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038974076,0.00045579835,0.000786699,0.0034213217,0.0008626275,0.00075980264,0.0005819352,0.0010207954,0.009764225],"category_scores_gemma":[0.0133324405,0.00045168918,0.0007247498,0.0043946225,0.00041874519,0.00035328986,0.00045963132,0.0008933751,0.0021489442],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005057655,0.00045663022,0.91574883,0.00046914772,0.0013950875,0.0014789945,0.0008221045,0.0019974865,0.007761489,0.0018102996,0.021381604,0.041620564],"study_design_scores_gemma":[0.0007911287,0.0008838794,0.9207434,0.00013855779,0.000992799,0.0039711124,0.0006184912,0.0033826346,0.0024395743,0.0028634265,0.06309513,0.000079833524],"about_ca_topic_score_codex":0.007564809,"about_ca_topic_score_gemma":0.004263509,"teacher_disagreement_score":0.009764225,"about_ca_system_score_codex":0.00030014326,"about_ca_system_score_gemma":0.0005588154,"threshold_uncertainty_score":0.032664597},"labels":[],"label_agreement":null},{"id":"W2789589646","doi":"10.1038/sdata.2018.18","title":"A database of chlorophyll a in Australian waters","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and coastal ecosystems","field":"Earth and Planetary Sciences","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Department of Science, Information Technology and Innovation, Queensland Government; Australian Antarctic Division; Directorate for Biological Sciences; School of Civil, Environmental and Mining Engineering, University of Adelaide; South Australian Research and Development Institute; Griffith University; Flinders University; Antarctic Climate and Ecosystems Cooperative Research Centre; University of New South Wales; University of Technology Sydney; Macquarie University; University of Tasmania; Universität Bremen; Australian Government; Commonwealth Scientific and Industrial Research Organisation; Great Barrier Reef Marine Park Authority; University of Queensland; NSW Office of Environment and Heritage; Université de Genève; Australian Institute of Marine Science","keywords":"Phytoplankton; Environmental science; Biomass (ecology); Zooplankton; Trophic level; Chlorophyll a; Abundance (ecology); Oceanography; Chlorophyll; Ecosystem; Food web; Fishery; Ecology; Biology; Nutrient; Botany","score_opus":0.04969863586566379,"score_gpt":0.2530766676749467,"score_spread":0.2033780318092829,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2789589646","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17527656,0.0018520503,0.0058289776,0.0004747481,0.00010912335,0.0005590417,0.78374517,0.0013924458,0.030761823],"genre_scores_gemma":[0.17453662,0.0025175516,0.025569333,0.00053135713,0.000070898575,0.0012172813,0.77489364,0.0002717548,0.020391554],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990783,0.00007194978,0.00018766546,0.00020009278,0.00038687992,0.00007510445],"domain_scores_gemma":[0.99583554,0.0003302089,0.00088248466,0.0005223238,0.0020536843,0.0003756371],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00077583815,0.0005521065,0.0006681717,0.0065441057,0.0008421134,0.0010664073,0.0010368619,0.00050625554,0.010373883],"category_scores_gemma":[0.0047520055,0.00047117888,0.000599958,0.013345985,0.0002434918,0.0009315969,0.0015895708,0.0006840478,0.0053290427],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006384235,0.00029316088,0.27126512,0.006153812,0.0005190707,0.0012915223,0.005224683,0.0038781601,0.024364386,0.004204551,0.30555904,0.37660795],"study_design_scores_gemma":[0.000032814376,0.00007179982,0.5734671,0.00028271807,0.00010343247,0.00033407426,0.00061582227,0.0020112777,0.0022187717,0.001072599,0.41971794,0.00007160432],"about_ca_topic_score_codex":0.08565457,"about_ca_topic_score_gemma":0.079059675,"teacher_disagreement_score":0.08565457,"about_ca_system_score_codex":0.0018395678,"about_ca_system_score_gemma":0.0035877868,"threshold_uncertainty_score":0.17031199},"labels":[],"label_agreement":null},{"id":"W2791403874","doi":"10.1038/sdata.2018.36","title":"Global quantitative analysis of the human brain proteome in Alzheimer’s and Parkinson’s Disease","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Alzheimer's disease research and treatments","field":"Medicine","cited_by":268,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; Emory University; Alzheimer's Disease Research Center, Emory University; Weston Brain Institute; National Institute of Neurological Disorders and Stroke; Alzheimer's Association; Michael J. Fox Foundation for Parkinson's Research","keywords":"Isobaric labeling; Neuroscience; Tandem mass tag; Neuropathology; Human brain; Proteome; Parkinson's disease; Posterior cingulate; Disease; Alzheimer's disease; Alzheimer's Disease Neuroimaging Initiative; Proteomics; Cognitive impairment; Computational biology; Biology; Bioinformatics; Medicine; Quantitative proteomics; Cortex (anatomy); Gene; Cognition; Pathology; Genetics","score_opus":0.12041218200452913,"score_gpt":0.41467726213894535,"score_spread":0.2942650801344162,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2791403874","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98826444,0.0013316716,0.004379877,0.00003185831,0.00000783763,0.000025696098,0.005331253,0.000076786324,0.0005505791],"genre_scores_gemma":[0.9601767,0.0013172302,0.018485662,0.000092046495,0.00002124016,0.00012490929,0.018978745,0.000052376934,0.00075095607],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99976534,0.000029189649,0.000022880737,0.000093061535,0.00007061238,0.00001903683],"domain_scores_gemma":[0.9998311,0.000029983914,0.00005099463,0.000026817392,0.00004172652,0.000019304058],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045281023,0.00029222135,0.00036886803,0.0011755594,0.0002356208,0.0003402874,0.00012498663,0.0001589898,0.00043798235],"category_scores_gemma":[0.00039599097,0.00011134136,0.00027025153,0.0011968155,0.00019701493,0.0001696322,0.00033096262,0.00019687579,0.00019377605],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000875827,0.00006429965,0.037576586,0.00022812671,0.00023306777,0.00015807601,0.00013656187,0.0006270282,0.93616873,0.00015372832,0.00058308506,0.02319498],"study_design_scores_gemma":[0.000029186509,0.00043236674,0.8251932,0.000021056323,0.0002763496,0.0018287597,0.0001467196,0.0037822742,0.1633685,0.0005744918,0.0043158545,0.000031293057],"about_ca_topic_score_codex":0.0007273943,"about_ca_topic_score_gemma":0.0010847332,"teacher_disagreement_score":0.0011755594,"about_ca_system_score_codex":0.00018549086,"about_ca_system_score_gemma":0.00019212342,"threshold_uncertainty_score":0.0023947358},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"bench_or_experimental","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2791916662","doi":"10.1038/sdata.2018.8","title":"Dynamic contrast-enhanced magnetic resonance imaging for head and neck cancers","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"MRI in cancer diagnosis","field":"Medicine","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto; University Health Network","funders":"University of Texas Health Science Center at Houston; Division of Mathematical Sciences; National Cancer Institute; National Institutes of Health; National Science Foundation; Andrew Sabin Family Foundation; National Institute of Dental and Craniofacial Research; University of Texas MD Anderson Cancer Center; Cancer Prevention and Research Institute of Texas; Elekta; University of Texas at Austin","keywords":"DICOM; Generalizability theory; Flip angle; Magnetic resonance imaging; Computer science; Head and neck; Contrast (vision); Software; Dynamic contrast; Scanner; Data set; Nuclear medicine; Cancer imaging; Medical physics; Radiology; Medicine; Artificial intelligence; Cancer; Mathematics","score_opus":0.032866982660721536,"score_gpt":0.3416893605712519,"score_spread":0.30882237791053035,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2791916662","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.65496254,0.16119651,0.103144586,0.0042640623,0.00076047896,0.0018889471,0.0039272998,0.0010573133,0.06879832],"genre_scores_gemma":[0.83127683,0.038076468,0.11618738,0.0008215705,0.0003444866,0.00058018876,0.0030570491,0.00013881836,0.009517288],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99964726,0.0001365488,0.00005291626,0.00003353607,0.00010087305,0.000028890223],"domain_scores_gemma":[0.99953413,0.00018575034,0.00006215854,0.00007354338,0.000115876996,0.00002859139],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013045222,0.00022794004,0.00024054287,0.0023881064,0.00031027812,0.0006539847,0.00036615226,0.00039233902,0.0041778097],"category_scores_gemma":[0.0023434444,0.00017235432,0.00023426578,0.0011035716,0.00029400704,0.0004965412,0.00034336923,0.00035257623,0.001136753],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014092971,0.00034640415,0.071918346,0.002523356,0.00017738818,0.004446361,0.0002587361,0.0009291747,0.1270982,0.0028685792,0.019057736,0.7689665],"study_design_scores_gemma":[0.00026474518,0.001742637,0.50747335,0.000987079,0.0006437493,0.06884456,0.00053580897,0.008746111,0.113910355,0.0061888937,0.2904949,0.0001678857],"about_ca_topic_score_codex":0.0015972385,"about_ca_topic_score_gemma":0.003339493,"teacher_disagreement_score":0.0041778097,"about_ca_system_score_codex":0.00045596677,"about_ca_system_score_gemma":0.0009171369,"threshold_uncertainty_score":0.013976157},"labels":[],"label_agreement":null},{"id":"W2792092426","doi":"10.1038/sdata.2018.40","title":"A suite of global, cross-scale topographic variables for environmental and biodiversity modeling","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":858,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"McGill University; Deutsche Forschungsgemeinschaft; Yale University; National Aeronautics and Space Administration; National Science Foundation","keywords":"Digital elevation model; Landform; Scale (ratio); Terrain; Elevation (ballistics); Physical geography; Topographic Wetness Index; Variable (mathematics); Spatial analysis; Geography; Spatial variability; Cartography; Environmental science; Geology; Remote sensing; Statistics; Mathematics; Geometry","score_opus":0.058784412206377,"score_gpt":0.28447914166515603,"score_spread":0.22569472945877903,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2792092426","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15614587,0.0006394559,0.717315,0.0005679755,0.00024277363,0.00031394375,0.093941085,0.01701859,0.013815228],"genre_scores_gemma":[0.5435704,0.00077774125,0.36291742,0.0001688119,0.00010540722,0.0006398642,0.086222455,0.0015678386,0.0040299986],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998234,0.00003813785,0.000017090184,0.00004330017,0.00005899447,0.000019147185],"domain_scores_gemma":[0.9996315,0.00009538785,0.000040692797,0.000099835466,0.00010743078,0.000025092158],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00057521503,0.0007992317,0.0004911351,0.0014847069,0.00036023185,0.0010355627,0.0009655034,0.00056336034,0.005615673],"category_scores_gemma":[0.0019198585,0.00037869025,0.0011320199,0.0032919091,0.00021931845,0.0010153456,0.00090015755,0.0012217526,0.001796893],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006401205,0.0002947132,0.04635299,0.00021682441,0.00042020605,0.00019895255,0.00018562253,0.76287675,0.0050828015,0.013046719,0.039684016,0.13157643],"study_design_scores_gemma":[0.000032552045,0.000030313235,0.020625954,0.00002609461,0.000042347696,0.00006259575,0.000092204085,0.94486225,0.002060119,0.008049197,0.024071299,0.000045033685],"about_ca_topic_score_codex":0.018483318,"about_ca_topic_score_gemma":0.027227754,"teacher_disagreement_score":0.018483318,"about_ca_system_score_codex":0.00045847023,"about_ca_system_score_gemma":0.0007479517,"threshold_uncertainty_score":0.03675145},"labels":[],"label_agreement":null},{"id":"W2792542239","doi":"10.1038/sdata.2018.22","title":"GlobTherm, a global database on thermal tolerances for aquatic and terrestrial organisms","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Physiological and biochemical adaptations","field":"Environmental Science","cited_by":253,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia; Université du Québec à Rimouski","funders":"Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Universidad de Alcalá; Deutsche Forschungsgemeinschaft","keywords":"Database; Environmental science; Computer science; Ecology; Biology","score_opus":0.062428838694399545,"score_gpt":0.29662381250183206,"score_spread":0.2341949738074325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2792542239","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0041462756,0.0017200336,0.003521322,0.00008284239,0.000048104284,0.000051338622,0.98280454,0.0048964326,0.0027291218],"genre_scores_gemma":[0.0046749045,0.0008401048,0.005858091,0.00008129361,0.000014499196,0.00016807762,0.987324,0.0006842473,0.00035475843],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989506,0.00013443423,0.00028453703,0.0002913949,0.00023529274,0.000103789345],"domain_scores_gemma":[0.9973999,0.00073420064,0.0006186721,0.0005375411,0.00043986063,0.0002699338],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013499773,0.002297185,0.0023708188,0.009713241,0.0006606011,0.0022758273,0.0032356696,0.0017470514,0.013842777],"category_scores_gemma":[0.006305245,0.0008992758,0.0017398413,0.01276704,0.00047002398,0.0029101383,0.002925925,0.001608226,0.01448905],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013117229,0.00017868265,0.03963939,0.029862022,0.0012627945,0.0011641388,0.0018338802,0.010199525,0.016730517,0.011271867,0.7662068,0.12033868],"study_design_scores_gemma":[0.000116885996,0.00005682878,0.037456777,0.0010018519,0.0002662964,0.00055253506,0.00023379261,0.001763494,0.0017771218,0.005344685,0.95131296,0.00011680698],"about_ca_topic_score_codex":0.0066097365,"about_ca_topic_score_gemma":0.0068763425,"teacher_disagreement_score":0.013842777,"about_ca_system_score_codex":0.00087299204,"about_ca_system_score_gemma":0.0020696751,"threshold_uncertainty_score":0.046308756},"labels":[],"label_agreement":null},{"id":"W2793992791","doi":"10.1038/sdata.2018.39","title":"Simplifying research access to genomics and health data with Library Cards","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Data science; Process (computing); Data access; Authentication (law); Protocol (science); Genomics; World Wide Web; Database; Biology; Genome; Medicine; Computer security","score_opus":0.933510998259124,"score_gpt":0.7230294947196095,"score_spread":0.21048150353951445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2793992791","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028592395,0.0010108256,0.8669281,0.014244455,0.0021551945,0.012354347,0.002302369,0.008754205,0.06365819],"genre_scores_gemma":[0.181438,0.0013715547,0.73002857,0.009489579,0.0016870835,0.0173518,0.003546898,0.0014816175,0.053604916],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.83492893,0.106066994,0.02316699,0.009693822,0.019480988,0.0066622407],"domain_scores_gemma":[0.6593868,0.11883561,0.018379834,0.1628569,0.029850433,0.010690463],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.11475225,0.0011651209,0.0017362986,0.004290418,0.0032982577,0.016434528,0.0065848636,0.005047744,0.0341505],"category_scores_gemma":[0.19127786,0.0018366054,0.0018178567,0.0054397886,0.0054570977,0.02082063,0.020333475,0.0054870183,0.034332104],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004020168,0.0015610139,0.014183366,0.0017076934,0.000156839,0.0012818582,0.008588819,0.0027481406,0.009342514,0.40112692,0.10667886,0.44860372],"study_design_scores_gemma":[0.001177834,0.001168668,0.006175059,0.0021838206,0.00021148889,0.0023758807,0.0035394563,0.009761104,0.022770276,0.13574211,0.81442684,0.0004675679],"about_ca_topic_score_codex":0.0014149534,"about_ca_topic_score_gemma":0.0010095917,"teacher_disagreement_score":0.9934151,"about_ca_system_score_codex":0.0033371134,"about_ca_system_score_gemma":0.012874645,"threshold_uncertainty_score":0.6068752},"labels":[],"label_agreement":null},{"id":"W2794106486","doi":"10.1038/sdata.2018.20","title":"A new, short-recorded photoplethysmogram dataset for blood pressure monitoring in China","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Non-Invasive Vital Sign Monitoring","field":"Engineering","cited_by":207,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"B.C. Women's Hospital & Health Centre; BC Children's Hospital; Children's & Women's Health Centre of British Columbia; University of British Columbia","funders":"Division of Graduate Education; Guilin University of Electronic Technology; National Natural Science Foundation of China","keywords":"Photoplethysmogram; Diabetes mellitus; Medicine; Blood pressure; Disease; Data mining; Computer science; Intensive care medicine; Internal medicine","score_opus":0.03882169248641047,"score_gpt":0.2928124533598392,"score_spread":0.2539907608734287,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2794106486","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.37466973,0.0014770929,0.010268405,0.00059408,0.00019785385,0.0006769828,0.6080596,0.0008855168,0.0031708025],"genre_scores_gemma":[0.24752678,0.00051079557,0.00726356,0.00024475282,0.000129198,0.0011355669,0.7411327,0.000042169137,0.0020145418],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993955,0.000098404555,0.00009448071,0.00017202647,0.00015430093,0.00008527339],"domain_scores_gemma":[0.9991986,0.0001581223,0.00011441978,0.00023000452,0.00019914041,0.00009964654],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00072293484,0.0007001984,0.00096450944,0.0012941647,0.00034512978,0.0004524953,0.0010383927,0.0008970004,0.0024347433],"category_scores_gemma":[0.0019720392,0.00017807576,0.0006860188,0.002028443,0.00029012267,0.0002786198,0.00063062046,0.0005078224,0.0010848241],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0067917104,0.0024189642,0.38596722,0.0037923427,0.001574408,0.0034003844,0.00043964118,0.021676311,0.038886573,0.0019291888,0.2554466,0.27767664],"study_design_scores_gemma":[0.00095427583,0.0009127299,0.8814721,0.00016139811,0.0005581556,0.000983362,0.00019043843,0.029470716,0.008392275,0.0014094032,0.075315304,0.0001797815],"about_ca_topic_score_codex":0.018304588,"about_ca_topic_score_gemma":0.022627408,"teacher_disagreement_score":0.018304588,"about_ca_system_score_codex":0.00067118637,"about_ca_system_score_gemma":0.0018849138,"threshold_uncertainty_score":0.036396086},"labels":[],"label_agreement":null},{"id":"W2802068140","doi":"10.1038/sdata.2018.76","title":"An optimal filter for short photoplethysmogram signals","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Non-Invasive Vital Sign Monitoring","field":"Engineering","cited_by":230,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"B.C. Women's Hospital & Health Centre; BC Children's Hospital; Children's & Women's Health Centre of British Columbia; University of British Columbia","funders":"Division of Graduate Education; Guilin University of Electronic Technology; National Natural Science Foundation of China","keywords":"Photoplethysmogram; Chebyshev filter; Filter (signal processing); Computer science; SIGNAL (programming language); Signal processing; Wearable computer; Noise (video); Interference (communication); Artificial intelligence; Computer vision; Telecommunications; Channel (broadcasting); Digital signal processing; Computer hardware; Embedded system","score_opus":0.06778257964544786,"score_gpt":0.31755129122592674,"score_spread":0.24976871158047886,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2802068140","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.40859237,0.0010631849,0.58785033,0.00013292106,0.000108631015,0.00010550863,0.000107936445,0.0003893511,0.001649829],"genre_scores_gemma":[0.7077621,0.00048439455,0.28994417,0.00007347597,0.00003106214,0.00006350778,0.00012742085,0.000047898156,0.0014660554],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997024,0.000044513992,0.00002662035,0.00007061191,0.00012091368,0.000034906534],"domain_scores_gemma":[0.9991387,0.00033087,0.000083320476,0.00005092482,0.0003647072,0.000031440428],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051749265,0.00040581686,0.00033918343,0.00054471043,0.00027664253,0.000515951,0.00021218599,0.00064730446,0.001543858],"category_scores_gemma":[0.0019685351,0.00015991833,0.0004223088,0.0003576565,0.00025872252,0.0005001995,0.00014165981,0.00031780862,0.00047325573],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017610808,0.00021296657,0.005354543,0.00037594742,0.00006752052,0.00018979756,0.00017325026,0.007628135,0.74758023,0.0013793844,0.0007246897,0.23455247],"study_design_scores_gemma":[0.00013304633,0.0016259714,0.053080514,0.00009306715,0.00029736012,0.00093276676,0.00028107842,0.21637815,0.7187725,0.0016796114,0.006626654,0.00009929639],"about_ca_topic_score_codex":0.0017024518,"about_ca_topic_score_gemma":0.0019930995,"teacher_disagreement_score":0.0017024518,"about_ca_system_score_codex":0.00043144514,"about_ca_system_score_gemma":0.0005773046,"threshold_uncertainty_score":0.005164683},"labels":[],"label_agreement":null},{"id":"W2803508903","doi":"10.1038/sdata.2018.97","title":"HormoneBase, a population-level database of steroid hormone levels across vertebrates","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Animal Behavior and Reproduction","field":"Agricultural and Biological Sciences","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Biology; Hormone; Evolutionary biology; Phenotype; Population; Steroid hormone; Database; Bioinformatics; Endocrinology; Genetics; Gene; Demography; Computer science","score_opus":0.16280903919880477,"score_gpt":0.3270080089994612,"score_spread":0.16419896980065643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2803508903","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0037981882,0.0033560488,0.0026840302,0.00010189975,0.00007176753,0.000047996793,0.9848267,0.0019824007,0.003130938],"genre_scores_gemma":[0.00782596,0.0022622559,0.0064201704,0.0001270938,0.000036290225,0.00016781501,0.981736,0.00038253775,0.0010419559],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99899465,0.000078747245,0.00031756214,0.00033240384,0.00021451477,0.000062083716],"domain_scores_gemma":[0.99631566,0.0011595058,0.0008643365,0.00056127337,0.00075512717,0.00034418755],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014611329,0.0011572576,0.0017170954,0.0097258985,0.0006412624,0.0020434659,0.0018045007,0.0010806802,0.021752702],"category_scores_gemma":[0.006239647,0.0007246051,0.00095129537,0.010879751,0.00037995825,0.0017627984,0.0017291484,0.0009431869,0.013402827],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022395106,0.00018978395,0.056291696,0.028432243,0.0016888101,0.0013899565,0.0017485486,0.0031211541,0.04289488,0.0086881565,0.6281565,0.22515878],"study_design_scores_gemma":[0.00011538919,0.00009140415,0.0556558,0.0012343522,0.00063505385,0.00076281396,0.00019994169,0.00053430174,0.0028248893,0.0025847694,0.93525106,0.00011034166],"about_ca_topic_score_codex":0.008074528,"about_ca_topic_score_gemma":0.009733631,"teacher_disagreement_score":0.021752702,"about_ca_system_score_codex":0.0008113149,"about_ca_system_score_gemma":0.0025417383,"threshold_uncertainty_score":0.07277006},"labels":[],"label_agreement":null},{"id":"W2807820258","doi":"10.1038/sdata.2018.107","title":"Warping an atlas derived from serial histology to 5 high-resolution MRIs","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Neurological Institute and Hospital; Western University; McGill University; Douglas Mental Health University Institute","funders":"","keywords":"Atlas (anatomy); Image warping; Histology; Computer science; High resolution; Artificial intelligence; Computer vision; Medicine; Anatomy; Pathology; Remote sensing; Geology","score_opus":0.1620349504297906,"score_gpt":0.3946804462657199,"score_spread":0.2326454958359293,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2807820258","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039562073,0.00039713466,0.9393072,0.00034496206,0.00032561997,0.0004896051,0.0038746106,0.008237179,0.0074616335],"genre_scores_gemma":[0.14042842,0.0007281491,0.84325016,0.00019981198,0.00007045348,0.0006803962,0.005386339,0.0027771643,0.0064790975],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99906164,0.00013854903,0.000105804276,0.00028821098,0.00033100732,0.0000748221],"domain_scores_gemma":[0.9984666,0.00036022346,0.00016624143,0.0004564593,0.0004853019,0.000065130385],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023131226,0.0008530815,0.0006872631,0.0027625128,0.00091207,0.0026159785,0.001126259,0.0012208638,0.010420418],"category_scores_gemma":[0.003995907,0.00093729806,0.0017769776,0.0026800525,0.00080654345,0.0011457583,0.0017054805,0.0017160224,0.0038531227],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000996544,0.00026004226,0.009145962,0.001641314,0.00057430146,0.0014615741,0.0022042915,0.09810767,0.19534138,0.046991386,0.036515146,0.6067604],"study_design_scores_gemma":[0.00015174809,0.001110378,0.033890463,0.000412218,0.00061117753,0.005585765,0.0008759683,0.32154623,0.322941,0.06580077,0.24657808,0.00049622275],"about_ca_topic_score_codex":0.0046248706,"about_ca_topic_score_gemma":0.009782291,"teacher_disagreement_score":0.010420418,"about_ca_system_score_codex":0.0015208189,"about_ca_system_score_gemma":0.0030866312,"threshold_uncertainty_score":0.034859776},"labels":[],"label_agreement":null},{"id":"W2809461164","doi":"10.1038/sdata.2018.110","title":"MEG-BIDS, the brain imaging data structure extended to magnetoencephalography","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":178,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"Horizon 2020 Framework Programme; National Institutes of Health; Medical Research Council; AXA Research Fund; National Institute of Biomedical Imaging and Bioengineering; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Wellcome Trust; Fondation Brain Canada","keywords":"Magnetoencephalography; Metadata; Computer science; Neuroimaging; Data acquisition; Data sharing; Data science; Psychology; Neuroscience; Electroencephalography; World Wide Web; Medicine","score_opus":0.08310562985708746,"score_gpt":0.32680331047269123,"score_spread":0.24369768061560376,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2809461164","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007930114,0.00060341327,0.8087653,0.002711318,0.0007529553,0.0004551897,0.1035739,0.057656344,0.017551547],"genre_scores_gemma":[0.12282546,0.0012151655,0.64361244,0.001991741,0.0005468741,0.0022173456,0.19665511,0.018774064,0.012161766],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99768615,0.000427799,0.00045489357,0.00040112968,0.00086341985,0.00016654642],"domain_scores_gemma":[0.9919869,0.0017575186,0.0005725805,0.0033312268,0.0019066607,0.0004451535],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.0039686984,0.0010603382,0.00092404697,0.0025496024,0.000897722,0.0044108126,0.0030265658,0.0011141696,0.016566494],"category_scores_gemma":[0.022027487,0.0010600489,0.0014601015,0.0034399,0.0009978566,0.005975462,0.0075762873,0.0023067005,0.0109760715],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015612126,0.00014271706,0.009506709,0.0014082552,0.0004298421,0.0011345245,0.0019954068,0.010702173,0.016187524,0.17013606,0.47062024,0.31617534],"study_design_scores_gemma":[0.00022046095,0.00009535155,0.0059921984,0.00026430702,0.00008885046,0.0011721425,0.00034551386,0.027845172,0.015956393,0.18782659,0.7599886,0.00020435714],"about_ca_topic_score_codex":0.0046206582,"about_ca_topic_score_gemma":0.0055344035,"teacher_disagreement_score":0.99697345,"about_ca_system_score_codex":0.00077019003,"about_ca_system_score_gemma":0.0024278841,"threshold_uncertainty_score":0.0554204},"labels":[],"label_agreement":null},{"id":"W2883298629","doi":"10.1038/sdata.2018.141","title":"MANTA2, update of the Mongo database for the analysis of transcription factor binding site alterations","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Children's Hospital; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Norges Forskningsråd; Universitetet i Oslo; European Molecular Biology Laboratory; Western Canada Research Grid; Weston Brain Institute; BC Children's Hospital; Children's Hospital Foundation; Compute Canada; Canadian Institutes of Health Research; Genome Canada","keywords":"DNA binding site; Genome; Binding site; Biology; Transcription factor; 1000 Genomes Project; Computational biology; Genetics; Database; Single-nucleotide polymorphism; Computer science; Gene; Promoter; Gene expression; Genotype","score_opus":0.033160377393625574,"score_gpt":0.2848577802092167,"score_spread":0.2516974028155911,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2883298629","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032812506,0.008882145,0.07264807,0.0009601549,0.000739237,0.00031562336,0.72743356,0.14405423,0.012154529],"genre_scores_gemma":[0.04520654,0.0034447054,0.10122699,0.0006207294,0.00020617481,0.0009721494,0.8270674,0.016199715,0.0050556134],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985929,0.00023888561,0.00029970572,0.00033945756,0.0004096507,0.00011940755],"domain_scores_gemma":[0.9964185,0.001193395,0.0006187247,0.0008218319,0.00043899138,0.00050850806],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031072886,0.0020711145,0.0024855079,0.0072825933,0.00096460176,0.0035129203,0.0042318283,0.0014527681,0.019963557],"category_scores_gemma":[0.008966248,0.0013001796,0.0022430462,0.006189994,0.00043442752,0.0020393187,0.0025220898,0.0017919955,0.015067984],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0066586416,0.0004266292,0.04478595,0.010543712,0.002282222,0.0022214912,0.0012586919,0.007280979,0.055560987,0.016101273,0.60730666,0.2455727],"study_design_scores_gemma":[0.0005696198,0.00022243202,0.022600729,0.00060167204,0.0007997963,0.0023121673,0.00025485374,0.014970413,0.015829757,0.016742969,0.9248027,0.0002928731],"about_ca_topic_score_codex":0.0032041138,"about_ca_topic_score_gemma":0.0040390366,"teacher_disagreement_score":0.019963557,"about_ca_system_score_codex":0.0006927533,"about_ca_system_score_gemma":0.0024509658,"threshold_uncertainty_score":0.0667848},"labels":[],"label_agreement":null},{"id":"W2884066197","doi":"10.1038/sdata.2018.129","title":"Leukocyte Tracking Database, a collection of immune cell tracks from intravital 2-photon microscopy videos","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Immunotherapy and Immune Responses","field":"Immunology and Microbiology","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"SystemsX.ch; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Ground truth; Computer science; Intravital microscopy; Visualization; Tracking (education); Benchmarking; Immune system; Artificial intelligence; Computer vision; Biology; In vivo; Immunology","score_opus":0.02924483496951172,"score_gpt":0.28607389184270215,"score_spread":0.2568290568731904,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2884066197","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14722301,0.011252633,0.1637283,0.00055908115,0.0005134279,0.0015437021,0.63753587,0.02950095,0.008142985],"genre_scores_gemma":[0.10866621,0.0037993286,0.12513609,0.00024096546,0.000078849305,0.0014637808,0.75630265,0.00077489915,0.0035372286],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993414,0.000060457463,0.00010703488,0.00021620815,0.00021325238,0.00006161885],"domain_scores_gemma":[0.9986614,0.00031254167,0.00025450997,0.0002755883,0.0003464774,0.0001494922],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007614147,0.0011080063,0.001284387,0.004401272,0.0006859567,0.0015951891,0.0020007885,0.0016614802,0.0027772393],"category_scores_gemma":[0.0026011406,0.00059586717,0.0007300225,0.0030776658,0.0002450146,0.00076682126,0.0010719784,0.0010068993,0.002794197],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031230524,0.001317301,0.05553776,0.0075009908,0.0008975299,0.0029714513,0.0007722286,0.018733647,0.1985814,0.00428534,0.3261294,0.38015],"study_design_scores_gemma":[0.0005825451,0.0012271673,0.1613185,0.0010976656,0.000985578,0.007891278,0.0008728715,0.11904611,0.1240674,0.005158333,0.5772436,0.00050894724],"about_ca_topic_score_codex":0.00740443,"about_ca_topic_score_gemma":0.01506645,"teacher_disagreement_score":0.00740443,"about_ca_system_score_codex":0.00078090193,"about_ca_system_score_gemma":0.0016417439,"threshold_uncertainty_score":0.014722645},"labels":[],"label_agreement":null},{"id":"W2893400158","doi":"10.1038/sdata.2018.187","title":"Mimi-PAGE, an open-source implementation of the PAGE09 integrated assessment model","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate Change Policy and Economics","field":"Economics, Econometrics and Finance","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Resources for the Future; Alfred P. Sloan Foundation; National Science Foundation","keywords":"Computer science; Documentation; Implementation; License; Source code; Modular design; Python (programming language); Open source; Coding (social sciences); World Wide Web; Software engineering; Database; Programming language; Operating system; Software","score_opus":0.3275247813131548,"score_gpt":0.3978441427766766,"score_spread":0.0703193614635218,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2893400158","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.035723034,0.00017949773,0.6615059,0.0010950456,0.0003939456,0.0006624433,0.032889646,0.22797418,0.039576266],"genre_scores_gemma":[0.16643253,0.00029644536,0.7396481,0.0004829333,0.000098094584,0.0016055392,0.034430787,0.043112192,0.013893293],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99899405,0.00026176346,0.00011437475,0.00019053817,0.0003434017,0.00009581336],"domain_scores_gemma":[0.99502194,0.0026148625,0.00029209332,0.0009466152,0.0009265194,0.00019790069],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029180257,0.00087405165,0.0006407535,0.0009537662,0.0004591958,0.0021757744,0.0035525127,0.0011089145,0.022365708],"category_scores_gemma":[0.014757954,0.0011446942,0.0016766567,0.00094526913,0.0006289905,0.0024902865,0.0026273194,0.0020044213,0.006505558],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011708204,0.00054914504,0.016525485,0.0011872138,0.0005258045,0.0006396569,0.0008674898,0.5101036,0.007430184,0.08924022,0.22997908,0.14178129],"study_design_scores_gemma":[0.0002712667,0.00008020097,0.0015656928,0.00012075916,0.000065560576,0.0001099746,0.000062735904,0.84815824,0.005305107,0.022615716,0.12155681,0.00008790093],"about_ca_topic_score_codex":0.007871112,"about_ca_topic_score_gemma":0.0071964758,"teacher_disagreement_score":0.022365708,"about_ca_system_score_codex":0.0010569626,"about_ca_system_score_gemma":0.002767605,"threshold_uncertainty_score":0.07482076},"labels":[],"label_agreement":null},{"id":"W2895302239","doi":"10.1038/sdata.2018.201","title":"Free-electron laser data for multiple-particle fluctuation scattering analysis","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atomic and Subatomic Physics Research","field":"Physics and Astronomy","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Burnaby Hospital","funders":"SLAC National Accelerator Laboratory; National Institute of Allergy and Infectious Diseases; National Institute of General Medical Sciences; Office of Science; Basic Energy Sciences; National Energy Research Scientific Computing Center; U.S. Department of Energy; National Science Foundation; National Institutes of Health; Advanced Scientific Computing Research; Max-Planck-Gesellschaft","keywords":"Scattering; Diffraction; Optics; Laser; Physics; Computational physics; Materials science","score_opus":0.08264920372629775,"score_gpt":0.3614613696515484,"score_spread":0.27881216592525065,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2895302239","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.058328312,0.00085215527,0.28673023,0.0006617325,0.00093376037,0.0010138861,0.56717247,0.056104958,0.028202549],"genre_scores_gemma":[0.07551082,0.0011458703,0.47670215,0.0003136092,0.00016030054,0.0028198112,0.4202218,0.012297709,0.010827962],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.99906343,0.00009725844,0.00014846458,0.00013412707,0.00046312757,0.0000934587],"domain_scores_gemma":[0.99711156,0.00044689307,0.00017500283,0.0011075486,0.001060285,0.00009875626],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001282182,0.0013751908,0.0013468176,0.0040584356,0.0013761747,0.0013070767,0.0017787201,0.0013026493,0.09308273],"category_scores_gemma":[0.0048462376,0.0006233022,0.00066633296,0.0068621174,0.00036088182,0.0017260615,0.0012512881,0.0025790464,0.026960948],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018334792,0.0009388768,0.0066674286,0.0032531738,0.00020974963,0.0014217547,0.0007863425,0.004607386,0.45394439,0.029675284,0.33003533,0.16662681],"study_design_scores_gemma":[0.00017813992,0.0002372524,0.011969026,0.0002371807,0.00007948295,0.0010327973,0.00026997423,0.028582599,0.3885498,0.010032768,0.55861044,0.00022054542],"about_ca_topic_score_codex":0.0013316466,"about_ca_topic_score_gemma":0.0027758915,"teacher_disagreement_score":0.09308273,"about_ca_system_score_codex":0.0006445463,"about_ca_system_score_gemma":0.0012980037,"threshold_uncertainty_score":0.31139272},"labels":[],"label_agreement":null},{"id":"W2896057756","doi":"10.1038/sdata.2018.209","title":"In-depth data on the network structure and hourly activity of the Central Chilean power grid","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Comisión Nacional de Investigación Científica y Tecnológica; Instituto de Sistemas Complejos de Ingeniería; National Research Foundation of Korea; National Research Foundation; University of Calgary","keywords":"Computer science; Raw data; Grid; Filter (signal processing); Stability (learning theory); Data mining; Distributed computing; Focus (optics); Real-time computing; Data structure; Machine learning; Geography","score_opus":0.03140422013737429,"score_gpt":0.28508400135533163,"score_spread":0.2536797812179573,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2896057756","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18841635,0.00049912394,0.0065480405,0.0007715595,0.0001039797,0.00012754872,0.7912767,0.0014143457,0.010842479],"genre_scores_gemma":[0.22581115,0.0004378069,0.005743782,0.000090331094,0.000058319467,0.00030991953,0.764112,0.00016146562,0.003275189],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99962497,0.00006405321,0.000037814574,0.00009570395,0.00012339505,0.0000539888],"domain_scores_gemma":[0.9979225,0.0003977424,0.00036217674,0.000518429,0.00064893975,0.00015029426],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041177685,0.00045554727,0.00039161387,0.0024547544,0.00033202293,0.00065282336,0.0008984593,0.0005102837,0.0040088366],"category_scores_gemma":[0.0028820292,0.00020426196,0.00036786965,0.005149092,0.00031002826,0.00086754275,0.0008822633,0.00075184356,0.0017247013],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042274862,0.0005115532,0.2754676,0.0021831887,0.00034439928,0.0011839307,0.0017072421,0.07991765,0.0057750028,0.0135271875,0.5071603,0.11179921],"study_design_scores_gemma":[0.000113277216,0.000076212855,0.52246433,0.0003305034,0.00008805824,0.00040245877,0.0018958817,0.055781215,0.0042230384,0.0055700107,0.40890706,0.00014800492],"about_ca_topic_score_codex":0.039899457,"about_ca_topic_score_gemma":0.054017268,"teacher_disagreement_score":0.039899457,"about_ca_system_score_codex":0.0010044029,"about_ca_system_score_gemma":0.0011966781,"threshold_uncertainty_score":0.07933444},"labels":[],"label_agreement":null},{"id":"W2896064861","doi":"10.1038/sdata.2018.229","title":"Combining tensile testing and structural analysis at the single collagen fibril level","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Collagen: Extraction and Characterization","field":"Materials Science","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University; Institut National de la Recherche Scientifique; Dalhousie University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Fibril; Ultimate tensile strength; Microscopy; Materials science; Tensile testing; Collagen fibril; Tendon; Biophysics; Composite material; Biomedical engineering; Anatomy; Optics; Biology","score_opus":0.1541988817359119,"score_gpt":0.3086584682149464,"score_spread":0.15445958647903452,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2896064861","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77373505,0.0018881053,0.17381296,0.00026208383,0.00018404255,0.0003084641,0.03821818,0.0032137395,0.008377478],"genre_scores_gemma":[0.7630731,0.0013749494,0.177457,0.0002866179,0.0001922902,0.00054310745,0.05346001,0.0009071611,0.0027057761],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9989919,0.00008390432,0.00010176852,0.00034440373,0.00035399737,0.00012388837],"domain_scores_gemma":[0.9973954,0.0005496424,0.0004241548,0.0007390868,0.0007620841,0.00012962478],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001054603,0.001321354,0.00086116075,0.0038270913,0.00062815566,0.0009914261,0.00060522545,0.0012460983,0.0023203935],"category_scores_gemma":[0.0017429756,0.00040198886,0.0010658584,0.0019225139,0.00059002475,0.0010183511,0.00087853154,0.0010145637,0.0012454594],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026867533,0.00020350853,0.024409197,0.00065452285,0.0002993342,0.00045669253,0.0001352663,0.004816597,0.9217644,0.0003915606,0.002454482,0.04414573],"study_design_scores_gemma":[0.000036127734,0.0009940218,0.37735543,0.00015373474,0.00050004054,0.0028503179,0.0006097312,0.082726225,0.50697607,0.0051094075,0.02243415,0.00025469912],"about_ca_topic_score_codex":0.0012731792,"about_ca_topic_score_gemma":0.003471466,"teacher_disagreement_score":0.0038270913,"about_ca_system_score_codex":0.00023879998,"about_ca_system_score_gemma":0.00029294213,"threshold_uncertainty_score":0.007762432},"labels":[],"label_agreement":null},{"id":"W2898434771","doi":"10.1038/sdata.2018.240","title":"Meta-mining of copy number profiles of high-risk neuroblastoma tumors","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neuroblastoma Research and Treatments","field":"Medicine","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Neuroblastoma; Copy-number variation; Biology; Segmentation; Cluster (spacecraft); Oncology; Bioinformatics; Internal medicine; Computational biology; Computer science; Medicine; Gene; Genetics; Genome; Artificial intelligence","score_opus":0.10163417764522263,"score_gpt":0.3493303079652971,"score_spread":0.24769613032007448,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2898434771","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.73779696,0.050983105,0.040533014,0.0012006089,0.0002932396,0.0001778432,0.16515756,0.0014627205,0.0023949207],"genre_scores_gemma":[0.8304968,0.005868936,0.020925058,0.00023765405,0.00013747529,0.00016877802,0.14114651,0.0002381634,0.00078059285],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99704546,0.0006892628,0.00041763583,0.0012697993,0.0004311265,0.00014669418],"domain_scores_gemma":[0.99368733,0.004000294,0.00069230533,0.0010264653,0.00036492536,0.00022867264],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034911132,0.00085221377,0.0019787203,0.0074298466,0.0004626025,0.0016245145,0.0010327233,0.0007118976,0.0014254643],"category_scores_gemma":[0.009117483,0.00038640696,0.0028990365,0.0061717113,0.00023930644,0.0006890751,0.0010934939,0.00064032426,0.0004366113],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018209588,0.0001318624,0.812945,0.0040846416,0.028848335,0.0014785255,0.0002619215,0.018503025,0.01970347,0.0008580609,0.008448518,0.10291571],"study_design_scores_gemma":[0.00024720753,0.000624192,0.8165135,0.00084531377,0.036886025,0.007003759,0.00050667214,0.060239747,0.02194733,0.00877196,0.046200667,0.0002138197],"about_ca_topic_score_codex":0.0031100104,"about_ca_topic_score_gemma":0.0058339788,"teacher_disagreement_score":0.0074298466,"about_ca_system_score_codex":0.0005577466,"about_ca_system_score_gemma":0.0011279607,"threshold_uncertainty_score":0.018463016},"labels":[],"label_agreement":null},{"id":"W2899177940","doi":"10.1038/sdata.2018.236","title":"Datasets from a research project examining the role of politics in social psychological research","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Academic Freedom and Politics","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Marsden Fund; Knut och Alice Wallenbergs Stiftelse","keywords":"Politics; Leverage (statistics); Relevance (law); Social research; Empirical research; Perception; Psychology; Psychological research; Publication bias; Test (biology); Social science; Social psychology; Political science; Data science; Sociology; MEDLINE; Epistemology; Computer science; Law; Ecology; Biology","score_opus":0.5869583855402817,"score_gpt":0.5763533724244022,"score_spread":0.01060501311587947,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2899177940","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.062332276,0.0016112266,0.00804207,0.003125916,0.00029619547,0.0016668666,0.9123687,0.0007751894,0.009781581],"genre_scores_gemma":[0.0357341,0.0005140308,0.021029225,0.0007788379,0.00010613059,0.005986197,0.93245155,0.00014421334,0.003255739],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9897607,0.004335786,0.001369389,0.0012256128,0.002796299,0.00051221374],"domain_scores_gemma":[0.93995553,0.033027448,0.0062802606,0.009811009,0.008153081,0.0027726395],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.007674327,0.00085861544,0.0008437511,0.0075479522,0.0023116397,0.0021798508,0.0022101947,0.0034415952,0.010714871],"category_scores_gemma":[0.046601098,0.0005753565,0.0012638253,0.01404067,0.0011715445,0.0012543543,0.003509083,0.0023720192,0.0061010886],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012932995,0.0016635602,0.06650898,0.004144378,0.00044263288,0.00080522033,0.003056524,0.0042718365,0.0019733913,0.012504093,0.8404894,0.062846735],"study_design_scores_gemma":[0.0010803172,0.00030412246,0.1682622,0.000852074,0.00021014357,0.00093581434,0.002202558,0.0044921297,0.002157698,0.013911564,0.8053881,0.00020331214],"about_ca_topic_score_codex":0.01021557,"about_ca_topic_score_gemma":0.030373076,"teacher_disagreement_score":0.99232566,"about_ca_system_score_codex":0.0021830425,"about_ca_system_score_gemma":0.002929728,"threshold_uncertainty_score":0.040586174},"labels":[],"label_agreement":null},{"id":"W2900302900","doi":"10.1038/sdata.2018.244","title":"X-ray computed microtomography of Megachirella wachtleri","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Paleontology and Evolutionary Biology","field":"Earth and Planetary Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Killam Trusts; Natural Sciences and Engineering Research Council of Canada; Midwestern University","keywords":"X-ray microtomography; Computed tomography; Geology; Medicine; Radiology","score_opus":0.03846653945708901,"score_gpt":0.2500899514856179,"score_spread":0.2116234120285289,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2900302900","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99672323,0.00043647774,0.00085211685,0.000018538232,0.0000027485441,0.000013578481,0.00058742485,0.000038374288,0.0013275327],"genre_scores_gemma":[0.9933924,0.00036800603,0.0045625265,0.000020834033,0.000006996606,0.00002479058,0.0010051724,0.000027156899,0.00059210375],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9999062,0.000008251648,0.000009243337,0.000041255644,0.000020479685,0.000014547678],"domain_scores_gemma":[0.99963,0.00007770853,0.00015296004,0.000046708046,0.0000559879,0.000036590172],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00012144643,0.00026152396,0.00014171872,0.0026248053,0.000293548,0.00049896067,0.00024458353,0.0003045613,0.0017932457],"category_scores_gemma":[0.00045932713,0.00035453198,0.00017941042,0.0009342338,0.00039872873,0.00032348026,0.0005548457,0.00024028796,0.00060163235],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043083768,0.000057067547,0.5017244,0.00039795198,0.000093565555,0.002048961,0.002053647,0.0016801549,0.4387661,0.00051132217,0.00047872245,0.05175728],"study_design_scores_gemma":[0.0000046261976,0.00008081774,0.98753196,0.00003923562,0.00003114001,0.002323693,0.00047489122,0.001059368,0.007026759,0.00007695359,0.001335005,0.00001550381],"about_ca_topic_score_codex":0.0035176424,"about_ca_topic_score_gemma":0.0075484407,"teacher_disagreement_score":0.0035176424,"about_ca_system_score_codex":0.00014475509,"about_ca_system_score_gemma":0.00020149905,"threshold_uncertainty_score":0.006994307},"labels":[],"label_agreement":null},{"id":"W2901044913","doi":"10.1038/sdata.2018.263","title":"Generation and quality control of lipidomics data for the alzheimer’s disease neuroimaging initiative cohort","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":66,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Diabetes and Digestive and Kidney Diseases; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; National Institutes of Health; Genentech; National Institute of Neurological Disorders and Stroke; IXICO; H. Lundbeck A/S; Servier; Eisai; Northern California Institute for Research and Education; University of California, San Diego; Pfizer; Biogen; BioClinica; F. Hoffmann-La Roche; University of Southern California; Eli Lilly and Company; U.S. Department of Defense; Meso Scale Diagnostics; Alzheimer's Disease Neuroimaging Initiative; Novartis Pharmaceuticals Corporation; Bristol-Myers Squibb; National Institute on Aging; Alzheimer's Association; Foundation for the National Institutes of Health","keywords":"Lipidomics; Alzheimer's Disease Neuroimaging Initiative; Neuroimaging; Metabolomics; Cohort; Disease; Etiology; Alzheimer's disease; Medicine; Metabolome; Computational biology; Bioinformatics; Internal medicine; Biology; Psychiatry","score_opus":0.21299411330994492,"score_gpt":0.3785598212245248,"score_spread":0.1655657079145799,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2901044913","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.55777496,0.00085571,0.05475458,0.0009254938,0.0001833732,0.003218068,0.3790228,0.0013295728,0.0019354194],"genre_scores_gemma":[0.33308664,0.0003470128,0.054627515,0.00040860928,0.00013481754,0.0037735929,0.60604006,0.0003878034,0.0011938978],"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99201584,0.002081074,0.001186663,0.0022915395,0.0020124675,0.0004123524],"domain_scores_gemma":[0.9846112,0.002305092,0.0024371436,0.006355775,0.0036402724,0.0006504308],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014391093,0.00084031344,0.0010451429,0.0028438899,0.001051472,0.0015188776,0.0020012066,0.0011403718,0.0019743924],"category_scores_gemma":[0.025862392,0.000504472,0.0009506639,0.0028006702,0.0008560613,0.0004039355,0.0017076966,0.0010488069,0.001234579],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007466601,0.00093090243,0.79551643,0.00082913437,0.0027728274,0.0016518126,0.0008967877,0.00934787,0.025208868,0.0033746143,0.074857704,0.07714649],"study_design_scores_gemma":[0.0012736395,0.00076810917,0.8854,0.00021954354,0.0010105447,0.001609959,0.00029492422,0.013910591,0.016825551,0.0029636794,0.07552181,0.00020165111],"about_ca_topic_score_codex":0.010983306,"about_ca_topic_score_gemma":0.009026201,"teacher_disagreement_score":0.014391093,"about_ca_system_score_codex":0.0012830299,"about_ca_system_score_gemma":0.002396846,"threshold_uncertainty_score":0.07610828},"labels":[],"label_agreement":null},{"id":"W2901175140","doi":"10.1038/sdata.2018.257","title":"Anthropomorphic breast model repository for research and development of microwave breast imaging technologies","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Microwave Imaging and Scattering Analysis","field":"Engineering","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Alberta Innovates; Killam Trusts","keywords":"Breast imaging; Microwave imaging; Computer science; Documentation; Mammography; Breast MRI; Medical physics; Microwave; Flexibility (engineering); Imaging technology; Systems engineering; Breast cancer; Radiology; Medicine; Engineering","score_opus":0.0909992141427596,"score_gpt":0.3269898645212841,"score_spread":0.23599065037852449,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2901175140","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007939481,0.0016565036,0.8266454,0.0006603818,0.00051670475,0.0008722196,0.06544562,0.05058431,0.04567942],"genre_scores_gemma":[0.08391087,0.006329212,0.5944486,0.0006685005,0.00018404947,0.002808178,0.2314164,0.019187491,0.061046734],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992156,0.000103035,0.0000906996,0.00007044197,0.0004779809,0.00004225444],"domain_scores_gemma":[0.9977983,0.0006056516,0.0001553728,0.0007267091,0.0006103441,0.000103641454],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013130672,0.0012157428,0.0010247877,0.0025797028,0.0004367526,0.0014841068,0.0029251333,0.0012811885,0.05603188],"category_scores_gemma":[0.004221653,0.00091274636,0.0012133885,0.0020800598,0.00032925123,0.0015467891,0.0017355218,0.0011463566,0.024395848],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004602298,0.0003459945,0.0031325668,0.0022242349,0.00013925298,0.0012001056,0.00034995645,0.13876815,0.020861946,0.033393838,0.4716688,0.32745486],"study_design_scores_gemma":[0.00012508647,0.00009953123,0.0015341379,0.00038944796,0.000079426354,0.0014798951,0.00008201846,0.17419554,0.013485334,0.012466472,0.7959152,0.00014798688],"about_ca_topic_score_codex":0.0023365582,"about_ca_topic_score_gemma":0.0049522487,"teacher_disagreement_score":0.05603188,"about_ca_system_score_codex":0.00062931515,"about_ca_system_score_gemma":0.0014316699,"threshold_uncertainty_score":0.18744534},"labels":[],"label_agreement":null},{"id":"W2901467993","doi":"10.1038/sdata.2018.249","title":"GloPL, a global data base on pollen limitation of plant reproduction","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant and animal studies","field":"Agricultural and Biological Sciences","cited_by":64,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Fundação de Amparo à Pesquisa do Estado de São Paulo; Deutsche Forschungsgemeinschaft; Alexander von Humboldt-Stiftung; Case Western Reserve University; National Science Foundation","keywords":"Pollen; Plant reproduction; Pollination; Biology; Outcrossing; Pollen source; Biome; Stamen; Reproduction; Pollinator; Botany; Abiotic component; Zoophily; Ecology; Ecosystem","score_opus":0.292284396659378,"score_gpt":0.29168242477216005,"score_spread":0.0006019718872179447,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2901467993","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004154897,0.010027814,0.0013576024,0.00019807355,0.000050791543,0.000076247205,0.98027635,0.0010138294,0.0028443777],"genre_scores_gemma":[0.0084131025,0.011192385,0.0037786944,0.00037799776,0.00005436898,0.00039434934,0.9746731,0.00031025536,0.0008056684],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99644035,0.0004014023,0.0015268698,0.0007386401,0.00071522425,0.0001775256],"domain_scores_gemma":[0.9649321,0.016501555,0.008307098,0.003303225,0.005816285,0.001139731],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043386254,0.0020530205,0.0030662664,0.030936621,0.0005681406,0.002908065,0.0029466823,0.0018811156,0.012367815],"category_scores_gemma":[0.021304376,0.0008357896,0.0017600885,0.028264279,0.0006375363,0.00305171,0.0029434355,0.0009766216,0.0094982525],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017199882,0.00015245962,0.09026565,0.14077774,0.0030642285,0.0018345464,0.002315283,0.0025858423,0.014583634,0.0073369504,0.4938654,0.24149829],"study_design_scores_gemma":[0.000121688674,0.00009301303,0.07852859,0.0073433383,0.0013759164,0.000665629,0.00038894615,0.00030035572,0.0017207799,0.0017206387,0.90760547,0.00013561716],"about_ca_topic_score_codex":0.010270555,"about_ca_topic_score_gemma":0.0121431025,"teacher_disagreement_score":0.030936621,"about_ca_system_score_codex":0.0009706245,"about_ca_system_score_gemma":0.0043494115,"threshold_uncertainty_score":0.041374445},"labels":[],"label_agreement":null},{"id":"W2902209191","doi":"10.1038/sdata.2018.270","title":"In-vivo probabilistic atlas of human thalamic nuclei based on diffusion- weighted magnetic resonance imaging","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":118,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre Hospitalier Universitaire Sainte-Justine; Université de Sherbrooke","funders":"Centre d'Imagerie BioMédicale; Université de Lausanne; Université de Genève; Hôpitaux Universitaires de Genève; École Polytechnique Fédérale de Lausanne; Centre Hospitalier Universitaire Vaudois; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Atlas (anatomy); Segmentation; Computer science; Magnetic resonance imaging; Artificial intelligence; Thalamus; Cluster analysis; Diffusion MRI; Diffusion-Weighted Magnetic Resonance Imaging; Pattern recognition (psychology); Probabilistic logic; Brain atlas; Functional magnetic resonance imaging; Neuroimaging; Neuroscience; Biology; Anatomy; Medicine; Radiology","score_opus":0.05710554135325454,"score_gpt":0.3492608401373721,"score_spread":0.29215529878411756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2902209191","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16164559,0.0005542604,0.8263133,0.00021025799,0.00006197836,0.00044379645,0.0052419677,0.0019558794,0.0035729748],"genre_scores_gemma":[0.5592004,0.00085439044,0.42958638,0.00006483818,0.000032805445,0.00051127747,0.0068228804,0.00049422705,0.0024327692],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.9995813,0.00011701331,0.00003437581,0.00014315211,0.000096692296,0.000027601094],"domain_scores_gemma":[0.9995548,0.00014042282,0.00006893905,0.00011984204,0.000090823625,0.000025165647],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00063019834,0.00055521476,0.00046536786,0.0013418333,0.00041976615,0.0013230949,0.0008377954,0.0006902872,0.0029165319],"category_scores_gemma":[0.0017322775,0.00043029388,0.0007548281,0.0013891959,0.000542142,0.0005044677,0.00090004137,0.0004311857,0.0009817017],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010775963,0.00026222004,0.01664814,0.0010641628,0.00046856265,0.0011566452,0.0013386331,0.50336844,0.15598002,0.018712651,0.009465393,0.29045752],"study_design_scores_gemma":[0.00007666042,0.00030793334,0.03474864,0.00012218286,0.00019417731,0.0036456701,0.00036601396,0.85975164,0.0633345,0.01612948,0.021155871,0.00016730114],"about_ca_topic_score_codex":0.0071334075,"about_ca_topic_score_gemma":0.012632737,"teacher_disagreement_score":0.0071334075,"about_ca_system_score_codex":0.00079986145,"about_ca_system_score_gemma":0.0013196077,"threshold_uncertainty_score":0.01418376},"labels":[],"label_agreement":null},{"id":"W2905179958","doi":"10.1038/sdata.2018.286","title":"On the privacy-conscientious use of mobile phone data","year":2018,"lang":"en","type":"editorial","venue":"Scientific Data","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":168,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Belgian Federal Science Policy Office; Fonds De La Recherche Scientifique - FNRS; Agence Nationale de la Recherche; UNICEF; Natural Sciences and Engineering Research Council of Canada; Agence Française de Développement; Bill and Melinda Gates Foundation","keywords":"Mobile phone; Internet privacy; Computer science; Data science; Phone; Field (mathematics); Scale (ratio); Social media; Mobile device; Computational sociology; World Wide Web; Telecommunications; Geography; Cartography","score_opus":0.09751765212507633,"score_gpt":0.32438209284749736,"score_spread":0.226864440722421,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2905179958","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00011597394,0.039470516,0.0006724885,0.18169402,0.77392924,0.000016919996,0.00006529211,0.000066131026,0.0039694253],"genre_scores_gemma":[0.0017256156,0.018780222,0.00022426053,0.05312867,0.92094743,0.000020694602,0.000029452878,0.00006769186,0.005075885],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99181366,0.0023050504,0.00090700644,0.0010969029,0.0034039759,0.0004734695],"domain_scores_gemma":[0.9349944,0.05141765,0.0012515227,0.0017573887,0.008448194,0.0021308516],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011249588,0.0015201913,0.0019371749,0.002365848,0.0044332305,0.008277452,0.003091662,0.020045215,0.0055915527],"category_scores_gemma":[0.048319735,0.0009073409,0.0013898424,0.0014533796,0.008828446,0.008262188,0.0029616316,0.031251844,0.0037792963],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021672304,0.000008159086,0.000037512826,0.00024107129,0.000020022035,0.00015031398,0.00014251299,0.000037901304,0.00005546263,0.0065736403,0.9816481,0.011063626],"study_design_scores_gemma":[0.000018304338,0.000013323028,0.00012122411,0.0005458153,0.000022574855,0.00021170954,0.0000686967,0.00008361677,0.00008676487,0.0048464243,0.99396294,0.000018653525],"about_ca_topic_score_codex":0.0013867987,"about_ca_topic_score_gemma":0.0025059364,"teacher_disagreement_score":0.020045215,"about_ca_system_score_codex":0.002788511,"about_ca_system_score_gemma":0.0029473985,"threshold_uncertainty_score":0.059494197},"labels":[],"label_agreement":null},{"id":"W2908989331","doi":"10.1038/s41597-018-0005-2","title":"Author Correction: A compendium of geochemical information from the Saanich Inlet water column","year":2019,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Geological Studies and Exploration","field":"Earth and Planetary Sciences","cited_by":186,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia; Fisheries and Oceans Canada; Concordia University; Canadian Institute for Advanced Research; University of British Columbia","funders":"","keywords":"Compendium; Table (database); Inlet; Column (typography); Water column; Geology; Chemistry; Oceanography; Mathematics; Computer science; Database; History; Archaeology","score_opus":0.04358805088912733,"score_gpt":0.2293282728724539,"score_spread":0.18574022198332657,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2908989331","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00052553334,0.0010257278,0.0038155974,0.025306331,0.92967004,0.000060540595,0.02238054,0.0017245564,0.01549121],"genre_scores_gemma":[0.020085867,0.007012729,0.023818916,0.04348011,0.11639339,0.00040342135,0.06597521,0.012394079,0.7104362],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99625033,0.00033943242,0.00054854306,0.0005012275,0.0021686817,0.00019171384],"domain_scores_gemma":[0.97257125,0.0042757858,0.0009970521,0.0020710502,0.01950349,0.00058133254],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026120425,0.0016549227,0.0012201541,0.004294404,0.0026911676,0.0031345915,0.0024316069,0.0019654676,0.1010518],"category_scores_gemma":[0.03770755,0.00096352166,0.0011947157,0.004406062,0.0013068527,0.0022011045,0.0023623607,0.0061076386,0.07324398],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014290672,0.0000026258924,0.000050652754,0.000071062204,0.0000044336944,0.000041334595,0.000018671124,0.000036804657,0.00005541878,0.00042031537,0.9940975,0.0051868595],"study_design_scores_gemma":[0.000011693129,0.0000059843755,0.00038467068,0.00015299306,0.000013960571,0.0000904598,0.000043187803,0.000104698694,0.00031735015,0.0006162143,0.99824405,0.000014690785],"about_ca_topic_score_codex":0.03337427,"about_ca_topic_score_gemma":0.044424158,"teacher_disagreement_score":0.96662575,"about_ca_system_score_codex":0.0028391234,"about_ca_system_score_gemma":0.0071436716,"threshold_uncertainty_score":0.33805197},"labels":[],"label_agreement":null},{"id":"W2909942535","doi":"10.1038/s41597-024-04278-w","title":"The International Bathymetric Chart of the Arctic Ocean Version 5.0","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Methane Hydrates and Related Phenomena","field":"Environmental Science","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Hydrographic Service; Geological Survey of Canada","funders":"Executive Agency for Small and Medium-sized Enterprises; Jet Propulsion Laboratory; European Climate, Infrastructure and Environment Executive Agency; Stockholms Universitet; Universitat de Barcelona; Vetenskapsrådet; Polarforskningssekretariatet; Agencia Estatal de Investigación; California Institute of Technology; National Aeronautics and Space Administration","keywords":"Bathymetry; Arctic; Bathymetric chart; Oceanography; Sea ice; Seafloor spreading; Submarine pipeline; Arctic ice pack; The arctic; Marine geology; Seabed; Geology; Environmental science","score_opus":0.020803981717288794,"score_gpt":0.24122598158175676,"score_spread":0.22042199986446798,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2909942535","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028383879,0.0021226853,0.11376814,0.0008176017,0.0020596103,0.0012173483,0.6711566,0.008798181,0.17167604],"genre_scores_gemma":[0.06146396,0.0020250187,0.1665681,0.0004016726,0.00021554825,0.0011431071,0.7403158,0.0028293212,0.025037488],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99896944,0.00016027567,0.0001477454,0.00011778415,0.0004858631,0.00011883027],"domain_scores_gemma":[0.9977943,0.000117975,0.00024042698,0.0002611122,0.0014562947,0.00013000067],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015432723,0.0008653327,0.00045427735,0.003332172,0.00052177475,0.0012791333,0.00067020295,0.00038857345,0.015213618],"category_scores_gemma":[0.0037220109,0.00032496426,0.0005482851,0.00621757,0.00034999807,0.00095180987,0.0010906637,0.0010919519,0.0077931467],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003432695,0.000073434014,0.04272372,0.0009005792,0.000109992834,0.00024246032,0.00081704743,0.0077114273,0.0097202705,0.013813969,0.6715308,0.25201315],"study_design_scores_gemma":[0.000035794874,0.000037904396,0.048696443,0.000184957,0.0000280951,0.00011788153,0.0002800935,0.0020347678,0.0013382522,0.0017320401,0.9454505,0.00006326675],"about_ca_topic_score_codex":0.09081776,"about_ca_topic_score_gemma":0.086691454,"teacher_disagreement_score":0.09081776,"about_ca_system_score_codex":0.00078278536,"about_ca_system_score_gemma":0.004067483,"threshold_uncertainty_score":0.18057829},"labels":[],"label_agreement":null},{"id":"W2910750340","doi":"10.1038/sdata.2018.280","title":"A global dataset of CO2 emissions and ancillary data related to emissions for 343 cities","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Air Quality and Health Impacts","field":"Environmental Science","cited_by":104,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"Centre National de la Recherche Scientifique; Fondation BNP Paribas; BNP Paribas Cardif","keywords":"Greenhouse gas; Environmental science; China; Sample (material); Data set; Data quality; Geography; Business; Statistics; Mathematics; Ecology","score_opus":0.11633315598266655,"score_gpt":0.38402478668171347,"score_spread":0.2676916306990469,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2910750340","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02932932,0.00009935449,0.00087392895,0.00010699512,0.0000180572,0.00006718552,0.9683792,0.00025297827,0.00087305973],"genre_scores_gemma":[0.025930174,0.00006643998,0.0016523487,0.000038580703,0.000008832663,0.0002092391,0.97171605,0.00003995139,0.00033835354],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986193,0.00023562151,0.00016376344,0.0004055191,0.00036371232,0.0002120894],"domain_scores_gemma":[0.9973435,0.000528269,0.00042619248,0.00055658387,0.0008991777,0.00024621192],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009467775,0.0011574692,0.0008751192,0.0037019008,0.00054787914,0.0011696406,0.0011940241,0.0012364502,0.0061942223],"category_scores_gemma":[0.0028716773,0.00048750153,0.0010922031,0.012694151,0.00037957338,0.0006750394,0.0016117156,0.0011444428,0.003591399],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009773468,0.00068918423,0.24921404,0.0026488751,0.000928501,0.00094421534,0.00042421868,0.0831659,0.0075711003,0.004270062,0.5927892,0.056377377],"study_design_scores_gemma":[0.00059349055,0.00015285639,0.5738997,0.00031826244,0.0002540797,0.00029039956,0.0007375333,0.023088831,0.0077251894,0.0019874733,0.39068407,0.00026815187],"about_ca_topic_score_codex":0.04170418,"about_ca_topic_score_gemma":0.029908108,"teacher_disagreement_score":0.04170418,"about_ca_system_score_codex":0.0013031308,"about_ca_system_score_gemma":0.0025266248,"threshold_uncertainty_score":0.082922876},"labels":[],"label_agreement":null},{"id":"W2911105719","doi":"10.1038/sdata.2018.299","title":"A long-term, temporally consistent, gridded daily meteorological dataset for northwestern North America","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cryospheric studies and observations","field":"Earth and Planetary Sciences","cited_by":104,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Environment and Climate Change Canada; Pacific Institute for Climate Solutions; University of Victoria","funders":"","keywords":"Term (time); Climatology; Environmental science; Geography; Geology","score_opus":0.07526351363347097,"score_gpt":0.2719372194882594,"score_spread":0.1966737058547884,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2911105719","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.100057356,0.000090297566,0.00072834134,0.000176931,0.000035470126,0.000054755186,0.89589584,0.0005000344,0.0024609386],"genre_scores_gemma":[0.082020596,0.000084150015,0.0029350198,0.00006318192,0.000020269035,0.00016898089,0.9135305,0.00005498315,0.0011223604],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9998336,0.000019022156,0.000019547757,0.000056329467,0.00004984606,0.000021713175],"domain_scores_gemma":[0.9990441,0.00011786003,0.00016952225,0.00015248731,0.00040922768,0.00010681646],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00025605134,0.0003769993,0.0002721629,0.0011129995,0.00044498188,0.00041347608,0.0005742479,0.00041132804,0.003511837],"category_scores_gemma":[0.0008496051,0.0001648388,0.00022937173,0.0029581124,0.00016988294,0.000331852,0.00038413715,0.00045850605,0.0017385299],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019669779,0.0006242318,0.3617974,0.0004868851,0.0002894798,0.00051123905,0.0006682404,0.0185426,0.0049831746,0.0009673259,0.5731209,0.037811834],"study_design_scores_gemma":[0.00012284782,0.00003903089,0.8136864,0.000072620955,0.000051080657,0.00013979658,0.00076331664,0.017152246,0.0017990202,0.0006010286,0.16552284,0.0000497984],"about_ca_topic_score_codex":0.17753813,"about_ca_topic_score_gemma":0.40806863,"teacher_disagreement_score":0.82246184,"about_ca_system_score_codex":0.00092388096,"about_ca_system_score_gemma":0.0017604831,"threshold_uncertainty_score":0.35300946},"labels":[],"label_agreement":null},{"id":"W2911747435","doi":"10.1038/s41597-018-0006-1","title":"Publisher Correction: GloPL, a global data base on pollen limitation of plant reproduction","year":2019,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Invertebrate Taxonomy and Ecology","field":"Agricultural and Biological Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Pollen; Reproduction; Plant reproduction; Base (topology); Computer science; Information retrieval; Statistics; Biology; Mathematics; Botany; Ecology; Pollination","score_opus":0.10894450563427192,"score_gpt":0.2556106430272448,"score_spread":0.14666613739297285,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2911747435","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012378026,0.0020602113,0.0059418487,0.0327826,0.7499992,0.00009959239,0.15625799,0.0035732768,0.048047446],"genre_scores_gemma":[0.01737403,0.007659715,0.016733855,0.017024186,0.04146403,0.00032615825,0.26913825,0.013245053,0.6170347],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99636084,0.00032290677,0.0006921953,0.00046705987,0.0019666075,0.00019044778],"domain_scores_gemma":[0.96896356,0.0057454533,0.0012148315,0.0030294852,0.020282667,0.0007640707],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003364313,0.0018710276,0.0014405155,0.008841529,0.0027965726,0.005255188,0.0023331705,0.0017695718,0.14657249],"category_scores_gemma":[0.050353426,0.00087519083,0.0010946702,0.009669792,0.0011944801,0.003370381,0.0032979301,0.0055752387,0.102629796],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014800706,0.0000027026856,0.00014026873,0.00008606061,0.0000052441114,0.00004727593,0.000032883716,0.000053531425,0.00004286424,0.0008928795,0.99129266,0.0073888586],"study_design_scores_gemma":[0.00000989336,0.0000047894846,0.0007560766,0.00018483434,0.000010618056,0.00008182746,0.00006132435,0.00006501513,0.0001368504,0.0006930227,0.997979,0.000016791177],"about_ca_topic_score_codex":0.049341045,"about_ca_topic_score_gemma":0.053261247,"teacher_disagreement_score":0.14657249,"about_ca_system_score_codex":0.003062683,"about_ca_system_score_gemma":0.0066671725,"threshold_uncertainty_score":0.4903338},"labels":[],"label_agreement":null},{"id":"W2912425033","doi":"10.1038/sdata.2018.310","title":"PEPCONF, a diverse data set of peptide conformational energies","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Chemical Synthesis and Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Okanagan University College; University of British Columbia, Okanagan Campus; University of British Columbia","funders":"British Columbia Knowledge Development Fund; Natural Sciences and Engineering Research Council of Canada; Western Canada Research Grid; Compute Canada","keywords":"Conformational isomerism; Peptide; Chemistry; Tripeptide; Set (abstract data type); Conformational ensembles; Data set; Computational chemistry; Crystallography; Database; Molecule; Computer science; Molecular dynamics; Artificial intelligence; Biochemistry","score_opus":0.05172673562870435,"score_gpt":0.29104771952590347,"score_spread":0.23932098389719914,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2912425033","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15596317,0.0053548855,0.020794297,0.00023250305,0.00016908343,0.0002931894,0.7947132,0.009525253,0.012954382],"genre_scores_gemma":[0.10136217,0.0014302758,0.023553165,0.0000929636,0.00003682089,0.0004997448,0.87128174,0.0009633288,0.0007798979],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99899393,0.00016295943,0.00011883409,0.00022470028,0.0004250692,0.00007447796],"domain_scores_gemma":[0.99881697,0.00038112342,0.0001505438,0.00028621726,0.00026669147,0.00009852183],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012819195,0.0023310215,0.0015166808,0.0032391392,0.00075679476,0.0012602133,0.0027296036,0.0014193625,0.009029722],"category_scores_gemma":[0.0036219521,0.0005803718,0.0008326823,0.005568103,0.00037927643,0.0016433039,0.0011134932,0.00170753,0.005075525],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004666341,0.0013258655,0.025641466,0.0130786495,0.0013951557,0.0020996213,0.00028773918,0.25376147,0.08028684,0.026061416,0.42716855,0.16422686],"study_design_scores_gemma":[0.001994251,0.0018899402,0.047124017,0.00054881856,0.0004846294,0.003645051,0.00029485606,0.34295914,0.101827204,0.028370913,0.4702418,0.00061945117],"about_ca_topic_score_codex":0.0015350787,"about_ca_topic_score_gemma":0.0014178265,"teacher_disagreement_score":0.009029722,"about_ca_system_score_codex":0.0005151444,"about_ca_system_score_gemma":0.00088369614,"threshold_uncertainty_score":0.030207396},"labels":[],"label_agreement":null},{"id":"W2913467375","doi":"10.1038/sdata.2019.13","title":"MicroRNA expression profiles across blood and different tissues in cattle","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"MicroRNA in disease regulation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Alberta Agriculture and Forestry","keywords":"microRNA; Biology; Microvesicles; Gene expression profiling; Computational biology; Gene expression; Transcriptome; Regulation of gene expression; Gene; Bioinformatics; Genetics","score_opus":0.01752199827010926,"score_gpt":0.2868957112692079,"score_spread":0.2693737129990986,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2913467375","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9883095,0.002313083,0.0029180301,0.000055989047,0.00002256121,0.000015237547,0.00526907,0.00006764913,0.0010289153],"genre_scores_gemma":[0.97644186,0.0017668878,0.009083631,0.00023044155,0.000035248177,0.000097198674,0.009789527,0.00009233415,0.0024628781],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99938846,0.00009612508,0.000047512356,0.00025608824,0.00012382306,0.00008797058],"domain_scores_gemma":[0.99966323,0.00012465674,0.00007021632,0.000022744089,0.000086198284,0.000032974756],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00056243833,0.00021568479,0.00042819313,0.001341015,0.00037678442,0.00097071513,0.00011510415,0.00039037678,0.000853889],"category_scores_gemma":[0.00080016284,0.00022522062,0.00028274834,0.0011353526,0.00025270638,0.0002255406,0.00028434684,0.0003494361,0.00037690223],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003070887,0.00008247621,0.22109696,0.00036248888,0.0002631459,0.00044340972,0.0015283943,0.00043852755,0.7345024,0.000293927,0.0008556387,0.037061762],"study_design_scores_gemma":[0.000039189443,0.00071449776,0.8660713,0.000087553664,0.0004269164,0.002124735,0.0012120254,0.0025127237,0.11298035,0.00048527613,0.0132792825,0.00006610921],"about_ca_topic_score_codex":0.0018594239,"about_ca_topic_score_gemma":0.00297403,"teacher_disagreement_score":0.0018594239,"about_ca_system_score_codex":0.00028856547,"about_ca_system_score_gemma":0.00017135851,"threshold_uncertainty_score":0.0036972165},"labels":[],"label_agreement":null},{"id":"W2914831628","doi":"10.1038/sdata.2018.301","title":"A collection of rumen bacteriome data from 334 mid-lactation dairy cows","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ruminant Nutrition and Digestive Physiology","field":"Agricultural and Biological Sciences","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Natural Science Foundation of China","keywords":"Biology; Rumen; Milking; Breed; Population; Dairy cattle; 16S ribosomal RNA; Lactation; Pyrosequencing; Animal science; Biotechnology; Food science; Bacteria; Genetics; Gene; Fermentation","score_opus":0.07573706693013013,"score_gpt":0.2746505349462183,"score_spread":0.1989134680160882,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2914831628","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8705134,0.003158779,0.007743391,0.00014711397,0.00006670155,0.00013658313,0.11602819,0.00028801645,0.001917966],"genre_scores_gemma":[0.5214997,0.0021287391,0.04731514,0.00047255465,0.00010714528,0.00058759423,0.42498735,0.00021572907,0.0026860582],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992335,0.00008556404,0.00007468753,0.00022164646,0.00025395176,0.00013067406],"domain_scores_gemma":[0.99939454,0.00010212499,0.00010998518,0.00006468334,0.00023923337,0.00008940692],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00082389935,0.0005856524,0.0008186529,0.0016601512,0.0010680885,0.00072527945,0.00032640062,0.00067890796,0.0019892978],"category_scores_gemma":[0.0009372629,0.00026030975,0.00064213015,0.0020865023,0.00023658585,0.0003392159,0.00079633103,0.00053739676,0.0011053375],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027628709,0.00035008776,0.21129589,0.0016951103,0.00071367586,0.00085835217,0.0014031457,0.0012438837,0.6752404,0.00023905172,0.007060131,0.097137354],"study_design_scores_gemma":[0.00007070878,0.00057262834,0.8876384,0.00018039717,0.00047133316,0.0011014944,0.0010033331,0.0022070266,0.048877187,0.00032386256,0.05745338,0.000100263],"about_ca_topic_score_codex":0.0032979571,"about_ca_topic_score_gemma":0.0074408785,"teacher_disagreement_score":0.0032979571,"about_ca_system_score_codex":0.00033229747,"about_ca_system_score_gemma":0.00069562317,"threshold_uncertainty_score":0.0066548586},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2915100187","doi":"10.1038/sdata.2019.32","title":"A 1980–2018 global fire danger re-analysis dataset for the Canadian Fire Weather Indices","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":88,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Horizon 2020 Framework Programme","keywords":"Environmental science; Climatology; Meteorology; Geography; Geology","score_opus":0.024263497241997518,"score_gpt":0.2643931061104489,"score_spread":0.2401296088684514,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2915100187","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021129877,0.00007768019,0.00020541128,0.000062029096,0.000035637717,0.000034055858,0.9956735,0.00029177268,0.0015068819],"genre_scores_gemma":[0.003005034,0.000050956845,0.00052452495,0.000018043122,0.0000050403996,0.00004398821,0.9954762,0.000033992783,0.00084216066],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993255,0.000026237964,0.000042523647,0.0001229259,0.0003106064,0.00017223916],"domain_scores_gemma":[0.9970893,0.000114411276,0.00014523343,0.0002718109,0.002153117,0.0002262142],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005812307,0.0013289049,0.00081770483,0.0044658226,0.0014816858,0.00093780464,0.0020441164,0.00079171825,0.011149248],"category_scores_gemma":[0.0026034925,0.0003704251,0.0008816557,0.0067586508,0.00040989637,0.0006066857,0.0009253799,0.001386116,0.009142521],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010989386,0.00006391909,0.010460625,0.00034066668,0.00008673544,0.00009167853,0.00008983365,0.0027400723,0.000689043,0.0010824616,0.9707891,0.013456088],"study_design_scores_gemma":[0.0001357738,0.000023913557,0.10341225,0.00026118173,0.00007408078,0.00008852908,0.0003136779,0.0059556984,0.0017282967,0.00071188493,0.88716394,0.00013081967],"about_ca_topic_score_codex":0.9516663,"about_ca_topic_score_gemma":0.9754878,"teacher_disagreement_score":0.048333704,"about_ca_system_score_codex":0.0082530035,"about_ca_system_score_gemma":0.016544275,"threshold_uncertainty_score":0.09723663},"labels":[],"label_agreement":null},{"id":"W2915913218","doi":"10.1038/sdata.2019.25","title":"Map of physical interactions between extracellular domains of Arabidopsis leucine-rich repeat receptor kinases","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant-Microbe Interactions and Immunity","field":"Agricultural and Biological Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Austrian Science Fund; National Science Foundation","keywords":"Arabidopsis; Leucine-rich repeat; Kinase; Signalling; Biology; Receptor; Computational biology; Extracellular; Cell biology; Gene; Genetics","score_opus":0.04508187148470579,"score_gpt":0.26960323848637335,"score_spread":0.22452136700166755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2915913218","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97451496,0.0009955352,0.006249475,0.00007652031,0.000013110973,0.000043119955,0.014796253,0.0002570714,0.0030540114],"genre_scores_gemma":[0.96078455,0.00027510704,0.01116077,0.000057170826,0.000008041411,0.00006486278,0.026706597,0.00004395428,0.0008990203],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99974924,0.00002914134,0.000022392032,0.00008207736,0.00008091205,0.000036196216],"domain_scores_gemma":[0.99921477,0.00027809985,0.00024191201,0.000055752043,0.000074325035,0.00013510801],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00021849295,0.00047550793,0.00039208235,0.0013254024,0.00048519284,0.0004417146,0.00024150482,0.00029451793,0.0029260947],"category_scores_gemma":[0.0007395888,0.00018885417,0.0005715954,0.0010051577,0.00014338642,0.00026956855,0.00037317842,0.00040853568,0.0012158222],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011994861,0.00027466947,0.093442716,0.0006513981,0.00033418345,0.0012265196,0.00012631198,0.0055456003,0.87077063,0.0010259063,0.0021911187,0.023211518],"study_design_scores_gemma":[0.00012254225,0.00062604184,0.7814886,0.000057686466,0.00030588632,0.00272581,0.00039560854,0.043850787,0.15462497,0.0017223099,0.013993123,0.000086601605],"about_ca_topic_score_codex":0.0014550741,"about_ca_topic_score_gemma":0.0040079723,"teacher_disagreement_score":0.0029260947,"about_ca_system_score_codex":0.0003275387,"about_ca_system_score_gemma":0.00043452258,"threshold_uncertainty_score":0.009788692},"labels":[],"label_agreement":null},{"id":"W2929357616","doi":"10.1038/s41597-019-0038-1","title":"Statistically downscaled climate dataset for East Africa","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":111,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Downscaling; Climate change; Climatology; Precipitation; Climate model; Environmental science; Scale (ratio); Representative Concentration Pathways; Projection (relational algebra); Tanzania; Geography; Environmental resource management; Physical geography; Meteorology; Cartography; Environmental planning; Computer science","score_opus":0.06660062009485934,"score_gpt":0.2906167723772865,"score_spread":0.22401615228242716,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2929357616","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011248008,0.00012560085,0.00070001907,0.00008595511,0.000037533893,0.00005101348,0.9861968,0.00032272894,0.0012323397],"genre_scores_gemma":[0.0167412,0.00012931901,0.0023781497,0.00002961052,0.000019054683,0.00024608002,0.97964954,0.000074642034,0.000732384],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997913,0.00003091825,0.00003247063,0.00006985487,0.000038930324,0.00003651241],"domain_scores_gemma":[0.9994407,0.00005680657,0.000098516626,0.00014651661,0.00021209552,0.00004528617],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051016256,0.00079420995,0.00050088915,0.0013185436,0.00034989693,0.00043681092,0.0007200138,0.00036173713,0.008082132],"category_scores_gemma":[0.001143802,0.00027098792,0.00047603264,0.002786523,0.00011372923,0.0004301157,0.00050268875,0.0006705914,0.0053465213],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00073626864,0.00023392688,0.061753128,0.0018109349,0.0005178002,0.00081212376,0.00041684342,0.024080707,0.005321969,0.0030976674,0.8513742,0.04984439],"study_design_scores_gemma":[0.0006071944,0.00005839999,0.19971669,0.00028325125,0.00013018136,0.00024456214,0.00028835307,0.010623408,0.0048688897,0.0011440692,0.78195554,0.00007945853],"about_ca_topic_score_codex":0.033019904,"about_ca_topic_score_gemma":0.03251399,"teacher_disagreement_score":0.033019904,"about_ca_system_score_codex":0.00048712365,"about_ca_system_score_gemma":0.0012135763,"threshold_uncertainty_score":0.06565541},"labels":[],"label_agreement":null},{"id":"W2930808060","doi":"10.1038/s41597-019-0024-7","title":"Geo-referenced population-specific microsatellite data across American continents, the MacroPopGen Database","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genetic diversity and population structure","field":"Biochemistry, Genetics and Molecular Biology","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Winnipeg; Golder Associates (Canada); BGC Engineering (Canada); Concordia University","funders":"Concordia University; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Biology; Fish migration; Microsatellite; Biodiversity; Genetic diversity; Population; Ecology; Intraspecific competition; Zoology; Allele; Demography; Genetics","score_opus":0.05393613729136078,"score_gpt":0.3173182112775451,"score_spread":0.26338207398618435,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2930808060","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11784988,0.0006104119,0.007451069,0.00018137561,0.000024715011,0.00013313441,0.86653227,0.0010325492,0.0061845984],"genre_scores_gemma":[0.113301866,0.00046986176,0.01757813,0.000103128106,0.00002389164,0.0006593907,0.8664314,0.00026551585,0.0011668],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99925786,0.00016102758,0.00014592761,0.00022922806,0.00015532684,0.000050625717],"domain_scores_gemma":[0.9959417,0.0008759247,0.0011836357,0.0007668749,0.0007331286,0.00049869454],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020321317,0.00033996976,0.00045146726,0.005949183,0.0003707579,0.0012808901,0.0007889381,0.00023543397,0.008403929],"category_scores_gemma":[0.0053659603,0.00035589127,0.00017931208,0.01129196,0.00020293354,0.00073379383,0.0012163862,0.00045688698,0.002557059],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015751624,0.00015935392,0.5897497,0.002037471,0.0009829666,0.00051045656,0.0011343961,0.0029905846,0.006186806,0.0054770657,0.22881562,0.16038047],"study_design_scores_gemma":[0.0002317623,0.00007906641,0.7306705,0.0004913357,0.0003633633,0.0005644775,0.00087570184,0.0048026894,0.0024405075,0.003083048,0.2563467,0.000050856688],"about_ca_topic_score_codex":0.009253222,"about_ca_topic_score_gemma":0.012300322,"teacher_disagreement_score":0.009253222,"about_ca_system_score_codex":0.0002876848,"about_ca_system_score_gemma":0.0013565006,"threshold_uncertainty_score":0.028113961},"labels":[],"label_agreement":null},{"id":"W2936362429","doi":"10.1038/s41597-019-0016-7","title":"Physical activity, sleep and cardiovascular health data for 50,000 individuals from the MyHeart Counts Study","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Physical Activity and Health","field":"Medicine","cited_by":82,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ted Rogers Centre for Heart Research","funders":"U.S. National Library of Medicine","keywords":"Cardiovascular health; Physical activity; Scale (ratio); Gerontology; Data collection; Physical fitness; Sleep (system call); Medicine; Cardiovascular fitness; Health data; Psychology; Physical therapy; Health care; Disease; Computer science","score_opus":0.16373391544128407,"score_gpt":0.3929367127110389,"score_spread":0.22920279726975484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2936362429","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.36912066,0.0007556928,0.0016420039,0.00076876325,0.00013347532,0.00063595676,0.6194194,0.00015455658,0.00736951],"genre_scores_gemma":[0.3694569,0.0009245547,0.0038632257,0.0009370013,0.00029120082,0.0050668935,0.61214924,0.00011952436,0.007191464],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9986766,0.0003366344,0.0003373276,0.00026462504,0.00026673818,0.00011808068],"domain_scores_gemma":[0.99642134,0.00062731194,0.00095820636,0.0008213512,0.0007792875,0.0003925142],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016579201,0.00046547232,0.0006041414,0.0017192569,0.0006842584,0.0008851253,0.0007411354,0.00080529595,0.0037522486],"category_scores_gemma":[0.0071165822,0.00054248946,0.00063485233,0.002908192,0.00021509107,0.000605834,0.0014412167,0.0010170761,0.0023912862],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016402302,0.0003777598,0.82933784,0.00077454746,0.00068547396,0.00023347068,0.00146088,0.0004722903,0.0010370385,0.0005680841,0.14177911,0.021633163],"study_design_scores_gemma":[0.0002340094,0.00018045187,0.97287107,0.00014127433,0.00018924833,0.0001637878,0.0005190765,0.00019354085,0.00019313226,0.00016815825,0.025109455,0.000036762343],"about_ca_topic_score_codex":0.017956553,"about_ca_topic_score_gemma":0.026869925,"teacher_disagreement_score":0.017956553,"about_ca_system_score_codex":0.0003088076,"about_ca_system_score_gemma":0.00070904876,"threshold_uncertainty_score":0.035704076},"labels":[],"label_agreement":null},{"id":"W2937999720","doi":"10.1038/s41597-019-0034-5","title":"A statistical atlas of cerebral arteries generated using multi-center MRA datasets from healthy subjects","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cerebrovascular and Carotid Artery Diseases","field":"Medicine","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Brain Institute; University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Government of Canada","keywords":"Magnetic resonance imaging; Neuroimaging; Magnetic resonance angiography; Cerebral arteries; Atlas (anatomy); Medicine; Radiology; Computer science; Anatomy","score_opus":0.06136679331602411,"score_gpt":0.3246087648774315,"score_spread":0.2632419715614074,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2937999720","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12349626,0.0016453988,0.49033356,0.0010577299,0.00035993144,0.0010858504,0.3429287,0.024782686,0.014309897],"genre_scores_gemma":[0.34966207,0.0019713254,0.3676834,0.00041159184,0.00033178055,0.004149906,0.26579937,0.004211715,0.0057788217],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9990096,0.00020959425,0.00016272717,0.00030595792,0.00023765616,0.0000744143],"domain_scores_gemma":[0.99674666,0.0012225631,0.0004463407,0.000736412,0.00073602266,0.0001120482],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020274336,0.00073000055,0.0008575047,0.0050643645,0.00072264497,0.0017674572,0.0012448077,0.0009726625,0.01142911],"category_scores_gemma":[0.006210335,0.00061612244,0.0010440721,0.0053419457,0.0005913974,0.0007668947,0.0012934807,0.0009266678,0.0047840755],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025752578,0.000618313,0.06320692,0.0035569447,0.0017898254,0.0036042428,0.0021798133,0.06414706,0.041176293,0.024517119,0.3227172,0.46991095],"study_design_scores_gemma":[0.0005521687,0.0007643294,0.282183,0.00067473063,0.0011021348,0.011938532,0.0009607849,0.13755675,0.029722946,0.053717297,0.48017564,0.0006516007],"about_ca_topic_score_codex":0.009127547,"about_ca_topic_score_gemma":0.017154353,"teacher_disagreement_score":0.01142911,"about_ca_system_score_codex":0.000832399,"about_ca_system_score_gemma":0.0030219147,"threshold_uncertainty_score":0.038234174},"labels":[],"label_agreement":null},{"id":"W2942802270","doi":"10.1038/s41597-019-0040-7","title":"A long-term dataset of lake surface water temperature over the Tibetan Plateau derived from AVHRR 1981–2015","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Urban Heat Island Mitigation","field":"Environmental Science","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Oceanic and Atmospheric Administration; National Natural Science Foundation of China; Canada Excellence Research Chairs, Government of Canada; University of Reading","keywords":"Plateau (mathematics); Term (time); Environmental science; Surface water; Remote sensing; Climatology; Physical geography; Geography; Geology; Mathematics; Physics","score_opus":0.01800691870223713,"score_gpt":0.244610230621266,"score_spread":0.22660331191902885,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2942802270","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8996736,0.00025313575,0.0013515548,0.000084903295,0.00003145382,0.000046717818,0.09575125,0.00026818868,0.0025392638],"genre_scores_gemma":[0.738733,0.00013338724,0.0025940149,0.000031779415,0.000047638714,0.00008626528,0.25697252,0.000037994843,0.0013633207],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99980754,0.000023392759,0.000021174108,0.000056379376,0.00005798439,0.000033480854],"domain_scores_gemma":[0.999498,0.000028126444,0.0001006947,0.000081083184,0.00021719233,0.000074895055],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046102982,0.00037654417,0.00022217707,0.0013851224,0.00031512132,0.00048538714,0.0003988867,0.00028589,0.0011765385],"category_scores_gemma":[0.00047405827,0.00010299187,0.0002781187,0.002679692,0.00021004486,0.00028978154,0.0003816694,0.00021582776,0.00043386157],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00057994254,0.0006590247,0.7833128,0.00044901864,0.00054681615,0.001119843,0.0005640443,0.063684694,0.018488346,0.00094110763,0.045843195,0.08381128],"study_design_scores_gemma":[0.00005045063,0.00004660806,0.96484137,0.000036861784,0.000036836936,0.00006162946,0.00023786312,0.018928817,0.0019740998,0.00013464616,0.013618586,0.000032245105],"about_ca_topic_score_codex":0.072248414,"about_ca_topic_score_gemma":0.09125871,"teacher_disagreement_score":0.072248414,"about_ca_system_score_codex":0.0006400006,"about_ca_system_score_gemma":0.0009383332,"threshold_uncertainty_score":0.14365572},"labels":[],"label_agreement":null},{"id":"W2944268392","doi":"10.1038/s41597-019-0069-7","title":"Worldwide measurements of bioturbation intensity, ventilation rate, and the mixing depth of marine sediments","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine Biology and Ecology Research","field":"Earth and Planetary Sciences","cited_by":94,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Memorial University of Newfoundland","funders":"Natural Environment Research Council; Research Councils UK; Sight Research UK","keywords":"Bioturbation; Biogeochemical cycle; Environmental science; Ecology; Ecosystem engineer; Ecosystem; Burrow; Oceanography; Sediment; Paleolimnology; Intertidal zone; Marine ecosystem; Climate change; Geology; Biology; Geomorphology","score_opus":0.0681339636167772,"score_gpt":0.2746749684645114,"score_spread":0.2065410048477342,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2944268392","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.39218047,0.005748506,0.00657289,0.0001788312,0.0000858282,0.00008564808,0.57420033,0.0003582931,0.020589132],"genre_scores_gemma":[0.3954987,0.0054372214,0.011998362,0.00014129194,0.00008593925,0.00039706766,0.5810335,0.00020399701,0.0052039637],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99887973,0.00013379876,0.00022190253,0.00039381112,0.00028385612,0.000086769585],"domain_scores_gemma":[0.9934228,0.00084369763,0.003091299,0.00093654596,0.0014670854,0.00023862443],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011044435,0.00068302633,0.00047457145,0.009943749,0.0003256944,0.0006729346,0.0005674782,0.00026814415,0.0036810231],"category_scores_gemma":[0.004146531,0.00026561925,0.0004997465,0.014262988,0.00039364086,0.0008132581,0.0013660424,0.00034395847,0.0021927822],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001992176,0.000060951883,0.7996343,0.002645499,0.000625022,0.0001687051,0.0013663177,0.0014233013,0.008286809,0.0012482046,0.034294087,0.15004764],"study_design_scores_gemma":[0.0000075267144,0.000026648,0.9308806,0.00028055813,0.00011636161,0.0001734446,0.0006265757,0.00020170397,0.0017276956,0.0005015195,0.06542934,0.000027971111],"about_ca_topic_score_codex":0.018420234,"about_ca_topic_score_gemma":0.036915142,"teacher_disagreement_score":0.018420234,"about_ca_system_score_codex":0.0004482594,"about_ca_system_score_gemma":0.0006535215,"threshold_uncertainty_score":0.03662598},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2950886245","doi":"10.1038/sdata.2018.142","title":"A multi-omic atlas of the human frontal cortex for aging and Alzheimer’s disease research","year":2018,"lang":"en","type":"article","venue":"Scientific Data","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":688,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of Environmental Health Sciences; National Institute on Aging; National Institutes of Health; National Center for Research Resources; Broad Institute","keywords":"Dementia; Biology; Dorsolateral prefrontal cortex; Histone; Cognitive decline; Neuroscience; Alzheimer's disease; Biomarker; Disease; Prefrontal cortex; Computational biology; Genetics; Bioinformatics; Cognition; Medicine; Gene; Internal medicine","score_opus":0.10791270952260082,"score_gpt":0.3747868602896041,"score_spread":0.26687415076700327,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950886245","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.044536106,0.011587452,0.11658417,0.0018809225,0.00045967518,0.00058030605,0.79383075,0.006800547,0.02374011],"genre_scores_gemma":[0.12678559,0.008269346,0.2449132,0.0006897221,0.0001889287,0.0019511685,0.6064293,0.0013581413,0.009414589],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9994361,0.00014387924,0.000073496696,0.0001607221,0.00012284072,0.00006300574],"domain_scores_gemma":[0.99915516,0.00012173749,0.0001523281,0.0002215359,0.0002617932,0.000087480425],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001412351,0.00069773983,0.00062731147,0.005843497,0.00087367196,0.0013843193,0.0009091365,0.0005215806,0.008967502],"category_scores_gemma":[0.0018052962,0.00026942417,0.0011382038,0.005992091,0.00026329982,0.00053993054,0.0017023998,0.00062708615,0.0038228596],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007605717,0.000110860834,0.06707669,0.0056338613,0.0024076328,0.0014131671,0.0020119275,0.0063406825,0.08939153,0.026783215,0.48551208,0.31255782],"study_design_scores_gemma":[0.00006999279,0.00008159787,0.2467743,0.00058930484,0.0006820752,0.0020139113,0.00040521257,0.0039625834,0.006143714,0.015766257,0.7234269,0.00008409402],"about_ca_topic_score_codex":0.018073604,"about_ca_topic_score_gemma":0.041381482,"teacher_disagreement_score":0.018073604,"about_ca_system_score_codex":0.001232318,"about_ca_system_score_gemma":0.0037496646,"threshold_uncertainty_score":0.035936832},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2951103577","doi":"10.1038/sdata.2016.44","title":"The brain imaging data structure, a format for organizing and describing outputs of neuroimaging experiments","year":2016,"lang":"en","type":"article","venue":"Scientific Data","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":1910,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Neurological Institute and Hospital","funders":"National Institute on Alcohol Abuse and Alcoholism; European Regional Development Fund; National Institute of General Medical Sciences; National Institute of Mental Health; Medical Research Council; Center for Behavioral Brain Sciences; National Institute of Biomedical Imaging and Bioengineering; Wellcome Trust; National Institutes of Health; Laura and John Arnold Foundation; National Science Foundation","keywords":"Computer science; Neuroimaging; Metadata; Data sharing; Software; Field (mathematics); File format; Functional magnetic resonance imaging; Information retrieval; Artificial intelligence; Data mining; Data science; Database; World Wide Web; Programming language","score_opus":0.37461250903114907,"score_gpt":0.4171929214087385,"score_spread":0.042580412377589416,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2951103577","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042819544,0.0004939849,0.7014865,0.0021872702,0.0006131046,0.0018675742,0.22331715,0.054504786,0.011247694],"genre_scores_gemma":[0.032148663,0.0014146166,0.584708,0.0015279085,0.00043306197,0.0051686284,0.35449597,0.013526125,0.0065769637],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9910883,0.002009292,0.0032120182,0.0011356185,0.002194253,0.00036045286],"domain_scores_gemma":[0.93681264,0.021077558,0.005902303,0.022081668,0.012408974,0.0017168501],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.015346307,0.0019505816,0.0012099324,0.0095780315,0.0019107978,0.0072188964,0.0037190572,0.0021811104,0.0113897445],"category_scores_gemma":[0.057514187,0.001275127,0.001595287,0.008420941,0.0018684542,0.009435492,0.0054134503,0.0038146677,0.015626807],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012251998,0.00032015456,0.009918919,0.002311142,0.00021076565,0.00085900497,0.0020004176,0.008280424,0.015135695,0.15335716,0.5800146,0.22636658],"study_design_scores_gemma":[0.00026469526,0.00018839465,0.0061927657,0.0006590467,0.00008190318,0.00095887075,0.00061331154,0.013240179,0.02122471,0.08696954,0.86932933,0.00027728424],"about_ca_topic_score_codex":0.0071042543,"about_ca_topic_score_gemma":0.0054578255,"teacher_disagreement_score":0.99628097,"about_ca_system_score_codex":0.0025663343,"about_ca_system_score_gemma":0.008423593,"threshold_uncertainty_score":0.08116001},"labels":[],"label_agreement":null},{"id":"W2952906091","doi":"10.1038/sdata.2014.54","title":"A high resolution 7-Tesla resting-state fMRI test-retest dataset with cognitive and physiological measures","year":2015,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Max-Planck-Gesellschaft","keywords":"Resting state fMRI; Functional magnetic resonance imaging; Cognition; Prefrontal cortex; Session (web analytics); Psychology; Audiology; Computer science; Neuroscience; Cognitive psychology; Medicine","score_opus":0.2342273660817456,"score_gpt":0.3202344923102148,"score_spread":0.08600712622846918,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952906091","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.112211995,0.0009615976,0.023646092,0.00051765476,0.0002616915,0.0015755654,0.850136,0.0031873467,0.0075020473],"genre_scores_gemma":[0.065999895,0.00023333945,0.021623688,0.00024715747,0.000119551056,0.004070058,0.90247744,0.000737311,0.004491561],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9993026,0.00013133403,0.000101634774,0.00023428994,0.00016871377,0.0000614631],"domain_scores_gemma":[0.9966287,0.0006870827,0.00024255067,0.0010396959,0.0012476661,0.00015422165],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014945918,0.0009778659,0.0009730625,0.0013305373,0.00078949233,0.00065962755,0.0015768717,0.0011104738,0.0090843635],"category_scores_gemma":[0.0049480493,0.00040876804,0.0007986611,0.0014483643,0.0004065669,0.00046935354,0.0009896006,0.0007236297,0.00865532],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0035027084,0.0015441636,0.06501072,0.0038941388,0.0015843773,0.0016944121,0.001233452,0.005075644,0.103824355,0.0015550328,0.63380605,0.17727491],"study_design_scores_gemma":[0.00077794143,0.00096326973,0.597914,0.00031189708,0.00078494113,0.0045138225,0.00037974108,0.0038209818,0.0184764,0.004679493,0.3670163,0.00036118188],"about_ca_topic_score_codex":0.0071325023,"about_ca_topic_score_gemma":0.022200065,"teacher_disagreement_score":0.0090843635,"about_ca_system_score_codex":0.00044150706,"about_ca_system_score_gemma":0.0010277863,"threshold_uncertainty_score":0.030390263},"labels":[],"label_agreement":null},{"id":"W2953143431","doi":"10.1038/sdata.2017.147","title":"Linking FANTOM5 CAGE peaks to annotations with CAGEscan","year":2017,"lang":"en","type":"article","venue":"Scientific Data","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"RIKEN; Australian Research Council; Ministry of Education, Culture, Sports, Science and Technology; Conquer Cancer Foundation","keywords":"Cage; Computer science; Biology; Combinatorics; Mathematics","score_opus":0.046579806632690866,"score_gpt":0.3092296791531692,"score_spread":0.26264987252047833,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2953143431","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.035108272,0.0035011605,0.34370404,0.00085262134,0.0011020327,0.000624094,0.48744127,0.09779222,0.029874219],"genre_scores_gemma":[0.045363557,0.0012787754,0.2773239,0.0007655436,0.00013884688,0.0013269183,0.65000755,0.016263999,0.0075308545],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9970174,0.00036131046,0.0002318276,0.001275607,0.00090135855,0.00021242433],"domain_scores_gemma":[0.99631655,0.0013585064,0.0003805132,0.0007252376,0.0009815039,0.00023759871],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004070217,0.0038702586,0.0016922898,0.006932654,0.0018800372,0.0035001659,0.0033264125,0.0017518571,0.019662043],"category_scores_gemma":[0.012105208,0.0015658141,0.0027659289,0.0047000963,0.00072852545,0.0018359954,0.0037134944,0.002085863,0.0121057285],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004447121,0.00021098526,0.019648202,0.015468087,0.0028340018,0.003068388,0.004182121,0.0126447715,0.34653437,0.031014105,0.36092836,0.19901949],"study_design_scores_gemma":[0.0002652787,0.00027490943,0.028662398,0.0011017497,0.0012843042,0.0019665665,0.0010042257,0.03988723,0.19867013,0.026494559,0.69967633,0.0007123094],"about_ca_topic_score_codex":0.008712566,"about_ca_topic_score_gemma":0.017036716,"teacher_disagreement_score":0.019662043,"about_ca_system_score_codex":0.0021652828,"about_ca_system_score_gemma":0.0028659974,"threshold_uncertainty_score":0.06577611},"labels":[],"label_agreement":null},{"id":"W2954654581","doi":"10.1038/s41597-019-0105-7","title":"iEEG-BIDS, extending the Brain Imaging Data Structure specification to human intracranial electrophysiology","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":181,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek; National Institute of Neurological Disorders and Stroke; National Institute of Mental Health; Whitehall Foundation; Gordon and Betty Moore Foundation; Horizon 2020 Framework Programme; Alfred P. Sloan Foundation; European Commission; Alan Turing Institute; National Institutes of Health; National Science Foundation","keywords":"Computer science; Metadata; Data sharing; Neuroinformatics; Human brain; Neuroscience; Artificial intelligence; Data science; Psychology; World Wide Web","score_opus":0.07157132679559527,"score_gpt":0.3255467582382028,"score_spread":0.2539754314426075,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2954654581","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0069301156,0.00028322675,0.93865126,0.001228055,0.00044625782,0.0007673188,0.01563596,0.02696412,0.009093564],"genre_scores_gemma":[0.12458152,0.0011206128,0.7344847,0.003823648,0.00056334405,0.0039530224,0.08838096,0.024990197,0.01810206],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9900131,0.0020946355,0.0034286478,0.0007071903,0.003128985,0.00062731345],"domain_scores_gemma":[0.97416466,0.006947621,0.0016966988,0.008507229,0.007761708,0.00092210056],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.014233019,0.0014574121,0.00087546394,0.002549349,0.0015869841,0.0053728865,0.003102384,0.002341921,0.00561357],"category_scores_gemma":[0.031702828,0.0014344908,0.0018904284,0.0024686123,0.0020292336,0.00666793,0.0063226665,0.0038153927,0.0050850497],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013590595,0.00031205855,0.012582747,0.0013821854,0.00022389191,0.001852928,0.0032638712,0.022421058,0.023897553,0.5619343,0.23297054,0.13779981],"study_design_scores_gemma":[0.00020760251,0.00018139009,0.002124186,0.0005019087,0.0000871748,0.0012494026,0.00054032623,0.040907297,0.027457811,0.11957665,0.8069706,0.0001956221],"about_ca_topic_score_codex":0.008814824,"about_ca_topic_score_gemma":0.011940754,"teacher_disagreement_score":0.985767,"about_ca_system_score_codex":0.0018909134,"about_ca_system_score_gemma":0.005962422,"threshold_uncertainty_score":0.07527232},"labels":[],"label_agreement":null},{"id":"W2955660171","doi":"10.1038/s41597-019-0122-6","title":"A high-resolution gridded dataset to assess electrification in sub-Saharan Africa","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Impact of Light on Environment and Health","field":"Environmental Science","cited_by":121,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"Ministero dell’Istruzione, dell’Università e della Ricerca; International Institute for Applied Systems Analysis; Università degli Studi di Milano","keywords":"Electrification; Electricity; Rural electrification; Consistency (knowledge bases); Benchmarking; Data quality; Sustainability; Environmental economics; Consumption (sociology); Quality (philosophy); Computer science; Database; Geography; Business; Service (business); Economics","score_opus":0.06913132960950415,"score_gpt":0.2948649510468915,"score_spread":0.22573362143738734,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2955660171","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19006759,0.0005000709,0.002646079,0.0004384994,0.000050618786,0.0001211115,0.8016565,0.00046579386,0.004053684],"genre_scores_gemma":[0.2603274,0.000433338,0.008801869,0.000092359245,0.00003553808,0.0004248537,0.72850263,0.00008990617,0.0012921453],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99967873,0.00010184121,0.000046141224,0.000060184033,0.00006176657,0.000051424602],"domain_scores_gemma":[0.9990965,0.00024297184,0.00018823457,0.00015303129,0.00024614748,0.0000731686],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005298066,0.000449111,0.00036633172,0.002217017,0.00026579303,0.0004952845,0.0005978408,0.00051167083,0.0030949796],"category_scores_gemma":[0.0025588607,0.00019338162,0.0003608863,0.005457266,0.00020249169,0.00044589626,0.00065809384,0.0004019877,0.0010629128],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00067104306,0.0005241542,0.4285258,0.0030422225,0.00081053626,0.002029886,0.00226513,0.090214975,0.0055992194,0.008457419,0.3465824,0.11127722],"study_design_scores_gemma":[0.00031581987,0.000098089185,0.5752937,0.0005422763,0.00011373913,0.00048703136,0.0030442043,0.033419713,0.003878724,0.003620827,0.37908852,0.00009733727],"about_ca_topic_score_codex":0.032103434,"about_ca_topic_score_gemma":0.027249675,"teacher_disagreement_score":0.032103434,"about_ca_system_score_codex":0.0005226541,"about_ca_system_score_gemma":0.0007078734,"threshold_uncertainty_score":0.06383312},"labels":[],"label_agreement":null},{"id":"W2957869398","doi":"10.1038/s41597-019-0129-z","title":"A macaque connectome for large-scale network simulations in TheVirtualBrain","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":91,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Western University; McGill University; Montreal Neurological Institute and Hospital; Baycrest Hospital","funders":"Deutsche Forschungsgemeinschaft; European Commission","keywords":"Connectome; Macaque; Connectomics; Tractography; Computer science; Human Connectome Project; Diffusion MRI; Resting state fMRI; Neuroscience; Scale (ratio); Tracing; Functional connectivity; Biology; Cartography; Magnetic resonance imaging; Geography; Medicine","score_opus":0.11977953947096279,"score_gpt":0.40320857664028026,"score_spread":0.28342903716931744,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2957869398","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5409583,0.00043309433,0.43246716,0.0019188933,0.00017213811,0.00022394002,0.004813553,0.0043366593,0.014676235],"genre_scores_gemma":[0.7854897,0.00035619264,0.20571832,0.00028789227,0.000045373887,0.00084631424,0.0033333898,0.0009175236,0.0030052683],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99990785,0.000036900667,0.0000041585727,0.00001834717,0.000020844069,0.000011883049],"domain_scores_gemma":[0.9995023,0.00028004794,0.000037617225,0.0000616811,0.000068991794,0.00004930198],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004496218,0.00048229264,0.00034910394,0.00075527566,0.000650314,0.00055229384,0.0011115294,0.0010950346,0.0046307333],"category_scores_gemma":[0.0022555206,0.00034924046,0.00076166,0.00057567627,0.0005185508,0.00059399876,0.0010084411,0.000762691,0.00030822755],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059624006,0.000061744024,0.0021203083,0.00011120019,0.000065519074,0.0002628643,0.00024062391,0.9607181,0.005224232,0.021563176,0.0037309742,0.005841607],"study_design_scores_gemma":[0.000017144504,0.000013413338,0.000571825,0.000010177757,0.000006212574,0.000038339127,0.000019936622,0.9881991,0.0005594629,0.00788427,0.002669943,0.00001018266],"about_ca_topic_score_codex":0.0163245,"about_ca_topic_score_gemma":0.019112041,"teacher_disagreement_score":0.0163245,"about_ca_system_score_codex":0.00076278835,"about_ca_system_score_gemma":0.0010499811,"threshold_uncertainty_score":0.03245896},"labels":[],"label_agreement":null},{"id":"W2962754235","doi":"10.1038/s41597-019-0217-0","title":"An accurate registration of the BigBrain dataset with the MNI PD25 and ICBM152 atlases","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Medical Image Segmentation Techniques","field":"Computer Science","cited_by":81,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital; Western University","funders":"Canadian Institutes of Health Research","keywords":"Computer science; Segmentation; Brain atlas; Atlas (anatomy); Image registration; Magnetic resonance imaging; Artificial intelligence; Brain mapping; Pattern recognition (psychology); Neuroscience; Medicine; Biology; Anatomy; Radiology; Image (mathematics)","score_opus":0.037542854995450604,"score_gpt":0.3138715182451634,"score_spread":0.2763286632497128,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2962754235","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029171024,0.002583716,0.12080203,0.0014261628,0.0011392005,0.0011566638,0.76433206,0.05982179,0.019567406],"genre_scores_gemma":[0.03237657,0.0006458132,0.08746352,0.0002695542,0.000104892526,0.0011560577,0.86916196,0.0033299478,0.005491695],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9978942,0.00026486875,0.00024991107,0.00072075805,0.0006937745,0.00017648758],"domain_scores_gemma":[0.99849594,0.00020845397,0.00012205193,0.00062602514,0.00046514874,0.000082420884],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019294296,0.0023481795,0.0014702037,0.0046069734,0.0012171756,0.0026050617,0.0026618063,0.002022856,0.014618718],"category_scores_gemma":[0.0052169915,0.00082944456,0.0018639314,0.003816424,0.0008728149,0.001654754,0.0032405988,0.001958941,0.022784527],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046735365,0.00020469728,0.004674274,0.0017719545,0.00037100777,0.00045738675,0.000416508,0.009026266,0.010305055,0.0065349955,0.7956337,0.17013694],"study_design_scores_gemma":[0.00025702358,0.00019794343,0.022567581,0.000778282,0.0002505941,0.0025169365,0.0005235561,0.0371004,0.024698714,0.019060368,0.89177746,0.00027114214],"about_ca_topic_score_codex":0.01811305,"about_ca_topic_score_gemma":0.04098493,"teacher_disagreement_score":0.01811305,"about_ca_system_score_codex":0.001549807,"about_ca_system_score_gemma":0.0033923115,"threshold_uncertainty_score":0.04890448},"labels":[],"label_agreement":null},{"id":"W2968251458","doi":"10.1038/s41597-019-0150-2","title":"MOD-LSP, MODIS-based parameters for hydrologic modeling of North American land cover change","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Science Foundation","keywords":"Land cover; Environmental science; Cover (algebra); Remote sensing; Physical geography; Geography; Hydrology (agriculture); Land use; Ecology; Geology; Biology","score_opus":0.06376587501712246,"score_gpt":0.25469549726520796,"score_spread":0.19092962224808552,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2968251458","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20920758,0.0011370932,0.35627568,0.001277977,0.0009496882,0.0011459,0.3378,0.033695407,0.058510598],"genre_scores_gemma":[0.49864808,0.0010590757,0.289972,0.00040852002,0.00025037405,0.0017879296,0.18691553,0.0044254637,0.016533153],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99986887,0.000026389032,0.000011005978,0.000033283704,0.00004808649,0.000012414733],"domain_scores_gemma":[0.99973553,0.000046117275,0.00003836661,0.000049204915,0.00011027608,0.000020468144],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005475075,0.00063069287,0.00032383698,0.00048359382,0.0003350718,0.00069844833,0.0010395715,0.00033692166,0.0050746608],"category_scores_gemma":[0.00093223975,0.00035968382,0.00045538926,0.0011901831,0.00016000863,0.0010537334,0.00033985716,0.0006288397,0.0020752272],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017203504,0.00026808435,0.05697059,0.00031763583,0.00029990528,0.00017575952,0.00021635539,0.44355482,0.006027284,0.013312653,0.33549577,0.14318912],"study_design_scores_gemma":[0.00012635945,0.00002595883,0.025753919,0.000029376828,0.000053117765,0.00006944989,0.00007945637,0.8196609,0.0031055708,0.008155084,0.14289457,0.00004622219],"about_ca_topic_score_codex":0.028798634,"about_ca_topic_score_gemma":0.035365567,"teacher_disagreement_score":0.028798634,"about_ca_system_score_codex":0.00071692007,"about_ca_system_score_gemma":0.001294825,"threshold_uncertainty_score":0.057262003},"labels":[],"label_agreement":null},{"id":"W2971384141","doi":"10.1038/s41597-019-0174-7","title":"Creating reproducible pharmacogenomic analysis pipelines","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vector Institute; Institute of Cancer Research; Ontario Institute for Cancer Research; University Health Network; University of Toronto; Princess Margaret Cancer Centre","funders":"","keywords":"Workflow; Computer science; Pipeline (software); Scalability; Identifier; Pharmacogenomics; Data science; Process (computing); Pipeline transport; Metadata; Data access; Data mining; Database; Bioinformatics; World Wide Web; Engineering; Biology; Programming language","score_opus":0.2945933816599378,"score_gpt":0.45496221309025664,"score_spread":0.16036883143031883,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971384141","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011139224,0.00025178608,0.9542553,0.0017563702,0.00020360295,0.0010519461,0.0064695813,0.021974944,0.0028973094],"genre_scores_gemma":[0.090398364,0.00073981617,0.88163936,0.00067413534,0.00013625981,0.0017934791,0.017080395,0.005631137,0.0019070887],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98322785,0.0039386027,0.0026891232,0.003389529,0.005732147,0.0010227514],"domain_scores_gemma":[0.9483531,0.0137394825,0.0034871325,0.021335123,0.011568001,0.0015172518],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.043982837,0.0011750965,0.0015715918,0.004635633,0.0019540025,0.008135394,0.004047741,0.0016403627,0.0044176844],"category_scores_gemma":[0.072302215,0.0017552193,0.0037971584,0.0052850703,0.0028426596,0.0068828943,0.00841186,0.0042972737,0.003278712],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018386622,0.000982389,0.032068323,0.0018152059,0.0011519064,0.0020183346,0.0025172643,0.12752764,0.07834107,0.2219639,0.08317779,0.44659755],"study_design_scores_gemma":[0.00071195414,0.00034382674,0.0068255556,0.00037810445,0.00036720186,0.0005357172,0.0006627383,0.29620045,0.13764973,0.3041128,0.25170505,0.00050693745],"about_ca_topic_score_codex":0.0050338274,"about_ca_topic_score_gemma":0.0045801452,"teacher_disagreement_score":0.95601714,"about_ca_system_score_codex":0.0034596685,"about_ca_system_score_gemma":0.0167372,"threshold_uncertainty_score":0.23260623},"labels":[],"label_agreement":null},{"id":"W2980534007","doi":"10.1038/s41597-019-0213-4","title":"The FluPRINT dataset, a multidimensional analysis of the influenza vaccine imprint on the immune system","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Influenza Virus Research Studies","field":"Medicine","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute of Infection and Immunity","funders":"National Center for Advancing Translational Sciences; National Center for Research Resources; National Institutes of Health; National Institute of Allergy and Infectious Diseases; European Commission; Georgia Clinical and Translational Science Alliance; Howard Hughes Medical Institute","keywords":"Immunogenicity; Influenza vaccine; Immune system; Immunology; Biology; Computational biology; Immunity; Vaccination; Virology","score_opus":0.10350322837359491,"score_gpt":0.38000786687867,"score_spread":0.2765046385050751,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2980534007","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.046561155,0.0020630425,0.003822913,0.00041077338,0.00017681437,0.00019925421,0.9419962,0.0020845174,0.0026852943],"genre_scores_gemma":[0.038002536,0.00046327358,0.005249683,0.00019525063,0.0000399859,0.00032801647,0.9549768,0.00008214947,0.0006622792],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99931073,0.00014168459,0.00008629577,0.0002044676,0.00014805517,0.00010874074],"domain_scores_gemma":[0.99919456,0.0002860738,0.000100847654,0.00019384598,0.00012388868,0.00010078901],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010064379,0.0012347437,0.0011132896,0.0022897546,0.00066877756,0.0012226094,0.0014060048,0.0013542365,0.0042104768],"category_scores_gemma":[0.0027057838,0.00032332088,0.0016492152,0.0017554055,0.0003298721,0.0005740952,0.0013978905,0.0010960371,0.0034364231],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003509025,0.0009881306,0.12453814,0.0055455277,0.0018244932,0.0013129249,0.00038655763,0.017465843,0.033793315,0.0028592069,0.7127757,0.09500113],"study_design_scores_gemma":[0.0010663093,0.0015476656,0.27264136,0.00078775873,0.0009934555,0.0025570309,0.0007842491,0.043981288,0.02275025,0.0075624874,0.6449573,0.00037087654],"about_ca_topic_score_codex":0.008574372,"about_ca_topic_score_gemma":0.012955799,"teacher_disagreement_score":0.008574372,"about_ca_system_score_codex":0.00064029807,"about_ca_system_score_gemma":0.0011430597,"threshold_uncertainty_score":0.017048895},"labels":[],"label_agreement":null},{"id":"W2981232846","doi":"10.1038/s41597-019-0181-8","title":"Bile acids targeted metabolomics and medication classification data in the ADNI1 and ADNIGO/2 cohorts","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; Genentech; IXICO; H. Lundbeck A/S; Servier; Eisai; Northern California Institute for Research and Education; University of California, San Diego; Pfizer; Biogen; BioClinica; F. Hoffmann-La Roche; University of Southern California; Novartis Pharmaceuticals Corporation; U.S. Department of Defense; Eli Lilly and Company; Bristol-Myers Squibb; Alzheimer's Disease Neuroimaging Initiative; Meso Scale Diagnostics; Alzheimer's Association; National Institutes of Health; U.S. Department of Health and Human Services; Foundation for the National Institutes of Health","keywords":"Metabolomics; Disease; Dementia; Mechanism (biology); Alzheimer's Disease Neuroimaging Initiative; Neuroimaging; Cohort; Bioinformatics; Medicine; Computational biology; Biology; Internal medicine; Neuroscience","score_opus":0.03761647439491405,"score_gpt":0.2914497505758763,"score_spread":0.25383327618096224,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2981232846","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90945035,0.00045455777,0.00084946654,0.00032474377,0.000027953673,0.00009545865,0.08720295,0.0001319049,0.0014625788],"genre_scores_gemma":[0.7501334,0.00035543172,0.0036967972,0.0002339,0.000051779258,0.00032062308,0.24370153,0.00013589197,0.0013706152],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99885786,0.00020280169,0.00013832236,0.00040501717,0.0002597537,0.00013629219],"domain_scores_gemma":[0.99788505,0.00028368164,0.0005369083,0.00067532726,0.0003427842,0.00027627684],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019215946,0.00059811125,0.0007953168,0.0018403691,0.0005484092,0.001192124,0.0006535059,0.00058522774,0.0017053953],"category_scores_gemma":[0.0064008543,0.0002638615,0.0006330702,0.0015689736,0.00033614718,0.00026890426,0.0013735553,0.0007018811,0.00072792236],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005634696,0.00028013735,0.9532722,0.00017815581,0.0011693971,0.00055184733,0.0003553172,0.0012939806,0.0058962903,0.00064216205,0.014806925,0.015918784],"study_design_scores_gemma":[0.00019106483,0.00017513758,0.98785305,0.000040273462,0.00029594448,0.0007175801,0.00018409132,0.00093510683,0.0013793162,0.0005704207,0.0076155677,0.000042469117],"about_ca_topic_score_codex":0.013138121,"about_ca_topic_score_gemma":0.01792692,"teacher_disagreement_score":0.013138121,"about_ca_system_score_codex":0.00053317065,"about_ca_system_score_gemma":0.0007135525,"threshold_uncertainty_score":0.026123285},"labels":[],"label_agreement":null},{"id":"W2981521947","doi":"10.1038/s41597-019-0206-3","title":"A database for using machine learning and data mining techniques for coronary artery disease diagnosis","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Artificial Intelligence in Healthcare","field":"Health Professions","cited_by":107,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Université du Québec à Montréal","funders":"","keywords":"Coronary artery disease; Computer science; Disease; Database; Artificial intelligence; Information retrieval; Medicine; Internal medicine","score_opus":0.40365149775140063,"score_gpt":0.5364813959138394,"score_spread":0.13282989816243873,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2981521947","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060756025,0.005208482,0.04531861,0.0012917422,0.00055427215,0.0010386026,0.9109825,0.018450595,0.011079582],"genre_scores_gemma":[0.01353913,0.0033531322,0.07228238,0.0007155453,0.00026414447,0.001248894,0.9051983,0.0009517363,0.002446703],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.994635,0.0008093518,0.0024661375,0.00066245307,0.0012454385,0.00018167181],"domain_scores_gemma":[0.9731609,0.009836222,0.004136905,0.0054460336,0.0055704103,0.0018494779],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0042694667,0.0016115268,0.0021023874,0.01746431,0.0010741748,0.0042375973,0.0030226216,0.0019246322,0.019554254],"category_scores_gemma":[0.029260108,0.0007162143,0.0013054662,0.016613677,0.00041005327,0.004147949,0.0029596288,0.0019529818,0.017536106],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010529552,0.00046233786,0.016082054,0.005355937,0.00048739393,0.00074452243,0.000299118,0.0032151486,0.0045253253,0.011396864,0.7047892,0.2515892],"study_design_scores_gemma":[0.00043884257,0.00023300486,0.015580898,0.0011151579,0.00035036833,0.001148636,0.00025952925,0.006093452,0.004523474,0.012908207,0.95715624,0.00019222416],"about_ca_topic_score_codex":0.0040205186,"about_ca_topic_score_gemma":0.006017657,"teacher_disagreement_score":0.019554254,"about_ca_system_score_codex":0.00094199425,"about_ca_system_score_gemma":0.004167208,"threshold_uncertainty_score":0.0654155},"labels":[],"label_agreement":null},{"id":"W2981564286","doi":"10.1038/s41597-019-0242-z","title":"Integrated open-source software for multiscale electrophysiology","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neural dynamics and brain function","field":"Neuroscience","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"National Institute of Biomedical Imaging and Bioengineering; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Molson Foundation","keywords":"Open source software; Open source; Computer science; Software; Data science; Programming language","score_opus":0.06402038037987169,"score_gpt":0.300269165734127,"score_spread":0.2362487853542553,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2981564286","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001495407,0.00023489467,0.76891273,0.0002109236,0.000233646,0.00018597404,0.005147043,0.22012061,0.003458711],"genre_scores_gemma":[0.042817768,0.0008416948,0.77547926,0.00084433734,0.000263701,0.0027384344,0.030578911,0.13154273,0.014893147],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980718,0.00021049088,0.00028581178,0.0004314242,0.0008512638,0.00014922723],"domain_scores_gemma":[0.99639004,0.0015633297,0.0002493407,0.0007473551,0.0008029099,0.0002469991],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028691073,0.0017073911,0.0012707151,0.002059456,0.00063128676,0.0022816034,0.0039899447,0.0019206417,0.053342212],"category_scores_gemma":[0.009737268,0.0012051471,0.0021640202,0.0012074411,0.0009255004,0.002562099,0.0050550313,0.0033769398,0.025439091],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010656773,0.00042017494,0.0032761025,0.0029793843,0.0009073779,0.0011744337,0.0012511322,0.023054618,0.07835396,0.047045797,0.37213597,0.4683354],"study_design_scores_gemma":[0.0005991383,0.00018441271,0.005922872,0.00061355997,0.00023739795,0.0017492287,0.00013000683,0.20812106,0.0620878,0.10540313,0.61442465,0.0005266889],"about_ca_topic_score_codex":0.0011682175,"about_ca_topic_score_gemma":0.0015693465,"teacher_disagreement_score":0.053342212,"about_ca_system_score_codex":0.00054309296,"about_ca_system_score_gemma":0.0019945758,"threshold_uncertainty_score":0.17844748},"labels":[],"label_agreement":null},{"id":"W2981959691","doi":"10.1038/s41597-019-0241-0","title":"Distributed radiomics as a signature validation study using the Personal Health Train infrastructure","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Radiomics and Machine Learning in Medical Imaging","field":"Medicine","cited_by":67,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre","funders":"","keywords":"Radiomics; Computer science; Signature (topology); Data mining; Artificial intelligence; Mathematics","score_opus":0.03081040018237908,"score_gpt":0.3496671711163765,"score_spread":0.31885677093399745,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2981959691","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89823145,0.00027417231,0.09508477,0.0006568157,0.00016010963,0.00035195478,0.0016765861,0.0012289765,0.0023351777],"genre_scores_gemma":[0.9818032,0.00003580546,0.015805602,0.00007236186,0.00003801612,0.000079892365,0.0016051264,0.000029296565,0.00053065567],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.99716944,0.0012876982,0.00014238337,0.0006853271,0.00049268664,0.00022245111],"domain_scores_gemma":[0.99371225,0.0020441017,0.0007231584,0.0021238385,0.0009596382,0.00043688988],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047824364,0.0005518024,0.00064640935,0.000782558,0.00033221988,0.000998722,0.0013832531,0.001006087,0.0010610233],"category_scores_gemma":[0.009878643,0.00017816747,0.0005593488,0.0007615177,0.0006560026,0.0009651993,0.0011654255,0.00080280274,0.0004764557],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002627222,0.0017925303,0.17123681,0.0003518127,0.0006498237,0.0011255854,0.0005048535,0.6067833,0.015135755,0.0073321406,0.012583257,0.17987695],"study_design_scores_gemma":[0.0002617003,0.0011397429,0.041272327,0.000035269353,0.00010211279,0.0006154444,0.0003182021,0.9363297,0.011175378,0.0033538733,0.0053423382,0.00005386578],"about_ca_topic_score_codex":0.005429755,"about_ca_topic_score_gemma":0.0031228215,"teacher_disagreement_score":0.005429755,"about_ca_system_score_codex":0.00096711173,"about_ca_system_score_gemma":0.0010712811,"threshold_uncertainty_score":0.025292277},"labels":[],"label_agreement":null},{"id":"W2982168029","doi":"10.1038/s41597-019-0214-3","title":"Global humid tropics forest structural condition and forest structural integrity maps","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":85,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Northern British Columbia","funders":"Nuclear Safety and Security Commission; National Aeronautics and Space Administration","keywords":"Canopy; Biodiversity; Forest ecology; Tropics; Tree canopy; Ecosystem services; Agroforestry; Disturbance (geology); Secondary forest; Environmental science; Ecosystem; Environmental resource management; Geography; Ecology; Biology","score_opus":0.01630188727024519,"score_gpt":0.2600394197173262,"score_spread":0.24373753244708102,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2982168029","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5850234,0.0016912501,0.0062905797,0.0005431283,0.000061183746,0.00034539375,0.3412794,0.00080523296,0.06396044],"genre_scores_gemma":[0.8692681,0.00058351294,0.012677698,0.00013811736,0.00006911485,0.0002406541,0.113868654,0.00008625435,0.0030678513],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997894,0.000034231416,0.00001619765,0.000043834665,0.0000818709,0.000034412624],"domain_scores_gemma":[0.9994504,0.00004617871,0.00019612786,0.00006248974,0.00017008389,0.00007474422],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00047846578,0.00037510553,0.000134733,0.0026431836,0.00016364794,0.0005955322,0.00024698456,0.00016119436,0.0041940287],"category_scores_gemma":[0.0007207859,0.0001008295,0.00014370082,0.0037640564,0.0001882569,0.0004947086,0.00046777376,0.00020722805,0.00065002305],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024160724,0.00008399792,0.79631925,0.0004911221,0.00019083364,0.00032519375,0.0011038986,0.009695567,0.0060686814,0.0046225544,0.040458668,0.14039859],"study_design_scores_gemma":[0.000014560194,0.000021937969,0.97109795,0.000046322362,0.000014425492,0.00013950402,0.00037141886,0.0026893923,0.00040399402,0.00039842105,0.024790283,0.000011969125],"about_ca_topic_score_codex":0.047710016,"about_ca_topic_score_gemma":0.05202126,"teacher_disagreement_score":0.047710016,"about_ca_system_score_codex":0.00046199036,"about_ca_system_score_gemma":0.00036075077,"threshold_uncertainty_score":0.09486461},"labels":[],"label_agreement":null},{"id":"W2982600860","doi":"10.1038/s41597-019-0262-8","title":"Structural and functional multi-platform MRI series of a single human volunteer over more than fifteen years","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced MRI Techniques and Applications","field":"Medicine","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université de Sherbrooke; University of Alberta; Institut Universitaire de Gériatrie de Montréal; Université Laval; Institut Universitaire en Santé Mentale de Québec","funders":"Canadian Institutes of Health Research; Canada Research Chairs; University of Electronic Science and Technology of China; Government of Canada; Government of Ontario; Alzheimer's Society; Fonds de Recherche du Québec - Santé; Pfizer Canada; Pfizer; Ontario Brain Institute; Courtois Foundation; Consortium canadien en neurodégénérescence associée au vieillissement","keywords":"Volunteer; Series (stratigraphy); Computer science; Computational biology; Biology; Ecology; Paleontology","score_opus":0.0722936045505076,"score_gpt":0.3349037215651771,"score_spread":0.2626101170146695,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2982600860","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9883108,0.0004502825,0.0052545173,0.00016724163,0.00003113721,0.00011582103,0.0027283616,0.00012705527,0.0028148657],"genre_scores_gemma":[0.9876765,0.00025804163,0.0044027762,0.00012695785,0.00004669289,0.00012103399,0.005063487,0.000061024824,0.002243408],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998265,0.000017763597,0.000014077768,0.00007264934,0.000042160165,0.000026925001],"domain_scores_gemma":[0.99918765,0.00010962578,0.000103527556,0.00022467872,0.00023807648,0.00013644228],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006592052,0.00038362446,0.00042941349,0.0009382206,0.000775309,0.00047342607,0.00037110815,0.0006592589,0.0016661421],"category_scores_gemma":[0.0011613645,0.00026007433,0.00027468606,0.00062734226,0.0005168018,0.0004528726,0.0006436338,0.00039805405,0.00076528534],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0040098783,0.002037951,0.3931256,0.00080247753,0.0013845259,0.018931108,0.009812344,0.0042170985,0.37962452,0.0012764866,0.013373561,0.17140444],"study_design_scores_gemma":[0.000060453745,0.0024747194,0.9474048,0.000039643888,0.0003048365,0.01509533,0.0015287303,0.0013338528,0.01459243,0.0012248867,0.015836738,0.00010367682],"about_ca_topic_score_codex":0.004697516,"about_ca_topic_score_gemma":0.011642921,"teacher_disagreement_score":0.004697516,"about_ca_system_score_codex":0.00016925881,"about_ca_system_score_gemma":0.00023690233,"threshold_uncertainty_score":0.009340346},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2990212745","doi":"10.1038/s41597-019-0310-4","title":"Benefits and barriers in the design of harmonized access agreements for international data sharing","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Neurological Institute and Hospital; McGill University and Génome Québec Innovation Centre; McGill University; McGill University Health Centre","funders":"Canadian Institutes of Health Research; University College London Hospitals NHS Foundation Trust; National Institute for Health and Care Research; Horizon 2020 Framework Programme; McGill University","keywords":"Data sharing; Context (archaeology); Corporate governance; Core (optical fiber); Data access; Set (abstract data type); Computer science; Business; Data science; Database; Telecommunications; Medicine","score_opus":0.8224288789402373,"score_gpt":0.60965478037303,"score_spread":0.2127740985672073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2990212745","genre_codex":"methods","genre_gemma":"empirical","domain_codex":"methods","domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.067000784,0.0028085134,0.5644769,0.24807149,0.0016798484,0.0057324143,0.0002838178,0.0004663676,0.1094799],"genre_scores_gemma":[0.58058584,0.0012078332,0.3697811,0.033831205,0.0010259456,0.0063665486,0.000302216,0.00033562764,0.0065637105],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.41270813,0.4423372,0.05144475,0.018169023,0.059245054,0.016095871],"domain_scores_gemma":[0.3554321,0.4389127,0.036991417,0.092375636,0.06252695,0.013761208],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.5028338,0.00092395634,0.0018395256,0.0031464144,0.010568718,0.029431459,0.007931523,0.013635239,0.004352682],"category_scores_gemma":[0.50749123,0.002317612,0.002507261,0.004245403,0.02618363,0.032866515,0.024693502,0.023175681,0.0011793858],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000071599425,0.000106582964,0.003032471,0.00025315903,0.00005428996,0.00024339918,0.007749892,0.0026012065,0.0005235724,0.9631781,0.0040536183,0.018132191],"study_design_scores_gemma":[0.00024070323,0.00026452853,0.0039681834,0.0031549903,0.0001140243,0.00078154553,0.012869923,0.010436147,0.0017585194,0.835269,0.13091674,0.00022567532],"about_ca_topic_score_codex":0.0050831637,"about_ca_topic_score_gemma":0.004551451,"teacher_disagreement_score":0.99206847,"about_ca_system_score_codex":0.013966128,"about_ca_system_score_gemma":0.05335276,"threshold_uncertainty_score":0.6130945},"labels":[],"label_agreement":null},{"id":"W2991126022","doi":"10.1038/s41597-019-0312-2","title":"A global wildfire dataset for the analysis of fire regimes and fire behaviour","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":366,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Nutrition International","funders":"","keywords":"Environmental science; Work (physics); Product (mathematics); Computer science; Remote sensing; Meteorology; Geography; Engineering; Mathematics","score_opus":0.017380249280836636,"score_gpt":0.26873818777623465,"score_spread":0.251357938495398,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2991126022","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0148285255,0.00018967094,0.00203768,0.0000938292,0.000059554848,0.00014893465,0.9791987,0.0013837777,0.0020593898],"genre_scores_gemma":[0.0075414265,0.00005368039,0.003543364,0.000022727407,0.000009959511,0.00013633844,0.98822856,0.000065929715,0.00039790617],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993662,0.000065559405,0.00010911941,0.00016050288,0.00022529847,0.000073322],"domain_scores_gemma":[0.9988219,0.00017038088,0.0001984987,0.00033199633,0.00032306163,0.00015418473],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008914504,0.0012522008,0.00079295307,0.0028989261,0.0004959955,0.0006415741,0.0013519357,0.0010153456,0.005761836],"category_scores_gemma":[0.002068658,0.0003018248,0.0011701871,0.0034618885,0.00025427638,0.00070792966,0.0009336002,0.0010970628,0.0066791167],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006947281,0.00079596136,0.07292207,0.0021082663,0.0005459594,0.00073618395,0.0003644392,0.012963289,0.008890284,0.0031101438,0.8296214,0.06724725],"study_design_scores_gemma":[0.0005398373,0.00030040793,0.24585316,0.00043375758,0.00017677226,0.001059074,0.000498633,0.028830217,0.00981533,0.003931904,0.7084104,0.00015059259],"about_ca_topic_score_codex":0.014965955,"about_ca_topic_score_gemma":0.02914061,"teacher_disagreement_score":0.014965955,"about_ca_system_score_codex":0.00064496446,"about_ca_system_score_gemma":0.0011481079,"threshold_uncertainty_score":0.029757679},"labels":[],"label_agreement":null},{"id":"W2993872008","doi":"10.1038/s41597-019-0300-6","title":"Global hydro-environmental sub-basin and river reach characteristics at high spatial resolution","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":660,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Australian Research Council; Department of Education and Training; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Compendium; Structural basin; Drainage basin; Land cover; Hydrology (agriculture); Environmental science; Spatial analysis; Computer science; Environmental resource management; Land use; Database; Remote sensing; Geography; Cartography; Geology; Ecology","score_opus":0.01102404109151187,"score_gpt":0.20506397169694648,"score_spread":0.19403993060543462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2993872008","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024861814,0.00007902688,0.0070315753,0.00009889077,0.00002017078,0.00008545014,0.9563112,0.002796146,0.008715661],"genre_scores_gemma":[0.071633585,0.00013272354,0.013256584,0.000048722486,0.000019462033,0.0004773549,0.91032547,0.0006142484,0.0034918257],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9995554,0.00004509187,0.00006611465,0.000112911024,0.00017172763,0.000048770955],"domain_scores_gemma":[0.9990101,0.00011503227,0.00015049553,0.0002700343,0.00037601733,0.000078373414],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00052404485,0.00042262694,0.00048093701,0.0043883906,0.00023097597,0.00095613766,0.00062118465,0.00032385797,0.015023903],"category_scores_gemma":[0.0016979956,0.00023264757,0.00049316924,0.007871198,0.00018636415,0.0009028268,0.0010517702,0.00041139623,0.0063494695],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040968382,0.00012246528,0.10642864,0.0014937947,0.00038084367,0.00045316954,0.0009847829,0.030969748,0.010164082,0.018034806,0.6740935,0.15646447],"study_design_scores_gemma":[0.0001034661,0.00003108981,0.20425245,0.00017350957,0.00010048536,0.00022100001,0.0006049409,0.012430695,0.0067066248,0.0066107605,0.7686149,0.00015007699],"about_ca_topic_score_codex":0.02349462,"about_ca_topic_score_gemma":0.022761602,"teacher_disagreement_score":0.02349462,"about_ca_system_score_codex":0.00058589835,"about_ca_system_score_gemma":0.0014866487,"threshold_uncertainty_score":0.050260007},"labels":[],"label_agreement":null},{"id":"W2994025393","doi":"10.1038/s41597-019-0316-y","title":"CE-QUAL-W2 model of dam outflow elevation impact on temperature, dissolved oxygen and nutrients in a reservoir","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrological Forecasting Using AI","field":"Environmental Science","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Global Institute for Water Security; University of Saskatchewan","funders":"Global Institute for Water Security, University of Saskatchewan; Global Water Futures; University of Saskatchewan","keywords":"Outflow; Nutrient; Environmental science; Elevation (ballistics); Hydrology (agriculture); Oxygen; Oceanography; Ecology; Geology; Chemistry; Biology; Geotechnical engineering; Mathematics","score_opus":0.04989829200182872,"score_gpt":0.29832687360129106,"score_spread":0.24842858159946235,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2994025393","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.80998546,0.00027064653,0.016679682,0.0015466361,0.00021436485,0.00020862867,0.14556725,0.0037634852,0.021763843],"genre_scores_gemma":[0.93886507,0.00013414033,0.009026037,0.00019986299,0.000023725683,0.00017767012,0.048110012,0.00017130618,0.0032922907],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998074,0.000034508877,0.00001272978,0.000057538415,0.000032068714,0.00005577724],"domain_scores_gemma":[0.9992712,0.00024361021,0.00005398972,0.00005796777,0.00026599696,0.00010722965],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00059226097,0.0009101025,0.0007152086,0.0006034692,0.0004770083,0.00092316436,0.0019447041,0.0013172532,0.004886778],"category_scores_gemma":[0.0015500771,0.0005162654,0.00096227217,0.0013676201,0.00052498025,0.00062963227,0.00045469526,0.00093149306,0.00057156547],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000078602854,0.00006880898,0.006307815,0.000038454386,0.000034885765,0.00004677812,0.000015227834,0.9854403,0.00038740376,0.0007457539,0.0054113525,0.0014246237],"study_design_scores_gemma":[0.00008866254,0.000014913654,0.002416693,0.0000056000868,0.000009838154,0.0000055399237,0.000019566516,0.9955859,0.00028043793,0.00028575238,0.0012731659,0.000013902658],"about_ca_topic_score_codex":0.3925209,"about_ca_topic_score_gemma":0.28883904,"teacher_disagreement_score":0.3925209,"about_ca_system_score_codex":0.0028534255,"about_ca_system_score_gemma":0.0036232704,"threshold_uncertainty_score":0.78047234},"labels":[],"label_agreement":null},{"id":"W3000434011","doi":"10.1038/s41597-019-0344-7","title":"A global database for metacommunity ecology, integrating species, traits, environment and space","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":64,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; Université de Montréal","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico; Deutsche Forschungsgemeinschaft; Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Metacommunity; Ecology; Trait; Biodiversity; Ecosystem; Taxon; Community; Database; Biology; Macroecology; Spatial ecology; Geography; Computer science; Biological dispersal","score_opus":0.11391963621065433,"score_gpt":0.28919924332291835,"score_spread":0.17527960711226404,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3000434011","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002975977,0.00041941018,0.0058398396,0.000105905536,0.000030028667,0.00007065845,0.9872745,0.001843295,0.0014402563],"genre_scores_gemma":[0.0050223405,0.00024731248,0.00954261,0.00006459504,0.000011904815,0.0002474297,0.9843597,0.00017851497,0.00032559433],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99818677,0.0002837923,0.00060087367,0.0004151039,0.00038223751,0.00013118789],"domain_scores_gemma":[0.9939177,0.0016309638,0.0012777358,0.0013577114,0.001131942,0.00068389886],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019305262,0.0013782005,0.0017604915,0.009558388,0.0007258377,0.0026400504,0.0026428292,0.0012739057,0.012423692],"category_scores_gemma":[0.011027141,0.00072282366,0.0013085608,0.014511699,0.00036952604,0.002953732,0.0034112518,0.0015579084,0.009216169],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006741955,0.00023526045,0.058751963,0.00963573,0.0011249518,0.0005397948,0.00079503225,0.0064385184,0.008220137,0.017627632,0.7912259,0.10473086],"study_design_scores_gemma":[0.0002163204,0.00006519518,0.05228285,0.00081811496,0.00027931778,0.000499874,0.0003624631,0.005280606,0.0022466527,0.011438243,0.92640907,0.0001012817],"about_ca_topic_score_codex":0.009242784,"about_ca_topic_score_gemma":0.014682675,"teacher_disagreement_score":0.012423692,"about_ca_system_score_codex":0.0009721151,"about_ca_system_score_gemma":0.003078232,"threshold_uncertainty_score":0.041561425},"labels":[],"label_agreement":null},{"id":"W3005230983","doi":"10.1038/s41597-020-0375-0","title":"Nexus of grazing management with plant and soil properties in northern China grasslands","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Carbon and Nitrogen Dynamics","field":"Agricultural and Biological Sciences","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Agriculture and Agri-Food Canada","funders":"Agriculture and Agri-Food Canada; Gansu Agricultural University","keywords":"Grazing; Environmental science; Grassland; Biomass (ecology); Ecosystem; Agroforestry; Vegetation (pathology); Agronomy; Plant community; Ecology; Biology; Species richness","score_opus":0.039439262392051616,"score_gpt":0.19525176392634327,"score_spread":0.15581250153429166,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3005230983","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9170039,0.028704783,0.00042659626,0.0005384136,0.00004301797,0.000042003834,0.04943693,0.000035392,0.003768844],"genre_scores_gemma":[0.9775693,0.0036019632,0.0004088363,0.00013772793,0.00003600263,0.00006944754,0.017472493,0.0000057219463,0.0006985408],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9991762,0.00019253524,0.00019945753,0.00021111766,0.00013581212,0.000084822575],"domain_scores_gemma":[0.99831593,0.00062843994,0.0005289457,0.000153664,0.0002532689,0.00011975105],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017875938,0.00031731752,0.00047713882,0.0030478132,0.00027735185,0.0005168725,0.00042819965,0.00033800857,0.0014910181],"category_scores_gemma":[0.002233745,0.000120055745,0.00081569783,0.0049171904,0.00031720806,0.00038020083,0.00046080747,0.00012928137,0.00012072583],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000249454,0.00004262351,0.95759434,0.0029572404,0.0011293805,0.0005847967,0.0004941018,0.0009310008,0.0014582245,0.00044108054,0.003428693,0.030689074],"study_design_scores_gemma":[0.000018878596,0.000038824383,0.99039406,0.0002969138,0.00061266863,0.00013680279,0.00025747702,0.00055954885,0.00031872524,0.00018857786,0.0071614254,0.000015998541],"about_ca_topic_score_codex":0.03483645,"about_ca_topic_score_gemma":0.05312538,"teacher_disagreement_score":0.03483645,"about_ca_system_score_codex":0.00097631366,"about_ca_system_score_gemma":0.0014339329,"threshold_uncertainty_score":0.06926733},"labels":[],"label_agreement":null},{"id":"W3006404851","doi":"10.1038/s41597-020-0379-9","title":"High-resolution T2-FLAIR and non-contrast CT brain atlas of the elderly","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Medical Image Segmentation Techniques","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Foothills Medical Centre; University of Calgary","funders":"Canadian Open Neuroscience Platform; University of Calgary; Canada Research Chairs; Heart and Stroke Foundation of Canada","keywords":"Fluid-attenuated inversion recovery; Spatial normalization; Neuroimaging; Contrast (vision); Computer science; Modalities; Normalization (sociology); Artificial intelligence; Modality (human–computer interaction); Voxel; Brain atlas; Brain mapping; Pattern recognition (psychology); Medicine; Neuroscience; Magnetic resonance imaging; Radiology; Psychology","score_opus":0.030584969278722597,"score_gpt":0.27253003643218593,"score_spread":0.24194506715346334,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3006404851","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0969042,0.0027472835,0.8575717,0.00068089523,0.00024359566,0.00046278955,0.015236537,0.0048457975,0.021307249],"genre_scores_gemma":[0.3838488,0.004192167,0.5722892,0.00031534026,0.00017067092,0.00067043456,0.018408606,0.0015700399,0.018534744],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9997459,0.000049978487,0.00004061514,0.00006928432,0.00007305818,0.00002107341],"domain_scores_gemma":[0.99956363,0.00010242301,0.00006608759,0.00011490102,0.00012859545,0.000024362795],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00071114156,0.00044268643,0.00027421745,0.002098199,0.00040505952,0.0010813418,0.0005106883,0.000720149,0.0053843195],"category_scores_gemma":[0.0015608928,0.00022906663,0.00039766662,0.001824156,0.00036823386,0.00056067173,0.0005812878,0.0004577656,0.0022355204],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006912957,0.00015243486,0.015500509,0.0014915302,0.00031102702,0.0031980192,0.0028880981,0.031814825,0.15276092,0.03598422,0.06452248,0.6906847],"study_design_scores_gemma":[0.000081832986,0.00047340957,0.13511333,0.0004262924,0.00032895204,0.029733943,0.0015714212,0.073555455,0.13984501,0.05403051,0.56454676,0.0002930365],"about_ca_topic_score_codex":0.002993465,"about_ca_topic_score_gemma":0.0056050643,"teacher_disagreement_score":0.0053843195,"about_ca_system_score_codex":0.0004234571,"about_ca_system_score_gemma":0.001011738,"threshold_uncertainty_score":0.018012345},"labels":[],"label_agreement":null},{"id":"W3006538489","doi":"10.1038/s41597-020-0388-8","title":"The Rural Household Multiple Indicator Survey, data from 13,310 farm households in 21 countries","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Food Security and Health in Diverse Populations","field":"Health Professions","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Consortium of International Agricultural Research Centers; United States Agency for International Development","keywords":"Food security; Household income; Survey data collection; Livestock; Agriculture; Welfare; Productivity; Poverty; Index (typography); Agricultural economics; Geography; Production (economics); Scale (ratio); Agricultural productivity; Business; Socioeconomics; Economics; Economic growth; Statistics","score_opus":0.48539061247049725,"score_gpt":0.4517201502234974,"score_spread":0.03367046224699988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3006538489","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3824677,0.0004009871,0.0029200239,0.00027257667,0.000032569424,0.0013563951,0.5960415,0.00014104405,0.016367333],"genre_scores_gemma":[0.40528673,0.0011645317,0.009032288,0.00051192776,0.000048176935,0.0053887013,0.56451356,0.00006478675,0.0139894085],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99899226,0.00032981913,0.00019667091,0.00015444504,0.00022326405,0.00010353337],"domain_scores_gemma":[0.99765366,0.00043849036,0.0007801746,0.0002225322,0.0007262318,0.00017898894],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014782435,0.00029885347,0.0003065793,0.0021645927,0.00040746978,0.00042461028,0.00040974576,0.00029750032,0.0049318383],"category_scores_gemma":[0.0037324135,0.0002686731,0.00019209045,0.0057082586,0.00013605658,0.00048374876,0.00065151,0.0003601484,0.0025024107],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030770982,0.000301039,0.7885268,0.00073093484,0.00015156952,0.00021588421,0.0022015832,0.0014237105,0.0012360367,0.0010752825,0.13470174,0.06912776],"study_design_scores_gemma":[0.00005174634,0.000118478034,0.9626523,0.000094709605,0.000026153064,0.000110223344,0.0010633753,0.0003594382,0.00030731657,0.00013402192,0.035064764,0.000017577622],"about_ca_topic_score_codex":0.021352395,"about_ca_topic_score_gemma":0.026433969,"teacher_disagreement_score":0.021352395,"about_ca_system_score_codex":0.0005212071,"about_ca_system_score_gemma":0.0011167794,"threshold_uncertainty_score":0.04245627},"labels":[],"label_agreement":null},{"id":"W3007040014","doi":"10.1038/s41597-019-0346-5","title":"Global karst springs hydrograph dataset for research and management of the world’s fastest-flowing groundwater","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Karst Systems and Hydrogeology","field":"Earth and Planetary Sciences","cited_by":103,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"Natural Environment Research Council; Sight Research UK","keywords":"Karst; Aquifer; Groundwater; Hydrograph; Hydrology (agriculture); Spring (device); Water resource management; Environmental science; Groundwater flow; Environmental resource management; Geography; Geology; Cartography; Drainage basin; Engineering","score_opus":0.16147000265261754,"score_gpt":0.3372273776496487,"score_spread":0.17575737499703117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3007040014","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0053544464,0.000097009855,0.000524769,0.0001513038,0.000029985778,0.000026890564,0.99194545,0.000653796,0.0012163726],"genre_scores_gemma":[0.007500047,0.00006891375,0.0010609665,0.000028649632,0.000010470563,0.0000686384,0.99087536,0.00004332902,0.00034358105],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99957293,0.00005239742,0.00009536261,0.000106799234,0.00011309394,0.000059426286],"domain_scores_gemma":[0.9990439,0.00017446843,0.00015166451,0.00019675422,0.0003339671,0.00009936257],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00047395256,0.00068958953,0.0005312349,0.0022897392,0.00030089018,0.0006518324,0.0010651067,0.0007315393,0.0069508944],"category_scores_gemma":[0.0023072031,0.00021247084,0.00063691044,0.0046277693,0.00019880214,0.0008892891,0.0009571972,0.00056234264,0.0051901904],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001426316,0.000088165994,0.041086603,0.0010562907,0.00014230091,0.00020455405,0.00018470279,0.0044924854,0.0011025221,0.0017402072,0.9270036,0.022756059],"study_design_scores_gemma":[0.00027708104,0.00004202724,0.13107029,0.00033045653,0.000078606376,0.00022555978,0.0006270682,0.014844806,0.0015902347,0.0035462906,0.84727037,0.00009723343],"about_ca_topic_score_codex":0.036167618,"about_ca_topic_score_gemma":0.051033553,"teacher_disagreement_score":0.036167618,"about_ca_system_score_codex":0.000718613,"about_ca_system_score_gemma":0.0016808336,"threshold_uncertainty_score":0.071914196},"labels":[],"label_agreement":null},{"id":"W3007951719","doi":"10.1038/s41597-020-0415-9","title":"A NWB-based dataset and processing pipeline of human single-neuron activity during a declarative memory task","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neural dynamics and brain function","field":"Neuroscience","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Toronto Western Hospital","funders":"National Institute of Neurological Disorders and Stroke; National Institute of Mental Health; McKnight Foundation; Kavli Foundation; National Alliance for Research on Schizophrenia and Depression; Howard Hughes Medical Institute; National Science Foundation","keywords":"Task (project management); Pipeline (software); Computer science; Programming language","score_opus":0.1313626909694292,"score_gpt":0.30981835983481254,"score_spread":0.17845566886538333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3007951719","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.073130324,0.0005653884,0.14823712,0.0012508081,0.00043798005,0.000841714,0.71726674,0.052053984,0.006215932],"genre_scores_gemma":[0.094163634,0.00037453292,0.11645348,0.0003841888,0.00006190319,0.0020805092,0.78087634,0.002720829,0.0028846294],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99965906,0.000031314772,0.000059492762,0.000113260634,0.000100488825,0.000036485242],"domain_scores_gemma":[0.99894935,0.00018775329,0.0000724323,0.0003258706,0.0003541508,0.00011054006],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006488418,0.0009272471,0.0007017389,0.001299813,0.00059330935,0.00086269283,0.0012663661,0.00081809534,0.00904333],"category_scores_gemma":[0.0036078757,0.000490918,0.0008466682,0.0012492149,0.00036657017,0.0010868085,0.0019678578,0.0011255977,0.008079807],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024221505,0.00076434895,0.03923624,0.002568283,0.00049682095,0.0019306777,0.0011985146,0.021848237,0.12066719,0.0054933727,0.5514709,0.25190327],"study_design_scores_gemma":[0.00087729475,0.00085520226,0.19052827,0.0008437497,0.00030741497,0.0036406682,0.0010015629,0.17203797,0.14313582,0.034498475,0.45158175,0.00069178035],"about_ca_topic_score_codex":0.0072095627,"about_ca_topic_score_gemma":0.014964852,"teacher_disagreement_score":0.00904333,"about_ca_system_score_codex":0.00052947016,"about_ca_system_score_gemma":0.0018456023,"threshold_uncertainty_score":0.030252993},"labels":[],"label_agreement":null},{"id":"W3008267920","doi":"10.1038/s41597-020-00781-y","title":"Multimodal pathophysiological dataset of gradual cerebral ischemia in a cohort of juvenile pigs","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neurological Disease Mechanisms and Treatments","field":"Neuroscience","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"","keywords":"Juvenile; Pathophysiology; Ischemia; Cohort; Medicine; Biology; Internal medicine; Ecology","score_opus":0.07800368628217957,"score_gpt":0.31025987053816684,"score_spread":0.23225618425598726,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3008267920","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9616097,0.00025776177,0.005248198,0.00022774117,0.000056182806,0.0001296507,0.030998185,0.00023785928,0.0012346539],"genre_scores_gemma":[0.92463136,0.00035463634,0.005815459,0.00017242586,0.000060534214,0.00037756294,0.06722261,0.00003569244,0.0013297983],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99986815,0.000018897059,0.00001208668,0.0000502239,0.000023507679,0.00002718611],"domain_scores_gemma":[0.9997104,0.00006231577,0.000046085734,0.000049958417,0.000077111836,0.000054150587],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00030718994,0.00031090854,0.0003403046,0.00086051604,0.00022668712,0.00027995007,0.00030715243,0.00053983764,0.0012327901],"category_scores_gemma":[0.0005154504,0.00008922818,0.00029045978,0.0005041516,0.00027500498,0.0001317796,0.00045673567,0.0002482578,0.00042573767],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002635553,0.0009905215,0.7310391,0.0009240885,0.00078021485,0.011764675,0.00073609094,0.010333083,0.09604475,0.0008606325,0.027350554,0.11654073],"study_design_scores_gemma":[0.000050013972,0.0011859455,0.9632011,0.000103534054,0.0001518296,0.004626337,0.0007175737,0.011926623,0.006569832,0.0008093267,0.010594619,0.00006329531],"about_ca_topic_score_codex":0.002984558,"about_ca_topic_score_gemma":0.0061270134,"teacher_disagreement_score":0.002984558,"about_ca_system_score_codex":0.00021065598,"about_ca_system_score_gemma":0.00031212444,"threshold_uncertainty_score":0.0059343576},"labels":[],"label_agreement":null},{"id":"W3008978453","doi":"10.1038/s41597-020-0397-7","title":"Tesco Grocery 1.0, a large-scale dataset of grocery purchases in London","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Nutritional Studies and Diet","field":"Medicine","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bell (Canada)","funders":"","keywords":"Representativeness heuristic; Census; Scale (ratio); Grocery store; Geography; Business; Population; Marketing; Statistics; Environmental health; Medicine; Cartography; Mathematics","score_opus":0.0809298758706811,"score_gpt":0.32031605464777646,"score_spread":0.23938617877709537,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3008978453","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01377925,0.00041207246,0.00018713373,0.00012523012,0.000021898635,0.000033221026,0.9826362,0.00019289565,0.0026120383],"genre_scores_gemma":[0.0097848885,0.00018554887,0.00068322103,0.00005493708,0.000013040105,0.000081849226,0.98739386,0.000050505638,0.0017520891],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994741,0.000077531644,0.00008450338,0.00016225476,0.00013024264,0.00007140234],"domain_scores_gemma":[0.99862885,0.00025070232,0.000384476,0.00019230062,0.0003647429,0.000178917],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003192325,0.0008404363,0.000728664,0.003826297,0.000351908,0.0015371828,0.0012043541,0.0012160086,0.017531293],"category_scores_gemma":[0.0036974875,0.00035509604,0.00063804653,0.0072802673,0.0003333204,0.0008496755,0.0014835995,0.00065080286,0.023113849],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00058866444,0.00012735753,0.09532566,0.0023927807,0.00026468968,0.00087934785,0.0007320111,0.0015661472,0.0012052318,0.0017846612,0.8760948,0.019038664],"study_design_scores_gemma":[0.00024975048,0.00008863247,0.18694493,0.0006394103,0.000078325,0.0005935951,0.0013770787,0.0020542748,0.0007934216,0.0008611196,0.8062111,0.000108364904],"about_ca_topic_score_codex":0.06853257,"about_ca_topic_score_gemma":0.16260955,"teacher_disagreement_score":0.06853257,"about_ca_system_score_codex":0.0010873972,"about_ca_system_score_gemma":0.001041763,"threshold_uncertainty_score":0.13626736},"labels":[],"label_agreement":null},{"id":"W3009357978","doi":"10.1038/s41597-019-0320-2","title":"A reference library for Canadian invertebrates with 1.5 million barcodes, voucher specimens, and DNA samples","year":2019,"lang":"en","type":"article","venue":"Scientific Data","topic":"Environmental DNA in Biodiversity Studies","field":"Environmental Science","cited_by":110,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Dalhousie University; Public Health Agency of Canada; University of Guelph","funders":"Nature Conservancy of Canada; Canada First Research Excellence Fund; Ontario Genomics; Ontario Ministry of Research, Innovation and Science; Government of Canada; Churchill Northern Studies Centre; University of Guelph; Ministry of Environment; Canada Foundation for Innovation; Parks Canada; Genome Canada; Gordon and Betty Moore Foundation; Natural Sciences and Engineering Research Council of Canada; Smithsonian Institution","keywords":"GenBank; Biodiversity; Barcode; Biology; DNA barcoding; Taxonomic rank; Global biodiversity; Taxonomy (biology); Environmental DNA; DNA sequencing; Zoology; Ecology; Geography; Taxon; DNA; Genetics; Computer science","score_opus":0.03989836595563699,"score_gpt":0.2084597603429137,"score_spread":0.1685613943872767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3009357978","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028809248,0.006973092,0.13882717,0.0016207799,0.0009989478,0.0026345802,0.7032376,0.011552579,0.10534595],"genre_scores_gemma":[0.023559215,0.0037196954,0.2260142,0.0008458187,0.0001407905,0.0013173443,0.69185644,0.0014982133,0.05104822],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9969146,0.0001501747,0.00019795347,0.00053302624,0.0018458078,0.00035829603],"domain_scores_gemma":[0.99321485,0.00030243464,0.0004469984,0.0008363758,0.0046707103,0.0005286897],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002433097,0.0015048247,0.0014839387,0.019469691,0.00617836,0.0024597861,0.004215548,0.0014406521,0.038564954],"category_scores_gemma":[0.005368536,0.0008908445,0.000997924,0.021358369,0.0009849215,0.0010224252,0.0020945529,0.0015772441,0.01825126],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006398771,0.00028077036,0.011675232,0.0035277316,0.00021362184,0.00090876943,0.0014555829,0.0013500118,0.117583156,0.012115442,0.37024307,0.48000678],"study_design_scores_gemma":[0.00004992848,0.00007265154,0.02566871,0.0005698416,0.00016640367,0.00051476824,0.000260611,0.0007453895,0.01233076,0.0010416077,0.95848763,0.00009163289],"about_ca_topic_score_codex":0.663842,"about_ca_topic_score_gemma":0.8443224,"teacher_disagreement_score":0.33615798,"about_ca_system_score_codex":0.010048265,"about_ca_system_score_gemma":0.04425985,"threshold_uncertainty_score":0.6762754},"labels":[],"label_agreement":null},{"id":"W3009988161","doi":"10.1038/s41597-020-0422-x","title":"A survey of invasive plants on grassland soil microbial communities and ecosystem services","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant and fungal interactions","field":"Agricultural and Biological Sciences","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Saskatchewan","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Ecosystem; Ecosystem services; Bromus inermis; Environmental science; Ecology; Ecosystem health; Grassland; Agroforestry; Forage; Biology","score_opus":0.13996465667780603,"score_gpt":0.24953217304742834,"score_spread":0.10956751636962231,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3009988161","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9993462,0.00003966289,0.000049534523,0.000006165397,2.7002892e-7,0.00000609032,0.00035717426,0.0000017513394,0.00019312194],"genre_scores_gemma":[0.997981,0.00013377912,0.00056104816,0.000019315878,0.0000010996062,0.000012560909,0.00096445065,9.676091e-7,0.00032572396],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998079,0.000040395957,0.000010704109,0.00004915276,0.000049799786,0.000041984484],"domain_scores_gemma":[0.99918956,0.000100280544,0.00030307728,0.0000562745,0.0001758881,0.00017490321],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004289265,0.00017568475,0.00014274921,0.00062446995,0.00041322334,0.00029345957,0.000188108,0.0002111732,0.00037602856],"category_scores_gemma":[0.00050175434,0.000115075934,0.000100294135,0.0007957155,0.00027876356,0.00017985214,0.00034115533,0.00019494562,0.00009682893],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009482011,0.000045848905,0.96915084,0.000027028651,0.000029049546,0.000077750774,0.000719345,0.000105562496,0.024231756,0.000021497968,0.000093407405,0.0054032695],"study_design_scores_gemma":[9.271314e-7,0.00003578854,0.9991535,0.0000020983443,0.0000031388417,0.000026783888,0.00028190814,0.000050799754,0.00030178667,0.0000037858013,0.00013781361,0.000001674737],"about_ca_topic_score_codex":0.05690285,"about_ca_topic_score_gemma":0.19229507,"teacher_disagreement_score":0.05690285,"about_ca_system_score_codex":0.00072175433,"about_ca_system_score_gemma":0.0006689089,"threshold_uncertainty_score":0.113143265},"labels":[],"label_agreement":null},{"id":"W3010123800","doi":"10.1038/s41597-020-0420-z","title":"Author Correction: A global database for metacommunity ecology, integrating species, traits, environment and space","year":2020,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; Université de Montréal","funders":"","keywords":"Metacommunity; Ecology; Space (punctuation); Geography; Computer science; Biology; Sociology; Demography","score_opus":0.09680895019397591,"score_gpt":0.30215380738792247,"score_spread":0.20534485719394657,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3010123800","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00019861454,0.0004776373,0.0017150054,0.044523634,0.925294,0.000044749704,0.016483542,0.0015231783,0.009739681],"genre_scores_gemma":[0.011643397,0.0045099356,0.016986711,0.075321384,0.17520192,0.0005521184,0.05319466,0.0075821616,0.65500766],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99365646,0.00067330495,0.0011392331,0.0007483435,0.0033187438,0.0004638729],"domain_scores_gemma":[0.9453334,0.010702563,0.0022693863,0.004270187,0.03519716,0.0022272908],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005734165,0.0020255067,0.0018270435,0.0053750025,0.003731768,0.006403821,0.0039461977,0.004379805,0.1537523],"category_scores_gemma":[0.0702921,0.0012320669,0.0016611898,0.0050738663,0.0017093722,0.0033449999,0.0034351677,0.008438027,0.11210828],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001295509,0.000002870472,0.000040103943,0.000041585416,0.000003414877,0.000032964726,0.000009660976,0.000019133713,0.000016768969,0.00039244036,0.99705267,0.0023755033],"study_design_scores_gemma":[0.000023390201,0.00000711525,0.00040024144,0.00018180565,0.000014979497,0.00011342266,0.000047109705,0.00011894054,0.00014577752,0.001022453,0.9979055,0.000019278348],"about_ca_topic_score_codex":0.037000194,"about_ca_topic_score_gemma":0.045916606,"teacher_disagreement_score":0.1537523,"about_ca_system_score_codex":0.003704035,"about_ca_system_score_gemma":0.009338379,"threshold_uncertainty_score":0.5143527},"labels":[],"label_agreement":null},{"id":"W3010707172","doi":"10.1038/s41597-020-0435-5","title":"Multidisciplinary database of permeability of fault zones and surrounding protolith rocks at world-wide sites","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrocarbon exploration and reservoir analysis","field":"Engineering","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Hydrogeology; Geology; Geothermal gradient; Permeability (electromagnetism); Protolith; Groundwater; Petrology; Aquifer; Crust; Petroleum engineering; Geotechnical engineering; Geochemistry; Metamorphic rock; Geophysics","score_opus":0.06308179900604846,"score_gpt":0.2855893111818437,"score_spread":0.22250751217579526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3010707172","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006596249,0.008880559,0.0008056702,0.00010212837,0.00002451824,0.0000710913,0.98011464,0.00025810092,0.003146995],"genre_scores_gemma":[0.023966566,0.01240686,0.003394259,0.00010283876,0.00003564008,0.0005205731,0.95789486,0.000099381454,0.0015790174],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982665,0.00020214941,0.0005908452,0.00039255613,0.00043424498,0.00011373279],"domain_scores_gemma":[0.99514717,0.0015753157,0.0013778621,0.00045603423,0.001266551,0.00017708362],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012165797,0.0010232701,0.001644051,0.025482591,0.0004443361,0.0016505497,0.0014116485,0.0013140768,0.014037872],"category_scores_gemma":[0.005401472,0.0005100724,0.0015142852,0.033235434,0.00036312072,0.0016322285,0.0014315785,0.0004869363,0.0063311085],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011737284,0.00025702265,0.078729965,0.2068886,0.0029550968,0.0018166401,0.0015408327,0.0074699307,0.0074787876,0.0071738283,0.4829864,0.20152919],"study_design_scores_gemma":[0.00016986862,0.00009647582,0.14009538,0.010707327,0.0015284941,0.0008261885,0.0009515849,0.00090774294,0.0029408569,0.0018617702,0.8398017,0.00011261496],"about_ca_topic_score_codex":0.0101613095,"about_ca_topic_score_gemma":0.018397141,"teacher_disagreement_score":0.025482591,"about_ca_system_score_codex":0.000791954,"about_ca_system_score_gemma":0.0026346107,"threshold_uncertainty_score":0.046961367},"labels":[],"label_agreement":null},{"id":"W3011355476","doi":"10.1038/s41597-020-0428-4","title":"Genome assembly of six polyploid potato genomes","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Pathogens and Resistance","field":"Agricultural and Biological Sciences","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Agriculture and Agri-Food Canada; McGill University","funders":"Agriculture and Agri-Food Canada; Consortium of International Agricultural Research Centers; Compute Canada; Natural Sciences and Engineering Research Council of Canada; Centre Sève; McGill University","keywords":"Polyploid; Genome; Biology; Genetics; Genome size; DNA sequencing; Evolutionary biology; Gene","score_opus":0.057538862461047204,"score_gpt":0.2333634025700783,"score_spread":0.1758245401090311,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3011355476","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.824552,0.0022450835,0.08471623,0.00059152243,0.00017684886,0.00067345466,0.07713921,0.0028208694,0.007084756],"genre_scores_gemma":[0.42669922,0.002025498,0.27464226,0.00021181254,0.000028780854,0.00049451843,0.2890959,0.000998566,0.005803562],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9997365,0.000030120042,0.00003513788,0.000116788084,0.00005172979,0.000029718416],"domain_scores_gemma":[0.99960023,0.00007068083,0.00006117196,0.00007676824,0.000112836206,0.000078402365],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004006846,0.00046702966,0.0005409992,0.0009331658,0.0006856527,0.0008322658,0.0005281941,0.0005652274,0.0026043458],"category_scores_gemma":[0.0009342533,0.00044488983,0.0008999538,0.0012389241,0.00016454529,0.00046860738,0.00083603413,0.0011505816,0.001437967],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00096533494,0.00008698261,0.0059472024,0.0008358465,0.00010244903,0.0008410989,0.0011711484,0.003450637,0.94715285,0.0010704919,0.0020264096,0.03634958],"study_design_scores_gemma":[0.00040228022,0.0014822419,0.17318954,0.00047965156,0.0008411089,0.004492632,0.002044591,0.045696016,0.5122685,0.004313954,0.25445023,0.00033921527],"about_ca_topic_score_codex":0.0022214814,"about_ca_topic_score_gemma":0.0031032737,"teacher_disagreement_score":0.0026043458,"about_ca_system_score_codex":0.0005420824,"about_ca_system_score_gemma":0.00067893416,"threshold_uncertainty_score":0.008712351},"labels":[],"label_agreement":null},{"id":"W3011735246","doi":"10.1038/s41597-020-00655-3","title":"TILES-2018, a longitudinal physiologic and behavioral data set of hospital workers","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Heart Rate Variability and Autonomic Control","field":"Medicine","cited_by":91,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institut National de la Recherche Scientifique; Université du Québec à Montréal","funders":"Intelligence Advanced Research Projects Activity; NIH Office of the Director; Office of the Director of National Intelligence","keywords":"Set (abstract data type); Longitudinal data; Data set; Gerontology; Medicine; Psychology; Computer science; Data mining; Artificial intelligence","score_opus":0.21975587364632193,"score_gpt":0.36081602114383465,"score_spread":0.14106014749751272,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3011735246","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.87279695,0.0006165512,0.004515731,0.00054875546,0.0001270801,0.00024335067,0.11903787,0.00025840293,0.0018553504],"genre_scores_gemma":[0.8118498,0.00047557865,0.007876168,0.0004534849,0.00020451417,0.00090869353,0.17589544,0.000067961104,0.0022683863],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99941266,0.000119776785,0.00008046312,0.00018983864,0.00010048677,0.00009683396],"domain_scores_gemma":[0.99795365,0.00040247606,0.00034888415,0.0004239913,0.0004839492,0.00038703435],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000793455,0.00042100745,0.000548913,0.0011372312,0.00054488593,0.00070227013,0.0006070825,0.00092248863,0.0012058693],"category_scores_gemma":[0.0032233237,0.00016368869,0.0003770575,0.0011453343,0.00026646338,0.00042139288,0.0010101016,0.0007488546,0.0013836451],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012145906,0.0015910165,0.89551944,0.00057301397,0.00035823433,0.001195189,0.002110987,0.0015056963,0.0087322425,0.0004670499,0.037200443,0.04953213],"study_design_scores_gemma":[0.000058692644,0.00040726905,0.97815174,0.00009209166,0.00008122644,0.0007567315,0.0013443385,0.0017913149,0.0014780463,0.00037347703,0.015415805,0.000049102342],"about_ca_topic_score_codex":0.0073727635,"about_ca_topic_score_gemma":0.011641879,"teacher_disagreement_score":0.0073727635,"about_ca_system_score_codex":0.00032739044,"about_ca_system_score_gemma":0.0006617725,"threshold_uncertainty_score":0.014659703},"labels":[],"label_agreement":null},{"id":"W3013353379","doi":"10.1038/s41597-020-0446-2","title":"A global ensemble of ocean wave climate projections from CMIP5-driven models","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ocean Waves and Remote Sensing","field":"Earth and Planetary Sciences","cited_by":96,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"Lawrence Berkeley National Laboratory; Biological and Environmental Research; Japan Society for the Promotion of Science; Office of Science; Ministry of Education, Culture, Sports, Science and Technology; U.S. Department of Energy; Commonwealth Scientific and Industrial Research Organisation; National Energy Research Scientific Computing Center","keywords":"Downscaling; Climatology; Environmental science; Coupled model intercomparison project; Metadata; Climate model; Climate change; Significant wave height; Wind wave model; Wave model; Scale (ratio); Meteorology; Wind wave; Wind speed; Computer science; Geography; Geology; Cartography; Precipitation","score_opus":0.10439511858755074,"score_gpt":0.2496939400140287,"score_spread":0.14529882142647796,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3013353379","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17163333,0.00030060761,0.009614707,0.0003874664,0.0002329982,0.00010864832,0.81040365,0.0022609415,0.0050577456],"genre_scores_gemma":[0.13462058,0.00019970127,0.010440778,0.00007750806,0.000045418263,0.000259093,0.8532917,0.00023259838,0.0008326261],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99968326,0.000051498057,0.000030295,0.000110748784,0.00008085951,0.000043313634],"domain_scores_gemma":[0.9991198,0.00014069946,0.000081540245,0.00024164324,0.00032895713,0.00008729249],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009575849,0.00097223726,0.00066676486,0.0010552995,0.00032521563,0.0008261124,0.0014506999,0.0008113837,0.0041616885],"category_scores_gemma":[0.0018202229,0.00041066474,0.0009483086,0.003348711,0.00019361428,0.000941377,0.00071938214,0.0011485312,0.0024362437],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005087579,0.0002832582,0.061186,0.0005910256,0.0013258808,0.00030496911,0.00014571418,0.6202612,0.0032468643,0.004064681,0.25366345,0.054418292],"study_design_scores_gemma":[0.00070281915,0.00016294977,0.14928606,0.00023650653,0.00037362662,0.00015818053,0.00024134468,0.6698828,0.0065163565,0.004849406,0.16730087,0.0002890896],"about_ca_topic_score_codex":0.03288445,"about_ca_topic_score_gemma":0.029086567,"teacher_disagreement_score":0.03288445,"about_ca_system_score_codex":0.00066496106,"about_ca_system_score_gemma":0.0015318701,"threshold_uncertainty_score":0.06538606},"labels":[],"label_agreement":null},{"id":"W3013517662","doi":"10.1038/s41597-020-0445-3","title":"A global database of Holocene paleotemperature records","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":300,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Okanagan University College; University of British Columbia, Okanagan Campus; Bishop's University; Okanagan College; Geological Survey of Canada; University of Ottawa; University of New Brunswick; Center for Northern Studies; Concordia University; Université du Québec à Montréal; University of British Columbia; Fisheries and Oceans Canada; Mount Royal University; Université Laval; Natural Resources Canada","funders":"National Oceanic and Atmospheric Administration; Sight Research UK; Climate Program Office; Heising-Simons Foundation; Past Global Changes; Natural Environment Research Council; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Holocene; Paleoclimatology; Proxy (statistics); Peat; Database; Geology; Context (archaeology); Physical geography; Climate change; Northern Hemisphere; Climatology; Oceanography; Geography; Archaeology; Paleontology","score_opus":0.06975029518787,"score_gpt":0.29128594666734886,"score_spread":0.22153565147947885,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3013517662","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032679856,0.0015819636,0.0025159917,0.00012276955,0.00004117739,0.000068033405,0.95680076,0.00045768474,0.0057317223],"genre_scores_gemma":[0.033110965,0.0016279094,0.00768666,0.00006583627,0.00004105264,0.00023983102,0.9556622,0.00011347252,0.0014519864],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99932754,0.00005547748,0.00018907558,0.00021230117,0.0001572455,0.00005837909],"domain_scores_gemma":[0.99566627,0.00071966683,0.0011826438,0.0008663521,0.001154326,0.00041076163],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009525679,0.0006664441,0.0007729141,0.0117003955,0.0003838088,0.0014277243,0.0010968434,0.0004183673,0.010614906],"category_scores_gemma":[0.0037029898,0.00040895515,0.00035260848,0.024006823,0.00019138733,0.0013877479,0.0011995621,0.00048032834,0.004684452],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000493195,0.00020739537,0.39432833,0.003699545,0.0006973925,0.0004912535,0.0011956996,0.0037497962,0.008789846,0.0049124765,0.18986414,0.39157096],"study_design_scores_gemma":[0.00004416872,0.00006135559,0.6048981,0.0004157429,0.00019547922,0.00043898006,0.00050043775,0.0006995624,0.0022990166,0.001024316,0.3893653,0.000057529844],"about_ca_topic_score_codex":0.0119018145,"about_ca_topic_score_gemma":0.013189572,"teacher_disagreement_score":0.0119018145,"about_ca_system_score_codex":0.00048477354,"about_ca_system_score_gemma":0.0020857132,"threshold_uncertainty_score":0.03551036},"labels":[],"label_agreement":null},{"id":"W3017300021","doi":"10.1038/s41597-020-0465-z","title":"FiCli, the Fish and Climate Change Database, informs climate adaptation and management for freshwater fishes","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fish Ecology and Management Studies","field":"Environmental Science","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Natural Resources and Forestry","funders":"","keywords":"Climate change; Adaptation (eye); Environmental resource management; Geography; Climate change adaptation; Fish <Actinopterygii>; Effects of global warming; Database; Freshwater fish; Ecology; Global warming; Fishery; Environmental science; Biology; Computer science","score_opus":0.0835278061003806,"score_gpt":0.25542169910849427,"score_spread":0.17189389300811367,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3017300021","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017228877,0.0032358214,0.0015616661,0.00061523035,0.00008933639,0.00022954831,0.97621983,0.002034596,0.014291048],"genre_scores_gemma":[0.011114075,0.005780868,0.01688475,0.0008377203,0.0001308629,0.0016146788,0.95914,0.0008456972,0.0036514127],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980197,0.00026367532,0.0007321459,0.0002834197,0.0005567557,0.000144315],"domain_scores_gemma":[0.984047,0.007670799,0.0026276973,0.0015693629,0.0032193775,0.0008657313],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040292945,0.0012389802,0.0015331774,0.024457846,0.00085641583,0.0026837168,0.0024193125,0.001477409,0.049194917],"category_scores_gemma":[0.02406373,0.0006329402,0.000975015,0.024411445,0.0004624953,0.0043047844,0.003477971,0.0008472137,0.01668433],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024526345,0.000029978131,0.009167393,0.015461914,0.00025963128,0.000162849,0.0007101632,0.00085914717,0.00075549993,0.0065094386,0.8596883,0.10615036],"study_design_scores_gemma":[0.00006602282,0.000021382797,0.016911177,0.0023508754,0.00013945848,0.000099852856,0.00018481133,0.00030162785,0.00032338352,0.0019250044,0.9775873,0.00008909839],"about_ca_topic_score_codex":0.032596864,"about_ca_topic_score_gemma":0.055570807,"teacher_disagreement_score":0.049194917,"about_ca_system_score_codex":0.0019833783,"about_ca_system_score_gemma":0.006917042,"threshold_uncertainty_score":0.16457337},"labels":[],"label_agreement":null},{"id":"W3018678682","doi":"10.1038/s41597-020-0467-x","title":"Simultaneous human intracerebral stimulation and HD-EEG, ground-truth for source localization methods","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced MRI Techniques and Applications","field":"Medicine","cited_by":74,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Sleep & Circadian Network; Université de Montréal; Hôpital du Sacré-Cœur de Montréal","funders":"Horizon 2020 Framework Programme; European Commission; Ministero della Salute; Università degli Studi di Milano","keywords":"Electroencephalography; Ground truth; Computer science; Epilepsy; Artificial intelligence; Clinical Practice; Ranging; Range (aeronautics); Sensitivity (control systems); Human brain; Pattern recognition (psychology); Neuroscience; Medicine; Psychology; Materials science","score_opus":0.10995956128060774,"score_gpt":0.43888664130918353,"score_spread":0.3289270800285758,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3018678682","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23970696,0.008761759,0.55669004,0.0010023158,0.00097311055,0.00089275086,0.16112557,0.016633166,0.014214372],"genre_scores_gemma":[0.48012844,0.0018531192,0.19365752,0.00032875294,0.0002711688,0.0009912722,0.31792665,0.00053594937,0.0043072486],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99885595,0.0002388886,0.00012120876,0.0003751318,0.00031308888,0.000095797826],"domain_scores_gemma":[0.9987716,0.00032508106,0.0001398111,0.00046938207,0.00024841275,0.000045765828],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010396244,0.0013878722,0.0007152655,0.0015928562,0.00036842312,0.001019634,0.0011787972,0.001197181,0.0038723107],"category_scores_gemma":[0.0044568367,0.00027472564,0.0006387974,0.0014389566,0.0006946139,0.0007170693,0.0011869073,0.0007421096,0.0029439828],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025377132,0.0008838794,0.02086013,0.0035792117,0.00069014163,0.0011904149,0.00033616574,0.064744346,0.068208314,0.005262711,0.13676527,0.6949417],"study_design_scores_gemma":[0.0008472987,0.0016495634,0.13747393,0.000986532,0.0005432184,0.0069075716,0.00086390966,0.37446702,0.17062552,0.035884637,0.26944265,0.00030816192],"about_ca_topic_score_codex":0.0028160387,"about_ca_topic_score_gemma":0.005762003,"teacher_disagreement_score":0.0038723107,"about_ca_system_score_codex":0.00031457498,"about_ca_system_score_gemma":0.00078554545,"threshold_uncertainty_score":0.0129541755},"labels":[],"label_agreement":null},{"id":"W3021219695","doi":"10.1038/s41597-020-0477-8","title":"COVID-19 Disease Map, building a computational repository of SARS-CoV-2 virus-host interaction mechanisms","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":153,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"National Institute of General Medical Sciences; Agence Nationale de la Recherche","keywords":"Pandemic; Coronavirus disease 2019 (COVID-19); Host (biology); Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Virology; 2019-20 coronavirus outbreak; Virus; Biology; Disease; Computational biology; Outbreak; Infectious disease (medical specialty); Medicine; Genetics","score_opus":0.14070240531308462,"score_gpt":0.3903484463825193,"score_spread":0.2496460410694347,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3021219695","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.061002567,0.0036815486,0.10135863,0.0027908797,0.00080482115,0.0006745555,0.7117009,0.099886246,0.018099869],"genre_scores_gemma":[0.12015545,0.0023508943,0.1134996,0.00053380063,0.00012030272,0.0008448099,0.75488573,0.0048983693,0.0027110637],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994918,0.00011688227,0.000059040118,0.00015661638,0.00012610322,0.000049472572],"domain_scores_gemma":[0.9988881,0.000499623,0.00007304201,0.0002504562,0.00014285828,0.0001459354],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014456034,0.001625336,0.0011070141,0.003533127,0.0009871668,0.0024141658,0.0030412248,0.0014126268,0.013706589],"category_scores_gemma":[0.006571057,0.0008372084,0.001961372,0.0027464817,0.0005418797,0.0017893431,0.0016583177,0.0016359299,0.004749492],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017021751,0.0007045896,0.026918605,0.004160829,0.0015965512,0.0014837144,0.0007898227,0.14804295,0.0088890735,0.03145965,0.69200003,0.08225206],"study_design_scores_gemma":[0.0013616785,0.00030786335,0.009796596,0.00051469373,0.00064844225,0.0009112212,0.0004853163,0.5208783,0.008514777,0.034407876,0.42189348,0.00027969375],"about_ca_topic_score_codex":0.015496335,"about_ca_topic_score_gemma":0.025763616,"teacher_disagreement_score":0.015496335,"about_ca_system_score_codex":0.0010008714,"about_ca_system_score_gemma":0.0034561995,"threshold_uncertainty_score":0.045853138},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ocean Networks Canada Society; University of Victoria","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute of Diabetes and Digestive and Kidney Diseases; Economic and Social Research Council","keywords":"World Wide Web; Computer science; Metadata; Scholarly communication; Data science; Internet privacy; Knowledge management; Publishing; Political science","score_opus":0.2872569636919766,"score_gpt":0.3757555810503705,"score_spread":0.08849861735839387,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3024722698","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014035875,0.004986748,0.73493654,0.05464263,0.000987072,0.0005095543,0.00053216994,0.00045864805,0.1889108],"genre_scores_gemma":[0.67685527,0.00660453,0.24168862,0.0075926455,0.002554784,0.00174643,0.00062455755,0.00047589268,0.061857287],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.96825856,0.012740256,0.0031459986,0.0034274762,0.010405793,0.00202186],"domain_scores_gemma":[0.9001599,0.049039,0.006541979,0.025877008,0.014824716,0.00355735],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.028046958,0.00093637337,0.0015784438,0.004178528,0.007923909,0.017969154,0.0039207893,0.009916371,0.008444413],"category_scores_gemma":[0.10084422,0.0015941104,0.0020431702,0.0050748847,0.021686148,0.042062435,0.012147124,0.009426426,0.0035309184],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011587728,0.000007573176,0.00011217675,0.000044743723,0.000012463287,0.00008289494,0.00055837986,0.0005705766,0.00004801716,0.99206793,0.0023059866,0.0041777277],"study_design_scores_gemma":[0.000026045023,0.000011499163,0.000075587275,0.000073034775,0.000016063605,0.00017814356,0.00016065604,0.0019779706,0.00011600847,0.9675925,0.029752064,0.000020525844],"about_ca_topic_score_codex":0.008645134,"about_ca_topic_score_gemma":0.0036322968,"teacher_disagreement_score":0.9960792,"about_ca_system_score_codex":0.0066887187,"about_ca_system_score_gemma":0.008105069,"threshold_uncertainty_score":0.14832824},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3026775718","doi":"10.1038/s41597-020-0489-4","title":"The Ontario Climate Data Portal, a user-friendly portal of Ontario-specific climate projections","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"York University","funders":"Natural Sciences and Engineering Research Council of Canada; Ministère de l’Environnement, de la Protection de la nature et des Parcs; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Climate change; User Friendly; Internet portal; Geography; Physical geography; Environmental science; Environmental resource management; Computer science; World Wide Web; Data science; Ecology; Biology; The Internet","score_opus":0.09271490599768245,"score_gpt":0.2789525976296391,"score_spread":0.18623769163195664,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3026775718","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002396401,0.00024937728,0.0069938623,0.0007046108,0.000099265555,0.00021861293,0.92716306,0.0077989027,0.054375783],"genre_scores_gemma":[0.026350968,0.0012101476,0.035544213,0.0002864661,0.00007163408,0.0007153431,0.8962601,0.004005808,0.0355554],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989555,0.00009586953,0.00008082761,0.00013015544,0.0005606766,0.00017699653],"domain_scores_gemma":[0.9941578,0.0007697872,0.00033468034,0.0009946265,0.0028885305,0.00085461425],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018719238,0.0006638963,0.000666763,0.0038766651,0.0018928797,0.0036078172,0.0014703681,0.0006346684,0.083383024],"category_scores_gemma":[0.007835015,0.00079022057,0.00048270612,0.013587518,0.0005277261,0.0023447769,0.0022328724,0.0009826359,0.024493236],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010685396,0.000020266405,0.006271612,0.0004325673,0.000030831132,0.00010142561,0.00050305173,0.0011471336,0.0005905188,0.0067932494,0.9535656,0.030436955],"study_design_scores_gemma":[0.000041592968,0.0000039770484,0.006486181,0.00007864669,0.000009314491,0.000029175388,0.00019436306,0.0011897259,0.00039072422,0.0012761699,0.9902587,0.00004150611],"about_ca_topic_score_codex":0.89103395,"about_ca_topic_score_gemma":0.92963874,"teacher_disagreement_score":0.10896605,"about_ca_system_score_codex":0.013132315,"about_ca_system_score_gemma":0.028544186,"threshold_uncertainty_score":0.27894402},"labels":[],"label_agreement":null},{"id":"W3028338699","doi":"10.1038/s41597-020-0488-5","title":"Global high-resolution emissions of soil NOx, sea salt aerosols, and biogenic volatile organic compounds","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric chemistry and aerosols","field":"Earth and Planetary Sciences","cited_by":186,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"National Natural Science Foundation of China","keywords":"Environmental science; NOx; Pollutant; Air quality index; Methane; Greenhouse gas; Pollution; Air pollution; Sea salt; Atmospheric sciences; Climate change; Environmental chemistry; Carbon cycle; Atmosphere (unit); Aerosol; Meteorology; Oceanography; Ecosystem; Chemistry; Ecology; Geography; Geology","score_opus":0.030461318930859977,"score_gpt":0.22887825537623427,"score_spread":0.19841693644537428,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3028338699","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9760685,0.00015019448,0.0027430386,0.0002147245,0.000049416783,0.000023118962,0.017450279,0.00070433115,0.0025963904],"genre_scores_gemma":[0.9671467,0.00011218743,0.005203287,0.000061012757,0.000022443797,0.000031687854,0.026718711,0.00012015252,0.00058382074],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.99986255,0.000016571867,0.000012362633,0.00004664248,0.000029880095,0.000031856554],"domain_scores_gemma":[0.999736,0.000048975075,0.000035256253,0.00006010231,0.00008744183,0.000032175907],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00040586098,0.0005738323,0.00037498772,0.0006298394,0.00026021173,0.0006332422,0.0004964962,0.0006248314,0.0019151517],"category_scores_gemma":[0.00063290104,0.00024529066,0.0009606787,0.0013197296,0.00027603182,0.0006453171,0.000515745,0.00055498455,0.0003549803],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024525163,0.00040037488,0.18359081,0.00014314859,0.00049240334,0.00022880989,0.00007123812,0.7737061,0.019399863,0.0013423174,0.0051754457,0.015204277],"study_design_scores_gemma":[0.00030513346,0.00007384061,0.27743304,0.00004490403,0.00015920638,0.00007622715,0.00021783,0.6929042,0.017990012,0.0017554605,0.008922436,0.00011782053],"about_ca_topic_score_codex":0.05895097,"about_ca_topic_score_gemma":0.038826387,"teacher_disagreement_score":0.05895097,"about_ca_system_score_codex":0.00056453614,"about_ca_system_score_gemma":0.0007590562,"threshold_uncertainty_score":0.11721563},"labels":[],"label_agreement":null},{"id":"W3033693292","doi":"10.1038/s41597-020-00712-x","title":"The Building Data Genome Project 2, energy meter data from the ASHRAE Great Energy Predictor III competition","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Building Energy and Comfort Optimization","field":"Engineering","cited_by":223,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"Ministry of Education, India; National Research Foundation","keywords":"ASHRAE 90.1; Benchmarking; Data set; Range (aeronautics); Energy (signal processing); Competition (biology); Metre; Measure (data warehouse)","score_opus":0.07491076497867988,"score_gpt":0.2493241337972274,"score_spread":0.17441336881854752,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3033693292","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0058087753,0.00021454605,0.0012001309,0.00024693392,0.000078404635,0.0000357936,0.9891956,0.0011858493,0.0020340434],"genre_scores_gemma":[0.0034918953,0.000045935143,0.0016730785,0.00004804244,0.000007653616,0.000048517784,0.99409556,0.00011308535,0.00047615683],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976132,0.00046732908,0.0001394267,0.00061833335,0.00091863406,0.0002431979],"domain_scores_gemma":[0.9962333,0.00091553584,0.0003300155,0.0007799336,0.0013188036,0.00042233965],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022164362,0.0013474873,0.0010214229,0.0027592042,0.00081145525,0.001777784,0.0017253984,0.0016873128,0.010311234],"category_scores_gemma":[0.008970054,0.00045735974,0.0009074597,0.0068144994,0.0004286232,0.00090150815,0.001976437,0.0016797292,0.013060412],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004536909,0.00024754694,0.023926765,0.00092577835,0.0001957873,0.00016145853,0.0001445374,0.005192666,0.0017892901,0.0023503227,0.940915,0.023697136],"study_design_scores_gemma":[0.00041872624,0.00019024781,0.09207874,0.00033263574,0.00011208949,0.0002664194,0.0004045693,0.010940652,0.0035392367,0.0039189644,0.88767916,0.00011862473],"about_ca_topic_score_codex":0.032734837,"about_ca_topic_score_gemma":0.054666854,"teacher_disagreement_score":0.032734837,"about_ca_system_score_codex":0.0010203546,"about_ca_system_score_gemma":0.0025789482,"threshold_uncertainty_score":0.06508863},"labels":[],"label_agreement":null},{"id":"W3035205089","doi":"10.1038/s41597-020-00611-1","title":"Publisher Correction: A global database of Holocene paleotemperature records","year":2020,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":63,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Okanagan University College; University of British Columbia, Okanagan Campus; Bishop's University; Okanagan College; Geological Survey of Canada; University of Ottawa; University of New Brunswick; Center for Northern Studies; Concordia University; Université du Québec à Montréal; University of British Columbia; Fisheries and Oceans Canada; Mount Royal University; Université Laval; Natural Resources Canada","funders":"","keywords":"Holocene; Database; Information retrieval; Geology; Computer science; Oceanography","score_opus":0.04864310150019533,"score_gpt":0.282948224771884,"score_spread":0.2343051232716887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3035205089","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00036733327,0.00086761377,0.00092718704,0.02976696,0.9244474,0.00005216909,0.025613226,0.00088220666,0.017075831],"genre_scores_gemma":[0.008894466,0.0057203867,0.008369798,0.03532883,0.1164309,0.0003721505,0.064524464,0.0039741504,0.7563849],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99659026,0.0002874994,0.0006370605,0.000382161,0.001881256,0.00022178698],"domain_scores_gemma":[0.9647996,0.005294263,0.0014266884,0.0030942715,0.024008892,0.0013763616],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032949317,0.0016240621,0.0013445595,0.0063191685,0.002674593,0.0054439926,0.002610643,0.002431823,0.14031246],"category_scores_gemma":[0.039095648,0.0008798858,0.0010525399,0.006324608,0.001018598,0.002571197,0.0020399643,0.005636609,0.09670686],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010932885,0.0000032349706,0.000060180148,0.00004353031,0.0000026546215,0.000043781154,0.0000081379285,0.000018433635,0.000018513372,0.00030125535,0.9958509,0.0036383914],"study_design_scores_gemma":[0.000016206539,0.0000066336474,0.00088010303,0.00017553865,0.00001192543,0.00009646669,0.00004585925,0.00006372271,0.00013106626,0.00048283854,0.9980749,0.000014846067],"about_ca_topic_score_codex":0.03681057,"about_ca_topic_score_gemma":0.053026076,"teacher_disagreement_score":0.14031246,"about_ca_system_score_codex":0.002612527,"about_ca_system_score_gemma":0.0068860534,"threshold_uncertainty_score":0.46939194},"labels":[],"label_agreement":null},{"id":"W3036205451","doi":"10.1038/s41597-020-0533-4","title":"Massive online data annotation, crowdsourcing to generate high quality sleep spindle annotations from EEG data","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Obstructive Sleep Apnea Research","field":"Medicine","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Canadian Sleep & Circadian Network","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Crowdsourcing; Annotation; Computer science; Electroencephalography; Data quality; Quality (philosophy); Metadata; Information retrieval; Data science; World Wide Web; Artificial intelligence; Biology; Neuroscience","score_opus":0.2777290831394593,"score_gpt":0.4268230892464325,"score_spread":0.14909400610697315,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3036205451","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15156078,0.0022164043,0.61996853,0.0039371606,0.0024590914,0.0052527697,0.14416733,0.045133032,0.025304822],"genre_scores_gemma":[0.33168986,0.00069994695,0.47534204,0.001173676,0.00059441855,0.006101859,0.17303942,0.003085758,0.00827298],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9924899,0.0025746624,0.0006088422,0.0020600283,0.0019605518,0.00030596906],"domain_scores_gemma":[0.9784693,0.008882377,0.0015928325,0.005813462,0.0042920825,0.0009499339],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007990924,0.0017366376,0.0011091204,0.0062704994,0.0018502902,0.001975982,0.0024496336,0.0014841654,0.004205951],"category_scores_gemma":[0.027745696,0.00048039222,0.0013723263,0.0037080739,0.0010842093,0.0015290966,0.006710973,0.0014752356,0.0036976857],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021864718,0.0013046828,0.05350225,0.0076359343,0.0011441569,0.002488826,0.0068391073,0.04085959,0.07273247,0.008663472,0.25277266,0.54987043],"study_design_scores_gemma":[0.000747872,0.0007178314,0.12851843,0.0013971967,0.0004108164,0.0010665267,0.005977207,0.322932,0.046271306,0.06473835,0.42652595,0.0006965301],"about_ca_topic_score_codex":0.010339773,"about_ca_topic_score_gemma":0.020229105,"teacher_disagreement_score":0.010339773,"about_ca_system_score_codex":0.0011888439,"about_ca_system_score_gemma":0.0024710407,"threshold_uncertainty_score":0.042260587},"labels":[],"label_agreement":null},{"id":"W3036577737","doi":"10.1038/s41597-020-0529-0","title":"Plant traits and vegetation data from climate warming experiments along an 1100 m elevation gradient in Gongga Mountains, China","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Chinese Academy of Sciences; National Natural Science Foundation of China; University of Arizona","keywords":"Elevation (ballistics); China; Vegetation (pathology); Physical geography; Environmental science; Global warming; Climate change; Ecology; Geology; Geography; Biology","score_opus":0.1121294371311376,"score_gpt":0.3025631863235091,"score_spread":0.1904337491923715,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3036577737","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9985868,0.000016027378,0.00005971722,0.0000049875025,7.875534e-7,0.000008646905,0.0012022192,0.000005750377,0.000115161965],"genre_scores_gemma":[0.99287224,0.000029989931,0.0003393285,0.000012309228,0.000002459133,0.000053142332,0.0065066624,0.0000034141092,0.00018040936],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99977404,0.000048954338,0.000025119058,0.00007871945,0.000038330934,0.000034754736],"domain_scores_gemma":[0.9995332,0.000065005566,0.00010723106,0.00012285955,0.000067584915,0.00010421945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00049244723,0.0002490228,0.00028508547,0.0010651209,0.00058276503,0.00022288297,0.0003036967,0.00015030848,0.00055151765],"category_scores_gemma":[0.0003161628,0.00013871561,0.0002545092,0.001485539,0.00041536806,0.0001631262,0.00028384113,0.00017004095,0.000096499956],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022791608,0.00013461304,0.9761169,0.000050285103,0.00009276389,0.00013214946,0.0007558947,0.00068219646,0.01401714,0.00006108394,0.00044621166,0.0072828894],"study_design_scores_gemma":[0.000006042939,0.000015221241,0.9994562,8.4057314e-7,0.000007836239,0.000011136012,0.00005590684,0.0001105837,0.00014560942,0.0000057772268,0.00018265446,0.0000022027712],"about_ca_topic_score_codex":0.03633555,"about_ca_topic_score_gemma":0.09778533,"teacher_disagreement_score":0.03633555,"about_ca_system_score_codex":0.00054580125,"about_ca_system_score_gemma":0.0005790391,"threshold_uncertainty_score":0.0722481},"labels":[],"label_agreement":null},{"id":"W3036636131","doi":"10.1038/s41597-020-0522-7","title":"The Fire and Tree Mortality Database, for empirical modeling of individual tree mortality after fire","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Sport Centre Pacific; Natural Resources Canada; Canadian Forest Service","funders":"U.S. Forest Service; Joint Fire Science Program; Rocky Mountain Research Station; U.S. Department of Agriculture","keywords":"Woodland; Geography; Tree (set theory); Ecology; Fire regime; Range (aeronautics); Database; Environmental resource management; Environmental science; Biology; Computer science; Ecosystem; Mathematics","score_opus":0.11759719441126627,"score_gpt":0.3233541803086803,"score_spread":0.20575698589741404,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3036636131","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029664269,0.000579685,0.019233666,0.00030727615,0.000086608256,0.00035751078,0.94161165,0.0038380558,0.0043213456],"genre_scores_gemma":[0.048252095,0.00040945507,0.025739342,0.00012048706,0.000040788225,0.0007929965,0.9222995,0.0003074856,0.0020378851],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99894434,0.00019744455,0.0002699169,0.00022236812,0.00029648785,0.00006942347],"domain_scores_gemma":[0.99542564,0.0016177094,0.0008141866,0.0010037217,0.00085191947,0.0002868905],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029577212,0.0011301409,0.0008701286,0.002532807,0.00039924108,0.0008993569,0.0025046184,0.0011017554,0.008670992],"category_scores_gemma":[0.010434982,0.0005728209,0.0009766481,0.0030899711,0.00016698793,0.001351274,0.00087623176,0.0011480565,0.006319728],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010142988,0.0009735996,0.19462988,0.0012417663,0.0007179342,0.00037528662,0.00028582476,0.06942168,0.0013416153,0.0060123755,0.63458407,0.08940174],"study_design_scores_gemma":[0.00085971097,0.00046197558,0.31911784,0.00060974876,0.0003438719,0.00086169085,0.0004371601,0.30576536,0.003078476,0.015005799,0.3532242,0.00023419676],"about_ca_topic_score_codex":0.027674692,"about_ca_topic_score_gemma":0.03232416,"teacher_disagreement_score":0.027674692,"about_ca_system_score_codex":0.0007532684,"about_ca_system_score_gemma":0.0015289001,"threshold_uncertainty_score":0.055027246},"labels":[],"label_agreement":null},{"id":"W3037249913","doi":"10.1038/s41597-020-0551-2","title":"The BenBioDen database, a global database for meio-, macro- and megabenthic biomass and densities","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine Biology and Ecology Research","field":"Earth and Planetary Sciences","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bedford Institute of Oceanography; Fisheries and Oceans Canada; Memorial University of Newfoundland; Université Laval; ArcticNet; St. Francis Xavier University; Université du Québec à Chicoutimi","funders":"Executive Agency for Small and Medium-sized Enterprises; Universität Rostock; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; European Commission; ArcticNet","keywords":"Benthic zone; Fauna; Biomass (ecology); Database; Environmental science; Bioturbation; Oceanography; Ecosystem; Benthos; Ecology; Macrobenthos; Abundance (ecology); Biology; Geology; Computer science; Sediment; Paleontology","score_opus":0.08922836685675936,"score_gpt":0.297443035589597,"score_spread":0.20821466873283764,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3037249913","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0039112065,0.006228788,0.0015578669,0.00010960435,0.00003912676,0.000112675036,0.9832896,0.0005676614,0.0041834717],"genre_scores_gemma":[0.0132876225,0.004590426,0.009483878,0.00012986246,0.000044090033,0.00086484326,0.9694728,0.00026253113,0.0018639329],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982482,0.00024398592,0.00071526115,0.00031805137,0.000391675,0.00008278513],"domain_scores_gemma":[0.99106723,0.003259247,0.0020951447,0.0007179978,0.002379135,0.00048114784],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022367216,0.0011171859,0.0018655875,0.026894202,0.0004679113,0.0018524517,0.0016843781,0.000819886,0.023779009],"category_scores_gemma":[0.009189755,0.00058884727,0.0006680596,0.025054565,0.00022724587,0.0016737873,0.0018526124,0.00062574644,0.008599776],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008013853,0.00013483778,0.05006522,0.057645988,0.0013431037,0.0005340822,0.0013132453,0.0014200947,0.0042647985,0.009605598,0.68060625,0.19226535],"study_design_scores_gemma":[0.00012801167,0.000043858763,0.0641261,0.0058326605,0.0007255074,0.00025614852,0.0004347089,0.00071780145,0.0014048731,0.0018956078,0.92435014,0.00008468894],"about_ca_topic_score_codex":0.010146851,"about_ca_topic_score_gemma":0.016081268,"teacher_disagreement_score":0.026894202,"about_ca_system_score_codex":0.00075239106,"about_ca_system_score_gemma":0.0027251327,"threshold_uncertainty_score":0.07954872},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3039006806","doi":"10.1038/s41597-020-0554-z","title":"ERA5-based global meteorological wildfire danger maps","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":214,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Horizon 2020 Framework Programme","keywords":"Environmental science; Meteorology; Earth system science; Climate change; Precipitation; Proxy (statistics); Interim; Probabilistic logic; Climatology; Environmental resource management; Computer science; Geography; Geology","score_opus":0.039449350082295526,"score_gpt":0.2532188669693745,"score_spread":0.213769516887079,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3039006806","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03799237,0.0003383382,0.005231522,0.00016409956,0.00009589412,0.00010648183,0.9465274,0.0015123781,0.008031567],"genre_scores_gemma":[0.06633596,0.00035043125,0.010910879,0.0000543302,0.000036948957,0.00015211191,0.92006457,0.0001843692,0.0019103661],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999818,0.00001969372,0.000019613555,0.000040478975,0.00006436572,0.000037771882],"domain_scores_gemma":[0.9996325,0.000041551506,0.000052302134,0.00006557499,0.00015547089,0.000052500578],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00035884965,0.00058035355,0.0002799087,0.0017121906,0.00020316934,0.0005974601,0.00053714425,0.00039044433,0.008037289],"category_scores_gemma":[0.0009920672,0.0001718353,0.0005306959,0.0026753498,0.000094580966,0.0006998933,0.00053988444,0.00054141553,0.0036494965],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000506467,0.00025419056,0.076731496,0.0012492352,0.0003890503,0.0004754955,0.0004710639,0.15212336,0.005174982,0.008699114,0.58743817,0.16648735],"study_design_scores_gemma":[0.00025292084,0.00006957851,0.22748268,0.00034119008,0.00011208608,0.000203291,0.000442455,0.101129904,0.0037711808,0.005836584,0.66020656,0.00015157886],"about_ca_topic_score_codex":0.040037178,"about_ca_topic_score_gemma":0.039953344,"teacher_disagreement_score":0.040037178,"about_ca_system_score_codex":0.0004813445,"about_ca_system_score_gemma":0.00070305425,"threshold_uncertainty_score":0.07960826},"labels":[],"label_agreement":null},{"id":"W3040736388","doi":"10.1038/s41597-020-0563-y","title":"A database of human gait performance on irregular and uneven surfaces collected by wearable sensors","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Context-Aware Activity Recognition Systems","field":"Computer Science","cited_by":119,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Centre Hospitalier Universitaire Sainte-Justine","funders":"National Institute for Occupational Safety and Health; U.S. Department of Health and Human Services","keywords":"Wearable computer; Gait; Computer science; Database; Gait analysis; Wearable technology; Physical medicine and rehabilitation; Embedded system; Medicine","score_opus":0.07286879744542986,"score_gpt":0.2707043208546184,"score_spread":0.19783552340918853,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3040736388","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2018567,0.0019088172,0.012077196,0.00020763741,0.00030442627,0.0007334819,0.7728419,0.0025911655,0.007478701],"genre_scores_gemma":[0.18038967,0.0009448946,0.012450701,0.00014226837,0.00008502779,0.0011669422,0.8011295,0.00013263489,0.0035583582],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9995622,0.00004250825,0.000101214435,0.00012846495,0.00012522092,0.000040450555],"domain_scores_gemma":[0.9988288,0.00015641356,0.00016719862,0.00023933224,0.00047969384,0.00012848595],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00031871887,0.0011840003,0.0011144116,0.0027013868,0.00036364346,0.0006775354,0.001059202,0.0011472441,0.006435498],"category_scores_gemma":[0.0018780553,0.00024670566,0.0006314044,0.0028667743,0.00020005961,0.0005586551,0.00073819637,0.0004083597,0.0055848937],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029658645,0.0017392868,0.13152656,0.0060531623,0.0008629441,0.0020009193,0.00072221254,0.007265855,0.014084617,0.0011631629,0.45632014,0.37529525],"study_design_scores_gemma":[0.00046557107,0.001278032,0.78050965,0.0009902226,0.0004348662,0.0042696665,0.0012894347,0.021042965,0.0076953196,0.0024498415,0.17927037,0.00030400333],"about_ca_topic_score_codex":0.0038192715,"about_ca_topic_score_gemma":0.011009333,"teacher_disagreement_score":0.006435498,"about_ca_system_score_codex":0.00023369823,"about_ca_system_score_gemma":0.00041201204,"threshold_uncertainty_score":0.02152884},"labels":[],"label_agreement":null},{"id":"W3040739689","doi":"10.1038/s41597-020-0534-3","title":"The FLUXNET2015 dataset and the ONEFlux processing pipeline for eddy covariance data","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Water Relations and Carbon Dynamics","field":"Environmental Science","cited_by":1766,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; University of Manitoba; Centre de Géomatique du Québec; University of Saskatchewan; Canadian Forest Service; Environment and Climate Change Canada; University of British Columbia; University of Waterloo; Global Institute for Water Security; Université Laval; Natural Resources Canada; Ministère des Ressources naturelles et des Forêts; Ontario Drive & Gear (Canada); McMaster University","funders":"Natural Environment Research Council; Sight Research UK","keywords":"Eddy covariance; Covariance; Pipeline (software); Computer science; Data mining; Statistics; Mathematics; Biology","score_opus":0.06542144998765222,"score_gpt":0.2774303750916456,"score_spread":0.21200892510399338,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3040739689","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018030179,0.00012562341,0.0015826443,0.0000836313,0.00007526749,0.000071570284,0.9926427,0.0023381033,0.0012774556],"genre_scores_gemma":[0.0018551988,0.000065054825,0.0021533677,0.000036695208,0.000016302822,0.0001879973,0.9950889,0.00024644655,0.0003500303],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99902105,0.00013384859,0.00019134967,0.00029161482,0.00023204637,0.00013013667],"domain_scores_gemma":[0.9988796,0.00023387071,0.00017127863,0.00028712812,0.00032524212,0.0001029487],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001555493,0.0023550796,0.0014625096,0.004605835,0.0006165845,0.0017722963,0.002230923,0.0018241615,0.018980775],"category_scores_gemma":[0.0037445861,0.00070565776,0.0015052819,0.006133919,0.0004025468,0.0019646194,0.0017682079,0.0015615998,0.02315832],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024562096,0.00017346551,0.011224647,0.0024216743,0.000244933,0.00020827816,0.0002342038,0.005565728,0.002398837,0.0037566426,0.9476436,0.025882479],"study_design_scores_gemma":[0.00026141907,0.000041563733,0.026523357,0.00045447584,0.00009847057,0.00018253995,0.00020275883,0.0073359213,0.0026731913,0.006077819,0.9559984,0.00015006581],"about_ca_topic_score_codex":0.016437009,"about_ca_topic_score_gemma":0.020130523,"teacher_disagreement_score":0.018980775,"about_ca_system_score_codex":0.001089525,"about_ca_system_score_gemma":0.0019501658,"threshold_uncertainty_score":0.06349701},"labels":[],"label_agreement":null},{"id":"W3041337723","doi":"10.1038/s41597-020-0520-9","title":"The International Bathymetric Chart of the Arctic Ocean Version 4.0","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geological Studies and Exploration","field":"Earth and Planetary Sciences","cited_by":311,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Hydrographic Service; Geological Survey of Canada","funders":"Stockholms Universitet; Natural Environment Research Council; Sight Research UK; Jet Propulsion Laboratory; National Aeronautics and Space Administration","keywords":"Bathymetric chart; Oceanography; Bathymetry; Chart; Arctic; The arctic; Geography; Environmental science; Geology; Mathematics; Statistics","score_opus":0.07080271622274797,"score_gpt":0.22058557326769745,"score_spread":0.14978285704494948,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3041337723","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030841867,0.0016656495,0.058360603,0.00066919444,0.0013515031,0.0008038623,0.7025667,0.006356692,0.197384],"genre_scores_gemma":[0.079235345,0.0024534848,0.10528091,0.000380176,0.00022158834,0.00092235365,0.778696,0.0021853212,0.030624777],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999111,0.00014352346,0.00012309905,0.00009795601,0.00042103787,0.0001032858],"domain_scores_gemma":[0.99845827,0.00010253597,0.00023234046,0.0002118935,0.0008984417,0.00009657989],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014237054,0.00071465026,0.00036529754,0.0028829582,0.0004528998,0.0011775907,0.0005404696,0.00029806985,0.013862736],"category_scores_gemma":[0.003523683,0.00027423026,0.0003632417,0.0054650432,0.00028801765,0.0008069055,0.0010201664,0.00081627636,0.008747269],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002454322,0.00004442724,0.041687623,0.0006953456,0.00007348561,0.00020115163,0.0008301028,0.004213542,0.00782748,0.017288586,0.6744179,0.25247502],"study_design_scores_gemma":[0.000019430525,0.000025251567,0.06056013,0.0001357056,0.000021026242,0.00009908447,0.0002737282,0.0011393813,0.0011027385,0.0014402352,0.93513864,0.000044654058],"about_ca_topic_score_codex":0.10194559,"about_ca_topic_score_gemma":0.09753112,"teacher_disagreement_score":0.10194559,"about_ca_system_score_codex":0.000880613,"about_ca_system_score_gemma":0.004344925,"threshold_uncertainty_score":0.20270437},"labels":[],"label_agreement":null},{"id":"W3043194960","doi":"10.1038/s41597-020-00583-2","title":"A comprehensive, multisource database for hydrometeorological modeling of 14,425 North American watersheds","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":119,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; École de Technologie Supérieure","funders":"","keywords":"Hydrometeorology; Environmental science; Watershed; Precipitation; Hydrology (agriculture); Snow; Hydrological modelling; Climatology; Scale (ratio); Rain gauge; Database; Meteorology; Geography; Geology; Cartography; Computer science","score_opus":0.0959559339871537,"score_gpt":0.2779758610765329,"score_spread":0.1820199270893792,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3043194960","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021729931,0.00032996017,0.009840495,0.00020806339,0.000030319168,0.0001207191,0.9532526,0.005302106,0.009185863],"genre_scores_gemma":[0.060760703,0.00055038993,0.016658366,0.000095546384,0.000030527975,0.00057573477,0.91802514,0.0004776884,0.0028258304],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99944526,0.00006656711,0.00010628003,0.00016189458,0.00018122862,0.000038762173],"domain_scores_gemma":[0.99840647,0.00027712277,0.00019836791,0.00036108724,0.00059860456,0.000158338],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00064499147,0.0008093459,0.0008351822,0.0032701355,0.0005572872,0.0016447917,0.0016006808,0.0006711697,0.016476154],"category_scores_gemma":[0.002245238,0.000544172,0.00058194273,0.008508258,0.0001819874,0.0016028587,0.00075751956,0.0005870479,0.00705449],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001358171,0.0003827656,0.09271629,0.0008110129,0.00030403936,0.00024984326,0.00053778366,0.07647162,0.001785087,0.008146904,0.7064549,0.112003975],"study_design_scores_gemma":[0.00018640743,0.000034275356,0.12958713,0.0002993293,0.00014929613,0.00014028203,0.0006381679,0.1285208,0.0021674118,0.008683094,0.72945535,0.00013853272],"about_ca_topic_score_codex":0.08324084,"about_ca_topic_score_gemma":0.08809076,"teacher_disagreement_score":0.08324084,"about_ca_system_score_codex":0.001426218,"about_ca_system_score_gemma":0.0036954999,"threshold_uncertainty_score":0.16551262},"labels":[],"label_agreement":null},{"id":"W3043520506","doi":"10.1038/s41597-020-00589-w","title":"Author Correction: COVID-19 Disease Map, building a computational repository of SARS-CoV-2 virus-host interaction mechanisms","year":2020,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Host (biology); Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); 2019-20 coronavirus outbreak; Computer science; Virology; Virus; Computational biology; Biology; Disease; Medicine; Infectious disease (medical specialty); Ecology; Outbreak","score_opus":0.04901385862372627,"score_gpt":0.3529667561876381,"score_spread":0.30395289756391186,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3043520506","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00019445566,0.00031473275,0.001731688,0.04909127,0.93838054,0.000033225526,0.004743776,0.0011996763,0.004310591],"genre_scores_gemma":[0.020468108,0.004732827,0.017012268,0.123210914,0.26398245,0.00040782033,0.018652761,0.010509653,0.5410232],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99496365,0.00069127214,0.00094402744,0.00081817125,0.0022816388,0.00030123934],"domain_scores_gemma":[0.95900327,0.010198649,0.0014713628,0.0029187673,0.024489727,0.0019182257],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045564617,0.001659583,0.0014812103,0.0037983584,0.0036167726,0.004707012,0.002985587,0.0046490403,0.11226249],"category_scores_gemma":[0.08257293,0.001006847,0.001345424,0.0030410876,0.0024270224,0.0025017504,0.0026478462,0.009243902,0.071659654],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016584034,0.000003040067,0.00003465103,0.000044794277,0.0000044963567,0.000093749186,0.000017506562,0.000027988535,0.000028703871,0.00046011244,0.99694294,0.002325371],"study_design_scores_gemma":[0.00003609296,0.000011183467,0.0003300037,0.00017472576,0.0000204872,0.00034381813,0.000075058655,0.00022178295,0.0003132175,0.0015623365,0.9968839,0.000027257156],"about_ca_topic_score_codex":0.021640144,"about_ca_topic_score_gemma":0.027934587,"teacher_disagreement_score":0.11226249,"about_ca_system_score_codex":0.0028166198,"about_ca_system_score_gemma":0.006592928,"threshold_uncertainty_score":0.37555546},"labels":[],"label_agreement":null},{"id":"W3043668475","doi":"10.1038/s41597-020-0557-9","title":"CerebrA, registration and manual label correction of Mindboggle-101 atlas for MNI-ICBM152 template","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":117,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"Fonds de Recherche du Québec - Santé","keywords":"Atlas (anatomy); Computer science; Preprocessor; Brain atlas; Magnetic resonance imaging; Artificial intelligence; Pattern recognition (psychology); Medicine; Radiology; Anatomy","score_opus":0.16964929103160334,"score_gpt":0.3270970872157216,"score_spread":0.15744779618411828,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3043668475","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.054116074,0.003437122,0.49694538,0.0012581766,0.0020843484,0.004144553,0.31572002,0.09520267,0.027091656],"genre_scores_gemma":[0.0810354,0.00091876223,0.473507,0.0007522351,0.00022990674,0.008683726,0.3918387,0.020341376,0.022692937],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9975789,0.0003451537,0.00037681236,0.0010150403,0.00050153874,0.00018254868],"domain_scores_gemma":[0.9977812,0.00032708832,0.00026404712,0.0009937594,0.00055358716,0.000080363956],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028853018,0.0023550077,0.001487715,0.004310327,0.0017192602,0.003236714,0.0029326696,0.0019580268,0.027202955],"category_scores_gemma":[0.009466619,0.0011894709,0.0017853947,0.0034315556,0.00094473205,0.0014743985,0.00274527,0.0024384428,0.024787772],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00091505254,0.00022208174,0.0075008087,0.001612612,0.0005016765,0.00063507,0.00082615466,0.00667863,0.02710909,0.010046198,0.60729885,0.3366537],"study_design_scores_gemma":[0.0003821406,0.00033462452,0.032967042,0.0005815008,0.00043638048,0.0051548863,0.00044011622,0.036842544,0.050796587,0.034333315,0.8373497,0.00038129045],"about_ca_topic_score_codex":0.024077507,"about_ca_topic_score_gemma":0.0483428,"teacher_disagreement_score":0.027202955,"about_ca_system_score_codex":0.0017402661,"about_ca_system_score_gemma":0.005219861,"threshold_uncertainty_score":0.091003},"labels":[],"label_agreement":null},{"id":"W3045777412","doi":"10.1038/s41597-020-00590-3","title":"Author Correction: Global karst springs hydrograph dataset for research and management of the world’s fastest-flowing groundwater","year":2020,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Karst Systems and Hydrogeology","field":"Earth and Planetary Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"Natural Environment Research Council; Sight Research UK","keywords":"Karst; Hydrograph; Groundwater; Hydrology (agriculture); Geology; Water resource management; Environmental science; Geography; Paleontology; Archaeology; Geotechnical engineering","score_opus":0.1316815406232029,"score_gpt":0.34645248710723464,"score_spread":0.21477094648403175,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3045777412","genre_codex":"editorial","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00064123195,0.00057606754,0.0025614102,0.037386853,0.5283678,0.00014173416,0.40410018,0.0025640852,0.023660578],"genre_scores_gemma":[0.014942662,0.0029740923,0.0139135625,0.027264621,0.04287737,0.0007940982,0.50597006,0.007964016,0.38329953],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9962787,0.00040978982,0.0006681146,0.0005335927,0.0018425873,0.0002673334],"domain_scores_gemma":[0.9614401,0.0060340343,0.0015294435,0.0037376096,0.026141042,0.0011178475],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037876861,0.0018013676,0.0012884695,0.005581714,0.001769022,0.0041468088,0.0026119917,0.0022050722,0.17574868],"category_scores_gemma":[0.053551722,0.00094054727,0.0013361253,0.0073510334,0.0010050151,0.0031690684,0.0029131216,0.004972752,0.113901295],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009819743,0.0000019716636,0.000071005066,0.00003158632,0.0000027428819,0.000013330165,0.0000056335807,0.00003294773,0.000009583993,0.00023740245,0.9981535,0.0014304277],"study_design_scores_gemma":[0.00003957099,0.0000057954594,0.0010448565,0.00018183603,0.000012127993,0.000044700184,0.000067738896,0.0001254485,0.00013676927,0.0011020664,0.99721694,0.000022181443],"about_ca_topic_score_codex":0.05486078,"about_ca_topic_score_gemma":0.064742215,"teacher_disagreement_score":0.17574868,"about_ca_system_score_codex":0.0026428392,"about_ca_system_score_gemma":0.007601958,"threshold_uncertainty_score":0.5879379},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3048219094","doi":"10.1038/s41597-020-00603-1","title":"A consistent Great Lakes ice cover digital data set for winters 1973–2019","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Arctic and Antarctic ice dynamics","field":"Earth and Planetary Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Oceanic and Atmospheric Administration; U.S. Department of Commerce","keywords":"Cover (algebra); Physical geography; Data set; Geology; Oceanography; Environmental science; Geography; Computer science; Engineering","score_opus":0.08784984032571287,"score_gpt":0.26159169881930355,"score_spread":0.1737418584935907,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3048219094","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052030466,0.00009456606,0.0007241444,0.00011395264,0.000043180204,0.00014894918,0.9420809,0.00038452898,0.0043792655],"genre_scores_gemma":[0.027948283,0.00005931354,0.0019904748,0.00002996342,0.00001709894,0.0003125096,0.9683498,0.000047670543,0.0012448746],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999585,0.00002759901,0.00007856538,0.00008452211,0.00017962986,0.000044815948],"domain_scores_gemma":[0.9988361,0.000077130484,0.00014330387,0.00019942236,0.0006326223,0.00011142851],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00053799455,0.00036151698,0.000307151,0.0033948196,0.0006927008,0.0005754026,0.0006478205,0.00027412933,0.004781109],"category_scores_gemma":[0.0016118515,0.0001788084,0.00032491967,0.005220559,0.00023239572,0.0004669442,0.0007068568,0.00030353921,0.0037051435],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00058599684,0.0002684225,0.2233801,0.0007224015,0.00023411267,0.0008848685,0.0010418169,0.0065506008,0.0066276942,0.0022079328,0.6726595,0.08483659],"study_design_scores_gemma":[0.00010789329,0.000047654696,0.59260994,0.000121159705,0.000051770632,0.00023753771,0.00061543094,0.004379725,0.003946509,0.00039835033,0.39742404,0.000059958154],"about_ca_topic_score_codex":0.111555554,"about_ca_topic_score_gemma":0.18409447,"teacher_disagreement_score":0.8884444,"about_ca_system_score_codex":0.0011349476,"about_ca_system_score_gemma":0.002271828,"threshold_uncertainty_score":0.22181243},"labels":[],"label_agreement":null},{"id":"W3048924598","doi":"10.1038/s41597-020-00594-z","title":"A comprehensive CHO SWATH-MS spectral library for robust quantitative profiling of 10,000 proteins","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Spinal Cord Injury BC","funders":"Biomedical Research Council; Agency for Science, Technology and Research","keywords":"Chinese hamster ovary cell; Computer science; Profiling (computer programming); Proteome; Computational biology; Data acquisition; Downstream (manufacturing); Biological system; Chemistry; Biology; Cell culture; Bioinformatics; Engineering","score_opus":0.1247426534309085,"score_gpt":0.32926990674785844,"score_spread":0.20452725331694993,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3048924598","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25806865,0.004753791,0.6331779,0.00057835004,0.0002312872,0.0012508795,0.06639565,0.018657386,0.016886061],"genre_scores_gemma":[0.1705522,0.0032229705,0.6632592,0.001012085,0.00012120647,0.0036663907,0.1421837,0.0046055573,0.011376694],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.99906105,0.00010580491,0.00008093404,0.0002858515,0.00038721933,0.00007909985],"domain_scores_gemma":[0.99924195,0.000118435644,0.00008748752,0.00015257717,0.00031617767,0.00008343755],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001983004,0.0016005336,0.0012392115,0.0032821454,0.0013732433,0.0010844117,0.00092547433,0.0006744797,0.0058090524],"category_scores_gemma":[0.0010957484,0.00052341336,0.0009519074,0.0029146424,0.00034598436,0.0012514998,0.0013903606,0.0010787237,0.0038765583],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034497006,0.00011036362,0.0010523958,0.00033971507,0.0001275242,0.000108127584,0.00007418747,0.0003641996,0.9572598,0.00052784145,0.0049133226,0.03477751],"study_design_scores_gemma":[0.00008386999,0.00041798205,0.010813136,0.00007130958,0.00020955014,0.0008661314,0.00006338203,0.007189026,0.91048366,0.00088875665,0.06882398,0.00008917074],"about_ca_topic_score_codex":0.0006178687,"about_ca_topic_score_gemma":0.0016723225,"teacher_disagreement_score":0.0058090524,"about_ca_system_score_codex":0.00059724995,"about_ca_system_score_gemma":0.0012791766,"threshold_uncertainty_score":0.0194332},"labels":[],"label_agreement":null},{"id":"W3080864197","doi":"10.1038/s41597-020-00610-2","title":"HIT-COVID, a global database tracking public health interventions to COVID-19","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":115,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McMaster University","funders":"National Institute of General Medical Sciences; Johnson and Johnson Foundation; Johns Hopkins University","keywords":"Coronavirus disease 2019 (COVID-19); 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Pandemic; Public health; Tracking (education); Global health; Geography; Data science; Computer science; Virology; Biology; Medicine; Outbreak; Infectious disease (medical specialty); Psychology; Nursing","score_opus":0.8307176163106647,"score_gpt":0.5738583493022208,"score_spread":0.25685926700844386,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3080864197","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025532295,0.0012827014,0.0059984215,0.0011987071,0.00023575246,0.001061178,0.97795993,0.0018501504,0.007859938],"genre_scores_gemma":[0.0093550505,0.001934716,0.026485642,0.0009141077,0.0001830885,0.0035460102,0.9552705,0.0005235365,0.0017872408],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9859746,0.0041884254,0.005767374,0.0019272222,0.0017631213,0.00037919122],"domain_scores_gemma":[0.87632656,0.05194239,0.025751723,0.020784095,0.019008685,0.006186545],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018581098,0.0013350849,0.0019930862,0.019000955,0.0012072691,0.0045966404,0.0031237255,0.0020628397,0.024977306],"category_scores_gemma":[0.09882324,0.0013609882,0.0017654123,0.024937099,0.0005634049,0.0040856204,0.0040634247,0.0027426302,0.010326954],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001089799,0.00016321632,0.049422383,0.011979674,0.000984741,0.00018345304,0.0013703391,0.0017049655,0.00068332505,0.01176102,0.81896293,0.10169416],"study_design_scores_gemma":[0.00070541893,0.00016694957,0.047070682,0.0053627947,0.00066376734,0.00017584905,0.0008433867,0.0020686646,0.0013694009,0.006718575,0.93468845,0.00016620706],"about_ca_topic_score_codex":0.02249446,"about_ca_topic_score_gemma":0.024390424,"teacher_disagreement_score":0.024977306,"about_ca_system_score_codex":0.002417534,"about_ca_system_score_gemma":0.014984108,"threshold_uncertainty_score":0.09826738},"labels":[],"label_agreement":null},{"id":"W3080911893","doi":"10.1038/s41597-020-00578-z","title":"Arctic tidal current atlas","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Oceanographic and Atmospheric Processes","field":"Earth and Planetary Sciences","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Fisheries and Oceans Canada","funders":"Norges Forskningsråd; Bundesministerium für Bildung und Forschung; National Science Foundation","keywords":"Baroclinity; Current (fluid); Oceanography; Arctic; Tidal power; Environmental science; Ocean current; Submarine pipeline; Tidal range; Climatology; Geology; Estuary; Ecology","score_opus":0.07603865341572055,"score_gpt":0.25238641647405463,"score_spread":0.17634776305833408,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3080911893","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011826783,0.0019360256,0.012049142,0.00044111078,0.0010290871,0.0002453722,0.87103635,0.0024966486,0.098939344],"genre_scores_gemma":[0.027164841,0.0018162258,0.019078074,0.00024569125,0.00022589696,0.00029933878,0.92387116,0.00064253935,0.02665621],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992429,0.000116300616,0.0001368161,0.00014925665,0.00026744258,0.00008729542],"domain_scores_gemma":[0.9969766,0.00019521861,0.00028465642,0.000321192,0.0019634354,0.00025896518],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013046166,0.0006628847,0.00059886195,0.004975643,0.0010209151,0.0013541855,0.00092966744,0.000392293,0.02824197],"category_scores_gemma":[0.0023196219,0.00020142547,0.00048010703,0.006560479,0.00017892456,0.0008821738,0.0010960501,0.0006856547,0.019689461],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032123,0.000048741676,0.03712947,0.00087607483,0.00010696578,0.00025888474,0.00061346864,0.0029487074,0.002176867,0.012842585,0.81927377,0.12340331],"study_design_scores_gemma":[0.000015122506,0.000013745713,0.02224341,0.00013487827,0.000025630558,0.00011459191,0.0001262459,0.00067541085,0.00029195563,0.001137744,0.97519934,0.00002191424],"about_ca_topic_score_codex":0.08710004,"about_ca_topic_score_gemma":0.0978399,"teacher_disagreement_score":0.08710004,"about_ca_system_score_codex":0.0011639916,"about_ca_system_score_gemma":0.0034182782,"threshold_uncertainty_score":0.17318612},"labels":[],"label_agreement":null},{"id":"W3082836448","doi":"10.1038/s41597-020-00584-1","title":"Author Correction: A global database of Holocene paleotemperature records","year":2020,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Okanagan University College; University of British Columbia, Okanagan Campus; Bishop's University; Okanagan College; Geological Survey of Canada; University of Ottawa; University of New Brunswick; Center for Northern Studies; Concordia University; Université du Québec à Montréal; University of British Columbia; Fisheries and Oceans Canada; Mount Royal University; Université Laval; Natural Resources Canada","funders":"","keywords":"Holocene; Database; Geology; Oceanography; Data science; Information retrieval; Computer science","score_opus":0.06046163971018304,"score_gpt":0.30597480834678126,"score_spread":0.24551316863659822,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3082836448","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00037682577,0.00071266637,0.001199298,0.02908159,0.9255247,0.000056262063,0.02864392,0.0009998564,0.01340489],"genre_scores_gemma":[0.013348059,0.005222917,0.013075703,0.041900348,0.12729669,0.00049014256,0.081185825,0.005803275,0.711677],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9958016,0.00040190737,0.0008390636,0.00052628934,0.0021466163,0.00028453578],"domain_scores_gemma":[0.95823735,0.006396545,0.0018293527,0.0038619447,0.028236901,0.0014378534],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040165107,0.001623971,0.0012130378,0.005644333,0.0026111607,0.004427794,0.0025311897,0.0024068775,0.14245985],"category_scores_gemma":[0.048222817,0.0008397643,0.0010962924,0.0052838447,0.0010258914,0.0023687896,0.002364471,0.0052637924,0.09293067],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013800179,0.0000028036095,0.00006911345,0.000055584496,0.000003284755,0.000041524636,0.000010440632,0.000018719888,0.000020494957,0.00030494688,0.9959896,0.0034695866],"study_design_scores_gemma":[0.000019893649,0.0000066275834,0.0007688172,0.00020610199,0.000014120316,0.00011100209,0.000050855608,0.00006632727,0.0001677972,0.00060192065,0.9979711,0.000015600666],"about_ca_topic_score_codex":0.029570185,"about_ca_topic_score_gemma":0.041551065,"teacher_disagreement_score":0.14245985,"about_ca_system_score_codex":0.0024847395,"about_ca_system_score_gemma":0.0075378674,"threshold_uncertainty_score":0.47657567},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3085670454","doi":"10.1038/s41597-020-00652-6","title":"Structural models of human ACE2 variants with SARS-CoV-2 Spike protein for structure-based drug design","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"SARS-CoV-2 and COVID-19 Research","field":"Medicine","cited_by":35,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children","funders":"Bundesministerium für Bildung und Forschung; European Regional Development Fund; Deutsche Forschungsgemeinschaft","keywords":"Coronavirus; Alanine scanning; Angiotensin-converting enzyme 2; Computational biology; Missense mutation; Mutagenesis; Biology; Genome; Mutant; Indel; Genetics; Coronavirus disease 2019 (COVID-19); Mutation; Gene; Medicine; Infectious disease (medical specialty)","score_opus":0.20444825342388231,"score_gpt":0.37951834475289553,"score_spread":0.17507009132901322,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3085670454","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8713232,0.005159982,0.029866533,0.0017313387,0.00022578555,0.00027578662,0.07420962,0.0030118383,0.014195894],"genre_scores_gemma":[0.82340384,0.002782922,0.029928243,0.00034615712,0.00005002657,0.00049474003,0.14009206,0.0004318962,0.0024701261],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99979204,0.000058374764,0.00001309969,0.0000386585,0.00005622092,0.00004157661],"domain_scores_gemma":[0.99973434,0.00013037921,0.000028091485,0.00002501274,0.000045738132,0.000036373403],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00035795567,0.0015989214,0.0013301515,0.00062264345,0.00065232796,0.00090593926,0.0015081299,0.0013123595,0.0059970603],"category_scores_gemma":[0.00081717403,0.00055298285,0.0017939697,0.00089543307,0.00027107605,0.00057012984,0.00044624304,0.0011280215,0.0010256569],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00094760227,0.00041020088,0.0064607644,0.00077114935,0.00033898794,0.0008100874,0.00011348135,0.94627357,0.015950672,0.003976201,0.013746503,0.010200926],"study_design_scores_gemma":[0.00064411305,0.00030501746,0.0019489329,0.00006480079,0.00013020521,0.00016348815,0.0001145855,0.97367555,0.0056053223,0.0037592053,0.013549023,0.000039772123],"about_ca_topic_score_codex":0.014177697,"about_ca_topic_score_gemma":0.01924268,"teacher_disagreement_score":0.014177697,"about_ca_system_score_codex":0.0014117783,"about_ca_system_score_gemma":0.0019155756,"threshold_uncertainty_score":0.028190374},"labels":[],"label_agreement":null},{"id":"W3087421614","doi":"10.1038/s41597-020-00644-6","title":"A high-resolution in vivo magnetic resonance imaging atlas of the human hypothalamic region","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neonatal and fetal brain pathology","field":"Medicine","cited_by":146,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Western Hospital; University of Toronto","funders":"McDonnell Center for Systems Neuroscience; National Institutes of Health; NIH Blueprint for Neuroscience Research; Canada Research Chairs; National Institute of Mental Health; Deutsche Forschungsgemeinschaft","keywords":"Atlas (anatomy); Magnetic resonance imaging; Nuclear magnetic resonance; In vivo; Functional magnetic resonance imaging; Biology; Physics; Medicine; Neuroscience; Anatomy; Radiology","score_opus":0.03990451522155245,"score_gpt":0.26367124791062385,"score_spread":0.2237667326890714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3087421614","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12737983,0.0014679395,0.85142356,0.00054714363,0.00012658013,0.0006594797,0.006811322,0.0024280103,0.009156122],"genre_scores_gemma":[0.342537,0.001688167,0.64192855,0.00020113969,0.0000785299,0.000895175,0.006766296,0.0007346859,0.0051704706],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.9998369,0.000043237127,0.000016013872,0.000042382366,0.000045689376,0.000015831727],"domain_scores_gemma":[0.9997676,0.00006649352,0.000036646932,0.00007174725,0.00004196678,0.000015663218],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00047586928,0.00035076097,0.00026245543,0.00085359823,0.00035962468,0.00082636985,0.000438352,0.00059411966,0.0037986704],"category_scores_gemma":[0.00094624807,0.00032633994,0.0003604627,0.000827058,0.00030350825,0.00036572458,0.0005924157,0.000494781,0.0013173153],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005838573,0.00022765248,0.013590284,0.0011490787,0.00024110198,0.0018957393,0.0014303665,0.048527133,0.5245884,0.016225409,0.021604251,0.36993676],"study_design_scores_gemma":[0.00019517426,0.001239036,0.17383671,0.0004897905,0.00046021547,0.02853035,0.0011348708,0.23518665,0.2828028,0.02722871,0.24856557,0.0003302067],"about_ca_topic_score_codex":0.0020355317,"about_ca_topic_score_gemma":0.005873187,"teacher_disagreement_score":0.0037986704,"about_ca_system_score_codex":0.00027010473,"about_ca_system_score_gemma":0.0011534672,"threshold_uncertainty_score":0.01270777},"labels":[],"label_agreement":null},{"id":"W3088562264","doi":"10.1038/s41597-020-00648-2","title":"A database of chlorophyll and water chemistry in freshwater lakes","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Aquatic Ecosystems and Phytoplankton Dynamics","field":"Environmental Science","cited_by":81,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Wilfrid Laurier University; York University","funders":"Natural Sciences and Engineering Research Council of Canada; York University; Ministry of Environment; McGill University; Center for Urban Science and Progress; Government of Canada","keywords":"Environmental science; Water quality; Chlorophyll a; Biomass (ecology); Proxy (statistics); Freshwater ecosystem; Watershed; Nutrient; Productivity; Ecosystem; Ecology; Database; Hydrology (agriculture); Biology; Computer science","score_opus":0.02431380870858216,"score_gpt":0.22138061792069283,"score_spread":0.19706680921211067,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3088562264","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008160444,0.0022950266,0.0016197508,0.00009525252,0.000015126043,0.00017415787,0.9849386,0.00059928786,0.0021023746],"genre_scores_gemma":[0.02252768,0.0020523758,0.009529055,0.00015387712,0.000030991898,0.0011635777,0.9634787,0.0001260236,0.0009377353],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980813,0.00026088112,0.0008041708,0.0003951333,0.00036543646,0.000093143244],"domain_scores_gemma":[0.99469393,0.0015636757,0.0016260081,0.0004823537,0.0013325276,0.00030140392],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016678704,0.000759933,0.0013979524,0.012163946,0.0005192959,0.0013233254,0.0012545837,0.0008463532,0.012387564],"category_scores_gemma":[0.008127668,0.00049478526,0.0011272873,0.019948324,0.00023422022,0.0014014964,0.0017564194,0.00056320545,0.004477472],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019679924,0.0003901191,0.186443,0.058051184,0.0027401727,0.0011796507,0.0019723864,0.0052740015,0.017721444,0.007868022,0.4364228,0.27996927],"study_design_scores_gemma":[0.0003767123,0.00013403044,0.20860066,0.002725814,0.0010177394,0.00043126236,0.0003795476,0.0013413895,0.0037567746,0.0031406723,0.7779358,0.0001594936],"about_ca_topic_score_codex":0.0090370765,"about_ca_topic_score_gemma":0.012673014,"teacher_disagreement_score":0.012387564,"about_ca_system_score_codex":0.0011120309,"about_ca_system_score_gemma":0.0033281916,"threshold_uncertainty_score":0.041440547},"labels":[],"label_agreement":null},{"id":"W3091795721","doi":"10.1038/s41597-020-00690-0","title":"GalliForm, a database of Galliformes occurrence records from the Indo-Malay and Palaearctic, 1800–2008","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Fisheries and Oceans Canada","funders":"Zoological museum, Lomonosov Moscow State University; Lomonosov Moscow State University; Leverhulme Trust","keywords":"Macroecology; Database; Biodiversity; Geography; Species distribution; Ecology; Galliformes; Taxon; Biology; Habitat","score_opus":0.09181699789533744,"score_gpt":0.276527385229699,"score_spread":0.18471038733436157,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3091795721","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.32655227,0.003299791,0.0019574868,0.00016569345,0.000071568575,0.0002153175,0.6492995,0.0007830234,0.017655386],"genre_scores_gemma":[0.2908215,0.002618025,0.009426796,0.00007606352,0.00005706824,0.0004947054,0.6884494,0.00015317471,0.007903332],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996612,0.000028552406,0.00011482504,0.000080262835,0.00008061618,0.000034506793],"domain_scores_gemma":[0.9977543,0.00028617244,0.0010530712,0.00033123407,0.00034784028,0.00022742792],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041228236,0.00056798605,0.00038488416,0.010225518,0.00037422206,0.000523146,0.00068183,0.00025414137,0.009034391],"category_scores_gemma":[0.0027261064,0.00022748542,0.00026968168,0.008073546,0.00035027295,0.00082313037,0.0011340799,0.0002613019,0.0036760387],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00088362553,0.00010906111,0.556338,0.009025031,0.00064807705,0.0019165755,0.0068308543,0.0010878175,0.016804682,0.002089503,0.123438396,0.2808284],"study_design_scores_gemma":[0.000019512927,0.00004631543,0.81181777,0.0003145952,0.00013691658,0.0009161421,0.0008258957,0.00040732615,0.0016010484,0.00024981407,0.18361634,0.000048384394],"about_ca_topic_score_codex":0.020442685,"about_ca_topic_score_gemma":0.045194656,"teacher_disagreement_score":0.020442685,"about_ca_system_score_codex":0.00046469824,"about_ca_system_score_gemma":0.0011009948,"threshold_uncertainty_score":0.040647388},"labels":[],"label_agreement":null},{"id":"W3092045521","doi":"10.1038/s41597-020-00642-8","title":"Large eQTL meta-analysis reveals differing patterns between cerebral cortical and cerebellar brain regions","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":330,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Pacific Centre for Reproductive Medicine; University of British Columbia; Canadian Institute for Advanced Research","funders":"National Institute of Neurological Disorders and Stroke; National Institute on Drug Abuse; National Institute on Aging; National Institutes of Health; Università degli Studi di Trento; National Institute of Mental Health; Arizona Biomedical Research Commission; Translational Genomics Research Institute; Rush University; Michael J. Fox Foundation for Parkinson's Research; Arizona Department of Health Services; CurePSP; University of Pennsylvania; Mayo Foundation for Medical Education and Research; F. Hoffmann-La Roche; Illinois Department of Public Health; University of Pittsburgh; Mayo Clinic","keywords":"Meta-analysis; Neuroscience; Cerebellum; Biology; Medicine; Internal medicine","score_opus":0.09744670688283534,"score_gpt":0.29753978829237937,"score_spread":0.20009308140954402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3092045521","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7304343,0.08801367,0.14437406,0.005547007,0.0005626442,0.0001539346,0.023371687,0.0021122403,0.005430394],"genre_scores_gemma":[0.9834277,0.0039167935,0.007858174,0.0005676959,0.00006379068,0.00008903714,0.0033802118,0.00026234187,0.00043421632],"study_design_codex":"observational","study_design_gemma":"meta_analysis","domain_scores_codex":[0.9928778,0.0033458765,0.0005505809,0.0023320713,0.0005485458,0.00034500242],"domain_scores_gemma":[0.9811271,0.0143107595,0.0009778842,0.0029221706,0.00044987234,0.00021227192],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01158346,0.0013495191,0.0019957153,0.0028824965,0.0007267126,0.0026279646,0.0011565292,0.0010521781,0.0047577056],"category_scores_gemma":[0.02303404,0.00053490873,0.01094259,0.004021651,0.0007462384,0.00089368667,0.0015506978,0.0014331848,0.00040984616],"study_design_candidate":"meta_analysis","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0053063896,0.00007075223,0.4798759,0.0041406425,0.393124,0.0013336973,0.00048583676,0.015203641,0.024226684,0.0046763765,0.005543955,0.06601207],"study_design_scores_gemma":[0.0012173806,0.00071043865,0.5515168,0.0009794885,0.3712292,0.001339557,0.00045722054,0.023930266,0.009062595,0.023786427,0.015618338,0.00015238802],"about_ca_topic_score_codex":0.00689423,"about_ca_topic_score_gemma":0.013487682,"teacher_disagreement_score":0.01158346,"about_ca_system_score_codex":0.00076730957,"about_ca_system_score_gemma":0.00086343754,"threshold_uncertainty_score":0.061259925},"labels":[],"label_agreement":null},{"id":"W3092143966","doi":"10.1038/s41597-020-00676-y","title":"A salmon diet database for the North Pacific Ocean","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and fisheries research","field":"Environmental Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Fisheries and Oceans Canada","funders":"Hakai Institute; University of British Columbia; G. Unger Vetlesen Foundation; Ambrose Monell Foundation","keywords":"Pacific ocean; Oceanography; Fishery; Geography; Biology; Geology","score_opus":0.0831538589209788,"score_gpt":0.28426685395865087,"score_spread":0.20111299503767205,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3092143966","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005500063,0.0005165177,0.0066807284,0.0002012336,0.00003815243,0.0003912528,0.97569036,0.001997976,0.008983667],"genre_scores_gemma":[0.0063088355,0.0006289423,0.02256424,0.00015608994,0.000022951206,0.0009632466,0.9658601,0.0003799754,0.0031156272],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998995,0.00009766016,0.00040555364,0.00016433469,0.00028296877,0.000054560376],"domain_scores_gemma":[0.9954786,0.0007029973,0.0009005684,0.0006657686,0.0018533403,0.00039872617],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001862861,0.0005800275,0.00069873856,0.0076902704,0.00070835825,0.0017402082,0.0016120295,0.0005386221,0.03391787],"category_scores_gemma":[0.0062657953,0.000545633,0.00054053013,0.014445795,0.00019233549,0.0019991773,0.0018584693,0.00086413056,0.014880765],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005262714,0.00016541602,0.034949362,0.0036682263,0.0002598649,0.0003781737,0.0006117768,0.00087555073,0.0025062687,0.0054703243,0.74286455,0.20772414],"study_design_scores_gemma":[0.00021165323,0.0000524748,0.07179609,0.0008462589,0.00013538702,0.00036088438,0.00040284797,0.0014623947,0.0010891894,0.003496991,0.9200557,0.00009003363],"about_ca_topic_score_codex":0.027263723,"about_ca_topic_score_gemma":0.0346422,"teacher_disagreement_score":0.03391787,"about_ca_system_score_codex":0.000937295,"about_ca_system_score_gemma":0.005855684,"threshold_uncertainty_score":0.11346662},"labels":[],"label_agreement":null},{"id":"W3092872384","doi":"10.1038/s41597-020-00784-9","title":"COVIDiSTRESS Global Survey dataset on psychological and behavioural consequences of the COVID-19 outbreak","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 and Mental Health","field":"Psychology","cited_by":134,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; University of Toronto","funders":"Vlaamse regering; Fonds Wetenschappelijk Onderzoek; Grantová Agentura České Republiky; Japan Society for the Promotion of Science; National Research University Higher School of Economics; Ministry of Education, Culture, Sports, Science and Technology; Institut Français de Bioinformatique","keywords":"Coronavirus disease 2019 (COVID-19); Outbreak; 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Psychology; Geography; Data science; Virology; Medicine; Computer science; Infectious disease (medical specialty); Pathology; Disease","score_opus":0.42882880990942107,"score_gpt":0.5209819841713665,"score_spread":0.09215317426194547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3092872384","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010773225,0.0001276765,0.00018514547,0.00029129474,0.0000385441,0.00007002843,0.9871941,0.00007201893,0.0012479569],"genre_scores_gemma":[0.016968897,0.0001339839,0.0007328022,0.00017243886,0.000029535462,0.00054346235,0.9805156,0.000021907445,0.0008814167],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99880886,0.0003907616,0.00018485857,0.00018814449,0.000258406,0.00016894654],"domain_scores_gemma":[0.9953499,0.001671233,0.00084168493,0.0005462499,0.0011986869,0.0003922668],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0014618523,0.0005594992,0.000616841,0.0025015753,0.00050793326,0.00092815974,0.0008454242,0.0010252157,0.010160072],"category_scores_gemma":[0.00906034,0.00025035426,0.0006666903,0.004719052,0.00021953184,0.00049455184,0.0016777607,0.0008322253,0.0058403043],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027066498,0.00016029993,0.12408243,0.0015697662,0.00021893025,0.00014046856,0.0006158224,0.001206661,0.0004401038,0.0013252667,0.8524199,0.017549712],"study_design_scores_gemma":[0.00020736092,0.00015177114,0.5283684,0.0008954204,0.00010153068,0.0002405322,0.0023361112,0.0017355991,0.00044282156,0.0016375848,0.463776,0.000106835374],"about_ca_topic_score_codex":0.039594565,"about_ca_topic_score_gemma":0.052467354,"teacher_disagreement_score":0.99915457,"about_ca_system_score_codex":0.0010121448,"about_ca_system_score_gemma":0.0014039149,"threshold_uncertainty_score":0.0787282},"labels":[],"label_agreement":null},{"id":"W3094042624","doi":"10.1038/s41597-020-00701-0","title":"Long-term surveys of age structure in 13 ungulate and one ostrich species in the Serengeti, 1926–2018","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Wildlife Ecology and Conservation","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Zoologische Gesellschaft Frankfurt; Natural Sciences and Engineering Research Council of Canada; Government of Canada; Wildlife Conservation Society","keywords":"Ungulate; Geography; Ecology; Biodiversity; Juvenile; Biology; Habitat","score_opus":0.06183281563456003,"score_gpt":0.24654999900480473,"score_spread":0.1847171833702447,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3094042624","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99556774,0.00014090078,0.00012308023,0.00001548983,0.0000060641346,0.000010872648,0.00348226,0.000007905932,0.00064569863],"genre_scores_gemma":[0.9869287,0.00020569572,0.00079972553,0.000038156315,0.000019696992,0.00008183376,0.011160086,0.0000071776576,0.00075891294],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998387,0.00002061645,0.000012317443,0.00005700027,0.000026646612,0.000044719127],"domain_scores_gemma":[0.99944097,0.00006162099,0.00024508813,0.0000506138,0.00009324772,0.0001084713],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003522856,0.00030597078,0.00024718078,0.001190783,0.00053785613,0.0002890473,0.0003751495,0.00025254427,0.001259234],"category_scores_gemma":[0.000692882,0.00018859144,0.00013493073,0.0011114511,0.00041754398,0.00031246443,0.0005699288,0.00039835094,0.0005420723],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012288622,0.000061678504,0.98643607,0.000042570602,0.00003539457,0.00015876835,0.0013004964,0.00021608791,0.0016287654,0.000041075928,0.0011204181,0.008835812],"study_design_scores_gemma":[0.0000030026683,0.000030012012,0.9983134,0.000013486987,0.000007957558,0.000082054736,0.00044524486,0.00012049251,0.000109251116,0.000011751337,0.00086033804,0.0000031163195],"about_ca_topic_score_codex":0.026122028,"about_ca_topic_score_gemma":0.096251495,"teacher_disagreement_score":0.026122028,"about_ca_system_score_codex":0.00042111037,"about_ca_system_score_gemma":0.00024717773,"threshold_uncertainty_score":0.051939964},"labels":[],"label_agreement":null},{"id":"W3095025955","doi":"10.1038/s41597-020-00711-y","title":"Individual back-calculated size-at-age based on otoliths from Pacific coral reef fish species","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and fisheries research","field":"Environmental Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Human Genome Research Institute; Fondation de France; Fondation BNP Paribas; BNP Paribas Cardif; Agence Nationale de la Recherche","keywords":"Coral reef; Coral reef fish; Fishery; Reef; Coral; Fish <Actinopterygii>; Biology; Oceanography; Geography; Ecology; Geology","score_opus":0.0891040629753068,"score_gpt":0.26381286344026394,"score_spread":0.17470880046495713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3095025955","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97392106,0.0004506161,0.0094451625,0.000024894369,0.000018381708,0.000023631814,0.0118536,0.00016128436,0.004101463],"genre_scores_gemma":[0.9725113,0.00031879233,0.013666219,0.00001724585,0.000010924245,0.00004464922,0.010846854,0.000055544704,0.0025284702],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997907,0.000011745147,0.000025973963,0.00007970249,0.00007211543,0.000019765415],"domain_scores_gemma":[0.9989712,0.00020927662,0.00033204124,0.00011768308,0.00031468467,0.000055224216],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004126429,0.00046235978,0.00020165723,0.0014001708,0.0003923046,0.0003512671,0.0002743487,0.00016804005,0.0029566514],"category_scores_gemma":[0.001654373,0.00015352081,0.00034038504,0.000769591,0.00023072945,0.00034577513,0.00055143255,0.00020095664,0.0009374539],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000070774775,0.000010691194,0.9332451,0.00008310416,0.00012150677,0.00008742344,0.00030795365,0.0021673169,0.012066041,0.00012100559,0.00078649935,0.050932545],"study_design_scores_gemma":[0.0000010463967,0.000014451538,0.99490607,0.000005472708,0.000027252128,0.00013306913,0.00006830855,0.0016626245,0.0016537538,0.000063826694,0.0014515442,0.000012559486],"about_ca_topic_score_codex":0.026588548,"about_ca_topic_score_gemma":0.09377628,"teacher_disagreement_score":0.026588548,"about_ca_system_score_codex":0.00043565425,"about_ca_system_score_gemma":0.00027128676,"threshold_uncertainty_score":0.05286759},"labels":[],"label_agreement":null},{"id":"W30988412","doi":"10.1038/s41597-019-0038-1","title":"Desenvolupament d’un mètode d’odometria visual per a un robot mòbil en entorns exteriors","year":2013,"lang":"en","type":"dissertation","venue":"Scientific Data","topic":"Robotics and Sensor-Based Localization","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Humanities; Physics; Computer science; Art","score_opus":0.018831324122114888,"score_gpt":0.2646404752014437,"score_spread":0.2458091510793288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W30988412","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07788217,0.0011528875,0.8460967,0.0010353826,0.00047592365,0.00014477884,0.00067925785,0.0036940244,0.06883896],"genre_scores_gemma":[0.31627035,0.0013852511,0.6460662,0.0001609116,0.00011065929,0.00020214583,0.0008200487,0.0004617191,0.034522712],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.999281,0.00015968736,0.000020022911,0.00020222426,0.00027009516,0.00006699775],"domain_scores_gemma":[0.9994628,0.000121259036,0.000048208447,0.00012536965,0.00019921055,0.000043161588],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008514857,0.0008447301,0.000432748,0.0011976138,0.00043221473,0.0015416463,0.00071541517,0.000600282,0.010375992],"category_scores_gemma":[0.0022148022,0.00042298262,0.00047935615,0.0008897892,0.0009097149,0.001671277,0.001642266,0.0007457582,0.0037530006],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020851509,0.00015020068,0.00703603,0.0002980308,0.00004950939,0.000127133,0.0010163838,0.030432383,0.028476791,0.021855587,0.013451537,0.896898],"study_design_scores_gemma":[0.00011782588,0.00053050404,0.035014696,0.0005967752,0.00009144969,0.0009657678,0.003051949,0.34721583,0.07312904,0.025971193,0.513095,0.00022005866],"about_ca_topic_score_codex":0.0064767217,"about_ca_topic_score_gemma":0.007975652,"teacher_disagreement_score":0.010375992,"about_ca_system_score_codex":0.0005326165,"about_ca_system_score_gemma":0.0010265925,"threshold_uncertainty_score":0.034711182},"labels":[],"label_agreement":null},{"id":"W3102112318","doi":"10.1038/s41597-020-00719-4","title":"Paired rRNA-depleted and polyA-selected RNA sequencing data and supporting multi-omics data from human T cells","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"RNA modifications and cancer","field":"Biochemistry, Genetics and Molecular Biology","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"European Commission; British Heart Foundation; Government of Canada","keywords":"Ribosomal RNA; Biology; RNA; RNA splicing; Epigenetics; Computational biology; Transcriptome; RNA-Seq; Gene; Genetics; Gene expression","score_opus":0.15829019945819303,"score_gpt":0.3294361991482268,"score_spread":0.17114599969003377,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3102112318","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4457792,0.0018536011,0.077531464,0.0007374209,0.00028836328,0.00024386348,0.46258965,0.0016493637,0.009327091],"genre_scores_gemma":[0.43614322,0.0012515389,0.11865583,0.0010999048,0.00019612137,0.0011907821,0.43204615,0.00091312267,0.008503305],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992681,0.0001050114,0.00007560918,0.00024588846,0.00021653996,0.00008876415],"domain_scores_gemma":[0.99907684,0.00030219866,0.00007989214,0.00022736634,0.00023083565,0.00008281041],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00079656363,0.00043092397,0.0005374116,0.0009909837,0.00076997903,0.00072575925,0.00031189361,0.0005432793,0.009600503],"category_scores_gemma":[0.001216246,0.00020264872,0.000598618,0.0016513357,0.0002804173,0.00016604528,0.0005432304,0.0006754243,0.0030497946],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011376193,0.00012870775,0.01772852,0.0009885599,0.00024584212,0.0006668965,0.00033561434,0.0029443956,0.92067903,0.0012480656,0.014075241,0.039821494],"study_design_scores_gemma":[0.00021721073,0.00050854427,0.18157238,0.00023289892,0.0005045704,0.002149171,0.0006097624,0.007444465,0.58886814,0.0053426223,0.21239871,0.00015148245],"about_ca_topic_score_codex":0.0016391083,"about_ca_topic_score_gemma":0.0035494738,"teacher_disagreement_score":0.009600503,"about_ca_system_score_codex":0.00024544343,"about_ca_system_score_gemma":0.00085931603,"threshold_uncertainty_score":0.03211683},"labels":[],"label_agreement":null},{"id":"W3108117466","doi":"10.1038/s41597-020-00737-2","title":"North American historical monthly spatial climate dataset, 1901–2016","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Natural Resources Canada; Canadian Forest Service","funders":"Canadian Forest Service; U.S. Forest Service; Natural Resources Canada; Environment and Climate Change Canada","keywords":"Precipitation; Environmental science; Climatology; Mean radiant temperature; Climate change; Maximum temperature; Statistics; Atmospheric sciences; Mathematics; Meteorology; Geography; Geology","score_opus":0.02849479680893761,"score_gpt":0.23064362183725673,"score_spread":0.20214882502831913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3108117466","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011407612,0.0001843198,0.00048585868,0.00015023614,0.000050025614,0.0000163392,0.98539793,0.00029960088,0.002008001],"genre_scores_gemma":[0.015959386,0.00017886641,0.0011559359,0.000054112203,0.000020810885,0.00010374507,0.98112476,0.000056357847,0.0013460566],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9997466,0.00003542744,0.000024824714,0.000080844264,0.000077945835,0.00003438133],"domain_scores_gemma":[0.99911755,0.00004909634,0.000112845315,0.00013151599,0.0005219069,0.000067174034],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00043542215,0.00058373535,0.00049288705,0.0013222554,0.0004059989,0.000552277,0.00084956904,0.00030859775,0.0064270576],"category_scores_gemma":[0.0012133457,0.0002793784,0.00051319646,0.0041498956,0.00015251363,0.0004211736,0.00053537544,0.00064547424,0.005318672],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000108468725,0.00008260857,0.058948494,0.00031433642,0.0002323539,0.00007365936,0.00011053067,0.010138436,0.00036574728,0.0010145985,0.91539884,0.013211876],"study_design_scores_gemma":[0.00013554102,0.000019571093,0.25548092,0.00017370582,0.00012145139,0.00011494422,0.00024774298,0.015902722,0.0009198015,0.0010348932,0.7257813,0.00006750228],"about_ca_topic_score_codex":0.32383883,"about_ca_topic_score_gemma":0.42361307,"teacher_disagreement_score":0.67616117,"about_ca_system_score_codex":0.0013430556,"about_ca_system_score_gemma":0.0025784639,"threshold_uncertainty_score":0.6439078},"labels":[],"label_agreement":null},{"id":"W3108537578","doi":"10.1038/s41597-020-00763-0","title":"ClimateEU, scale-free climate normals, historical time series, and future projections for Europe","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":124,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Climate change; Scale (ratio); Downscaling; Environmental science; Climate model; Climatology; Download; Adaptation (eye); Software; Computer science; Database; Environmental resource management; Meteorology; Geography; Cartography; Ecology; Geology","score_opus":0.04382605702617909,"score_gpt":0.2420340178003698,"score_spread":0.1982079607741907,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3108537578","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15993503,0.003156238,0.12393298,0.0012804586,0.00037166756,0.0001126298,0.6720228,0.0073328246,0.03185539],"genre_scores_gemma":[0.4115457,0.0030355148,0.116212286,0.00016494177,0.00018200344,0.00030879976,0.4627904,0.0010222533,0.0047381953],"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993748,0.00017499205,0.00009690253,0.0001436449,0.0001744545,0.000035320198],"domain_scores_gemma":[0.99803203,0.00055507896,0.00026593555,0.00056181307,0.00048078533,0.00010434894],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023681815,0.00042012622,0.00039017663,0.0034543243,0.00018019424,0.0016393859,0.00085274776,0.00046976912,0.006485949],"category_scores_gemma":[0.0068472526,0.0002662869,0.0005932248,0.0074072373,0.00016233919,0.0013732538,0.00087608525,0.0004627776,0.002000164],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00047557254,0.00017494,0.1288485,0.0011089569,0.00065053685,0.00039251556,0.00045418565,0.43461215,0.0022191051,0.03949884,0.17516918,0.21639556],"study_design_scores_gemma":[0.000171299,0.0001098105,0.1467962,0.00060245715,0.0002119786,0.00037220542,0.0005168606,0.29490736,0.0037142234,0.06408982,0.48830095,0.00020684372],"about_ca_topic_score_codex":0.014119221,"about_ca_topic_score_gemma":0.008366398,"teacher_disagreement_score":0.014119221,"about_ca_system_score_codex":0.00045477826,"about_ca_system_score_gemma":0.000808661,"threshold_uncertainty_score":0.028074026},"labels":[],"label_agreement":null},{"id":"W3109120520","doi":"10.1038/s41597-020-00762-1","title":"Bioinformatic analysis of membrane and associated proteins in murine cardiomyocytes and human myocardium","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"RNA and protein synthesis mechanisms","field":"Biochemistry, Genetics and Molecular Biology","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Alberta; Ted Rogers Centre for Heart Research; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canada First Research Excellence Fund; Canadian Institutes of Health Research; Alberta Innovates; University of Toronto; Government of Canada","keywords":"Cell biology; Membrane; Computational biology; Membrane protein; Biology; Chemistry; Biochemistry","score_opus":0.03046961960581235,"score_gpt":0.2608618438097863,"score_spread":0.23039222420397393,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3109120520","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9464925,0.0031963314,0.009688897,0.00015833993,0.000037749494,0.00005974179,0.038717702,0.0005665155,0.0010821902],"genre_scores_gemma":[0.84297216,0.0016801417,0.039741747,0.0002366971,0.00003245951,0.00020735,0.11390362,0.00020458967,0.0010213264],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99971586,0.00003341788,0.000029456658,0.0001290245,0.00006310024,0.000029224193],"domain_scores_gemma":[0.9997683,0.00007179473,0.000060525414,0.00002635846,0.000039352424,0.00003360244],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004596539,0.00033818328,0.00036071968,0.0012953706,0.00023826529,0.00045541316,0.00018302735,0.00026968837,0.0009636833],"category_scores_gemma":[0.00066275644,0.00008641152,0.00044652857,0.0009693095,0.00010910562,0.0001802006,0.0003243536,0.00018770345,0.0006295776],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021179037,0.00016359922,0.12906142,0.0014717825,0.0005272631,0.0014355102,0.0002915985,0.0063964063,0.7683877,0.0005374169,0.0047155037,0.084894],"study_design_scores_gemma":[0.00010148605,0.0005467194,0.760839,0.00019315451,0.0004891978,0.0034364115,0.00053312705,0.05567773,0.14537936,0.0015628993,0.03117174,0.00006928817],"about_ca_topic_score_codex":0.00063017226,"about_ca_topic_score_gemma":0.001115831,"teacher_disagreement_score":0.0012953706,"about_ca_system_score_codex":0.00020417703,"about_ca_system_score_gemma":0.00026972953,"threshold_uncertainty_score":0.0032238364},"labels":[],"label_agreement":null},{"id":"W3114591989","doi":"10.1038/s41597-021-01084-6","title":"Global daily 1 km land surface precipitation based on cloud cover-informed downscaling","year":2021,"lang":"en","type":"preprint","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Nuclear Safety and Security Commission; Deutsche Forschungsgemeinschaft; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; McGill University; Biodiversa+; Svenska Forskningsrådet Formas; Technology Agency of the Czech Republic; Agence Nationale de la Recherche; National Aeronautics and Space Administration; Vetenskapsrådet; National Science Foundation","keywords":"Downscaling; Precipitation; Environmental science; Orographic lift; Climatology; Cloud cover; Meteorology; Terrain; Orography; Cloud computing; Computer science; Geography; Geology","score_opus":0.055391883100898806,"score_gpt":0.3009565304965644,"score_spread":0.24556464739566558,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3114591989","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7157604,0.00034744042,0.26189223,0.00029195205,0.00014026821,0.000119622,0.009017984,0.0029793282,0.0094507085],"genre_scores_gemma":[0.86193824,0.00018397324,0.12222889,0.000060501716,0.0000497702,0.000060051207,0.0137801645,0.00031152836,0.0013868031],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998807,0.000020019574,0.0000064163173,0.000041255025,0.000038193502,0.000013449416],"domain_scores_gemma":[0.9997987,0.000025429988,0.00002633243,0.000059118203,0.00008071836,0.00000971118],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00029317822,0.00030920943,0.00022154542,0.00046077388,0.0001427394,0.000365919,0.00039287383,0.00023305851,0.0010792048],"category_scores_gemma":[0.0009987293,0.00018478544,0.000387362,0.0006728803,0.0001371145,0.0004384289,0.0003556986,0.0002935046,0.00042850836],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000112360285,0.000087965454,0.06768132,0.000081190745,0.00019575976,0.00014628767,0.00009216522,0.71673423,0.018079886,0.0037261806,0.008050808,0.18501183],"study_design_scores_gemma":[0.000022378308,0.000011332072,0.04051771,0.00001040958,0.000015681748,0.000023141005,0.000013587169,0.9507568,0.003743165,0.001841794,0.003024902,0.000019224493],"about_ca_topic_score_codex":0.010293002,"about_ca_topic_score_gemma":0.013854603,"teacher_disagreement_score":0.010293002,"about_ca_system_score_codex":0.00024913726,"about_ca_system_score_gemma":0.00043990812,"threshold_uncertainty_score":0.020466208},"labels":[],"label_agreement":null},{"id":"W3118241888","doi":"10.1038/s41597-020-00790-x","title":"Dataset of spiking and LFP activity invasively recorded in the human amygdala during aversive dynamic stimuli","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Memory and Neural Mechanisms","field":"Neuroscience","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Foundation for Research in Science and the Humanities; Russian Science Foundation; Universität Zürich; Mach-Gaensslen Foundation of Canada; Russian Foundation for Basic Research; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Amygdala; Neuroscience; Aversive Stimulus; Psychology","score_opus":0.24531642119854527,"score_gpt":0.3804034085416959,"score_spread":0.13508698734315064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3118241888","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.61840427,0.0020703655,0.030182794,0.0004773444,0.00023866662,0.00036570636,0.34030938,0.002398656,0.0055529033],"genre_scores_gemma":[0.51069236,0.0006799363,0.01743862,0.00020309509,0.0001329392,0.00082523405,0.46729216,0.00021047713,0.0025250597],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.99979633,0.000023028337,0.000023242665,0.00006385162,0.000058942926,0.000034610963],"domain_scores_gemma":[0.99952924,0.000116834715,0.00004873513,0.00011304942,0.00013919926,0.000053033582],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.00019417927,0.00057949545,0.00054863765,0.0008067299,0.00026101258,0.00038886658,0.00051833753,0.0007941331,0.0019790037],"category_scores_gemma":[0.00090143306,0.00011935977,0.0005265574,0.00085573556,0.00024787794,0.000189326,0.0006210787,0.00039869588,0.0012072019],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031153648,0.001468122,0.10725084,0.00452921,0.0014326393,0.00960805,0.0010684023,0.027857224,0.39435914,0.0016924817,0.16657974,0.28103885],"study_design_scores_gemma":[0.0003317279,0.0012104744,0.78174806,0.00029053463,0.0004653813,0.011745979,0.00076351094,0.034689475,0.05603596,0.0040806555,0.10832206,0.00031615744],"about_ca_topic_score_codex":0.0019277178,"about_ca_topic_score_gemma":0.004950724,"teacher_disagreement_score":0.9994817,"about_ca_system_score_codex":0.00014419481,"about_ca_system_score_gemma":0.00033293833,"threshold_uncertainty_score":0.006620407},"labels":[],"label_agreement":null},{"id":"W3122040552","doi":"10.1038/s41597-021-00800-6","title":"A complete dietary review of Japanese birds with special focus on molluscs","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Avian ecology and behavior","field":"Environmental Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Merck Canada Inc. (Canada)","funders":"","keywords":"Archipelago; Predation; Ecology; Biology; Resource (disambiguation); Ecosystem; Agriculture; Geography","score_opus":0.05785304932552797,"score_gpt":0.2803396562191138,"score_spread":0.2224866068935858,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3122040552","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04134502,0.85227084,0.0027365112,0.0032521707,0.0029653471,0.00038071565,0.067383245,0.00038973882,0.029276434],"genre_scores_gemma":[0.06634256,0.81743526,0.011165195,0.0025243366,0.001966441,0.00064310967,0.08207083,0.00022569911,0.017626593],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9990778,0.00013227841,0.00037648095,0.00011736028,0.00024098755,0.000055103053],"domain_scores_gemma":[0.99500906,0.0010258706,0.0009931478,0.00023612007,0.0024674465,0.00026836712],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012205312,0.0006064299,0.0012680781,0.020556742,0.0006719665,0.0008432801,0.0006279169,0.00043210867,0.00824471],"category_scores_gemma":[0.003734471,0.00033880514,0.00090179814,0.017143968,0.00036226943,0.0014580473,0.00087856967,0.00036730705,0.0018208944],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034063795,0.00009236476,0.020337906,0.10055176,0.0006585434,0.00051784277,0.0023052488,0.00019707375,0.005910558,0.0011507622,0.19983643,0.6681009],"study_design_scores_gemma":[0.000014325705,0.00011735269,0.10101572,0.016163142,0.0010573464,0.00062632916,0.0009165278,0.00007610969,0.0006935635,0.0003036805,0.87897146,0.00004443171],"about_ca_topic_score_codex":0.015502316,"about_ca_topic_score_gemma":0.041209128,"teacher_disagreement_score":0.020556742,"about_ca_system_score_codex":0.0008918581,"about_ca_system_score_gemma":0.0034434423,"threshold_uncertainty_score":0.030824125},"labels":[],"label_agreement":null},{"id":"W3124931592","doi":"10.1038/s41597-021-00820-2","title":"A compilation of North American tree provenance trials and relevant historical climate data for seven species","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Forest ecology and management","field":"Environmental Science","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada; Canadian Forest Service; Ontario Forest Research Institute; Ministry of Natural Resources and Forestry","funders":"","keywords":"Provenance; Climate change; Pinus <genus>; Abiotic component; Ecology; Range (aeronautics); Geography; Forestry; Biology; Botany; Engineering","score_opus":0.09550338386737549,"score_gpt":0.29889415795320823,"score_spread":0.20339077408583273,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3124931592","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08261338,0.0017659074,0.005898128,0.00008523468,0.00015797513,0.00021041295,0.8892962,0.00121318,0.018759709],"genre_scores_gemma":[0.06342022,0.0008811876,0.008328703,0.00009345869,0.000044936838,0.00077704375,0.92333597,0.00044840088,0.0026701433],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9984635,0.00027694073,0.00026589909,0.00036491491,0.00053530175,0.000093461385],"domain_scores_gemma":[0.9863803,0.002658576,0.002669241,0.003302101,0.0043068933,0.0006829045],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020311882,0.0006651075,0.00084736425,0.006421841,0.0010308041,0.0011680179,0.0009207079,0.0003830527,0.009834024],"category_scores_gemma":[0.0046068756,0.00046334762,0.0006486112,0.014351693,0.0004384813,0.0010225134,0.0008751312,0.00076218677,0.0047620675],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013134221,0.0005471634,0.3345553,0.0051048296,0.0014549507,0.000683494,0.0019337656,0.005353238,0.015195985,0.002586618,0.43317255,0.19809872],"study_design_scores_gemma":[0.000089734785,0.00010643512,0.4961045,0.00034413656,0.0002760138,0.00027512264,0.0004421996,0.00073729176,0.002397912,0.0008918428,0.49823198,0.00010286378],"about_ca_topic_score_codex":0.020730682,"about_ca_topic_score_gemma":0.08006574,"teacher_disagreement_score":0.020730682,"about_ca_system_score_codex":0.00066298485,"about_ca_system_score_gemma":0.0019882028,"threshold_uncertainty_score":0.04122001},"labels":[],"label_agreement":null},{"id":"W3125634994","doi":"10.1038/s41597-021-00805-1","title":"High-resolution, large-scale laboratory measurements of a sandy beach and dynamic cobble berm revetment","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Coastal and Marine Dynamics","field":"Earth and Planetary Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Australian Research Council; Horizon 2020 Framework Programme; Engineering and Physical Sciences Research Council; University of Bath; Deltares; European Commission; Research Councils UK","keywords":"Swash; Berm; Revetment; Cobble; Wave flume; Flume; Beach morphodynamics; Geology; Environmental science; Wave loading; Benthic zone; Wave height; Imbrication; Shoal; Hydrology (agriculture); Geomorphology; Geotechnical engineering; Oceanography; Breaking wave; Sediment; Sediment transport; Habitat; Ecology; Breakup; Submarine pipeline; Wave propagation","score_opus":0.024624088657585943,"score_gpt":0.23439753644425923,"score_spread":0.20977344778667328,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3125634994","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99474144,0.000019295749,0.0021133807,0.000025925876,0.000010426394,0.00007914207,0.0011798732,0.00009860019,0.0017318666],"genre_scores_gemma":[0.99152607,0.00003635052,0.0058217803,0.000052040727,0.0000056115236,0.0002054938,0.001127388,0.00003594591,0.0011894039],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99979824,0.000012785019,0.000008731466,0.00006153558,0.00007469098,0.000044022483],"domain_scores_gemma":[0.99968946,0.000036424186,0.000047345893,0.00004470938,0.0001158802,0.0000662521],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00030642122,0.0002444131,0.0002919593,0.00030683138,0.0006866902,0.0004859938,0.00027177786,0.00032468126,0.0014388176],"category_scores_gemma":[0.00042689688,0.0001622349,0.00019982265,0.0003395759,0.00033496125,0.0003528779,0.0004792826,0.00061867357,0.00029499622],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005706827,0.00060663227,0.112015404,0.00008468886,0.00004202065,0.00027326378,0.0006121457,0.002238964,0.86532503,0.00022491724,0.0010129319,0.016993374],"study_design_scores_gemma":[0.000082574385,0.0014457938,0.92964756,0.000012872928,0.000029048322,0.000100058234,0.0008684253,0.005886958,0.058333836,0.00014741436,0.0034024632,0.000043048218],"about_ca_topic_score_codex":0.0076603917,"about_ca_topic_score_gemma":0.02884124,"teacher_disagreement_score":0.0076603917,"about_ca_system_score_codex":0.00039647895,"about_ca_system_score_gemma":0.00057404186,"threshold_uncertainty_score":0.015231609},"labels":[],"label_agreement":null},{"id":"W3125770330","doi":"10.1038/s41597-021-00816-y","title":"Community-curated and standardised metadata of published ancient metagenomic samples with AncientMetagenomeDir","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"National Institute of General Medical Sciences; Arts and Humanities Research Council; Universidad Nacional Autónoma de México; Max-Planck-Gesellschaft; Carlsbergfondet; Deutsche Forschungsgemeinschaft; European Commission; Research Councils UK; Wellcome Trust","keywords":"Metadata; Metagenomics; Computer science; Interoperability; Data science; Information retrieval; World Wide Web; Geospatial metadata; Biology; Data element; Meta Data Services","score_opus":0.049485412251043394,"score_gpt":0.27135146186826553,"score_spread":0.22186604961722214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3125770330","genre_codex":"dataset","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02351314,0.0020880033,0.09437045,0.00097121176,0.0008991428,0.0016438821,0.8008519,0.044845596,0.030816672],"genre_scores_gemma":[0.016657777,0.0007347803,0.16244127,0.0004997978,0.00013474132,0.0015557999,0.80432117,0.008704306,0.0049503963],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9893929,0.0016640731,0.0034206489,0.002102553,0.0028456084,0.0005742413],"domain_scores_gemma":[0.961657,0.0077398084,0.00567337,0.013406389,0.00937056,0.0021529638],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.02232911,0.0019566733,0.0020340085,0.030483045,0.0034093861,0.0072259908,0.0032145784,0.0017375719,0.030468911],"category_scores_gemma":[0.046353746,0.0014278017,0.001838962,0.024529219,0.0018073019,0.0072692367,0.0102635585,0.0034293788,0.024104951],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030923984,0.0006563179,0.043644484,0.019175638,0.0008905954,0.0019882077,0.010703158,0.0026402548,0.12936977,0.045153428,0.45075223,0.29193354],"study_design_scores_gemma":[0.00013612006,0.00011698448,0.024373,0.0019199783,0.00028744427,0.00078315096,0.0012133213,0.0011294207,0.024154253,0.008365676,0.9371961,0.00032449575],"about_ca_topic_score_codex":0.006475277,"about_ca_topic_score_gemma":0.015141618,"teacher_disagreement_score":0.9967854,"about_ca_system_score_codex":0.002278565,"about_ca_system_score_gemma":0.012015776,"threshold_uncertainty_score":0.11808902},"labels":[],"label_agreement":null},{"id":"W3126221909","doi":"10.1038/s41597-021-00829-7","title":"The Cuban Human Brain Mapping Project, a young and middle age population-based EEG, MRI, and cognition dataset","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"Fonds de recherche du Québec – Nature et technologies; Helmholtz Association; University of Electronic Science and Technology of China; National Natural Science Foundation of China; Fondation Brain Canada","keywords":"Electroencephalography; Neuroimaging; Cognition; Population; Cognitive decline; Magnetic resonance imaging; Psychology; Audiology; Medicine; Neuroscience; Dementia; Disease; Pathology; Radiology","score_opus":0.15655278982966067,"score_gpt":0.3263458182746557,"score_spread":0.16979302844499503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3126221909","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017280972,0.00080661546,0.001029018,0.00025307058,0.00007697083,0.00025873687,0.9781542,0.00036717192,0.0017731611],"genre_scores_gemma":[0.018427538,0.00021173051,0.0014014158,0.00008277345,0.000027228913,0.0011259265,0.9771195,0.00007421702,0.0015296325],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99952114,0.00011574384,0.00005054936,0.00015199404,0.00007487236,0.000085753054],"domain_scores_gemma":[0.9990533,0.00012357289,0.000084485204,0.0002058263,0.00041515334,0.00011763159],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00091496273,0.0015061429,0.0013473103,0.0031521446,0.0009575987,0.0011837279,0.0023299784,0.00093910913,0.009312507],"category_scores_gemma":[0.0036340302,0.0003363589,0.00067381054,0.002873741,0.00026977016,0.0004276029,0.0016056013,0.0007369828,0.0067787436],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009989507,0.00021500641,0.02782238,0.0015509862,0.00042481237,0.00087754533,0.00029959288,0.0008522195,0.0015271209,0.00080530124,0.93446,0.030166157],"study_design_scores_gemma":[0.00074316416,0.00017865954,0.322063,0.000867019,0.00046771741,0.0021719446,0.0010098293,0.0036068924,0.0016104872,0.0024199334,0.66466457,0.00019674229],"about_ca_topic_score_codex":0.12737107,"about_ca_topic_score_gemma":0.17191233,"teacher_disagreement_score":0.12737107,"about_ca_system_score_codex":0.0017672982,"about_ca_system_score_gemma":0.0026077614,"threshold_uncertainty_score":0.25325936},"labels":[],"label_agreement":null},{"id":"W3129379202","doi":"10.1038/s41597-021-00851-9","title":"Author Correction: The FLUXNET2015 dataset and the ONEFlux processing pipeline for eddy covariance data","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Energy Load and Power Forecasting","field":"Engineering","cited_by":113,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; University of Manitoba; Centre de Géomatique du Québec; University of Saskatchewan; Canadian Forest Service; Environment and Climate Change Canada; University of British Columbia; University of Waterloo; Global Institute for Water Security; Université Laval; Natural Resources Canada; Ministère des Ressources naturelles et des Forêts; Ontario Drive & Gear (Canada); McMaster University","funders":"Natural Environment Research Council; Sight Research UK","keywords":"Pipeline (software); Eddy covariance; Covariance; Computer science; Data science; Statistics; Mathematics; Biology; Ecosystem; Ecology; Programming language","score_opus":0.08306270731006536,"score_gpt":0.3033050218793115,"score_spread":0.22024231456924614,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3129379202","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015642613,0.0016598123,0.022042897,0.02643679,0.5838168,0.00015708276,0.33195528,0.021145621,0.011221421],"genre_scores_gemma":[0.037338786,0.003947696,0.06623204,0.018448344,0.051130813,0.0013078143,0.5384342,0.05028815,0.23287222],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9946431,0.0007177777,0.0009701322,0.0011948692,0.002022774,0.00045126307],"domain_scores_gemma":[0.9718777,0.0047035175,0.0012008303,0.0053415275,0.015990296,0.00088605005],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0054899147,0.0024202506,0.001990959,0.0064675785,0.002227693,0.0055909376,0.0039428687,0.0035425222,0.25052562],"category_scores_gemma":[0.058295406,0.0016423763,0.00245556,0.006853397,0.0013828316,0.0038891935,0.0038833474,0.005475217,0.17333424],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026260914,0.000002886964,0.00016452486,0.00013466163,0.00001599202,0.000050056227,0.000021566624,0.00016810854,0.00016447638,0.0005659619,0.9940035,0.0046820347],"study_design_scores_gemma":[0.00006058899,0.000008656872,0.0011727166,0.0002802756,0.000033289085,0.00019081443,0.000076262506,0.00083291164,0.0008097712,0.003677038,0.9928,0.000057782836],"about_ca_topic_score_codex":0.014330665,"about_ca_topic_score_gemma":0.0290977,"teacher_disagreement_score":0.25052562,"about_ca_system_score_codex":0.0019697666,"about_ca_system_score_gemma":0.005740506,"threshold_uncertainty_score":0.83809173},"labels":[],"label_agreement":null},{"id":"W3136423134","doi":"10.1038/s41597-021-00874-2","title":"Author Correction: High-resolution, large-scale laboratory measurements of a sandy beach and dynamic cobble berm revetment","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Coastal and Marine Dynamics","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Cobble; Berm; Revetment; Scale (ratio); Environmental science; Geology; Geotechnical engineering; Hydrology (agriculture); Ecology; Geography; Cartography; Biology","score_opus":0.02847751722859263,"score_gpt":0.24724697988356958,"score_spread":0.21876946265497696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3136423134","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005638885,0.0009739826,0.002705839,0.018115275,0.97014636,0.000028441653,0.0028001054,0.0008666351,0.0037995875],"genre_scores_gemma":[0.07827114,0.012007929,0.030568166,0.05366253,0.307121,0.0004153634,0.018790979,0.009956969,0.48920593],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99576133,0.0003902282,0.00067554245,0.0007119218,0.0021715828,0.00028940092],"domain_scores_gemma":[0.9633514,0.005925308,0.0017067147,0.0033008913,0.024303608,0.0014120692],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026518474,0.0016320422,0.0013326381,0.0032964726,0.0031692404,0.0033492995,0.0028102603,0.004032535,0.06298206],"category_scores_gemma":[0.048676327,0.001064816,0.0013507766,0.0028719914,0.0016626164,0.0020511963,0.0024005326,0.0069347234,0.031149795],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004009861,0.0000041849544,0.00019857305,0.00017877735,0.000014254454,0.00028812356,0.000066358916,0.000115111296,0.00026439177,0.0007683504,0.98972064,0.0083410805],"study_design_scores_gemma":[0.000021260632,0.000011630893,0.000969439,0.00020451094,0.00002283501,0.0004337268,0.00010305604,0.00024783687,0.0006368781,0.0007740272,0.9965454,0.000029515522],"about_ca_topic_score_codex":0.012725715,"about_ca_topic_score_gemma":0.029826676,"teacher_disagreement_score":0.06298206,"about_ca_system_score_codex":0.0028037694,"about_ca_system_score_gemma":0.0053565796,"threshold_uncertainty_score":0.21069598},"labels":[],"label_agreement":null},{"id":"W3137174954","doi":"10.1038/s41597-022-01143-6","title":"A large-scale study on research code quality and execution","year":2022,"lang":"en","type":"preprint","venue":"Scientific Data","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Engineering and Physical Sciences Research Council; Alan Turing Institute; Alfred P. Sloan Foundation","keywords":"Computer science; Code reuse; Reuse; Replication (statistics); Coding (social sciences); Code (set theory); Transparency (behavior); Database; Set (abstract data type); Programming language; Software; Computer security","score_opus":0.7493102534435432,"score_gpt":0.600662037413241,"score_spread":0.1486482160303022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3137174954","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.938462,0.0049574464,0.03383323,0.0036526443,0.00024046833,0.00056295336,0.009512744,0.00283219,0.0059462474],"genre_scores_gemma":[0.9344575,0.0016308458,0.040129174,0.00087755936,0.00019876809,0.0005552792,0.018132193,0.0021014698,0.0019173463],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.91933435,0.032699525,0.008852768,0.012620357,0.024627244,0.0018657843],"domain_scores_gemma":[0.35984686,0.37536985,0.087146714,0.0977001,0.07511025,0.004826216],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.06990214,0.0008477963,0.00071599387,0.009103291,0.0020827507,0.0037262798,0.002248487,0.0012489336,0.0010960089],"category_scores_gemma":[0.39828172,0.00097637466,0.0012494839,0.015381799,0.0039957133,0.005115981,0.0029981479,0.0023441948,0.0010298535],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010982617,0.00048745197,0.7800267,0.0022200532,0.0011554295,0.00066827046,0.011453402,0.010389415,0.010064269,0.007132849,0.016188828,0.15911506],"study_design_scores_gemma":[0.00018299365,0.0008642187,0.8768828,0.0009295586,0.00052725425,0.0012115865,0.005136128,0.02658579,0.018764364,0.006757709,0.061800208,0.00035745348],"about_ca_topic_score_codex":0.012186054,"about_ca_topic_score_gemma":0.009795095,"teacher_disagreement_score":0.9300979,"about_ca_system_score_codex":0.003238899,"about_ca_system_score_gemma":0.0065768803,"threshold_uncertainty_score":0.36968225},"labels":[],"label_agreement":null},{"id":"W3139258240","doi":"10.1038/s41597-021-01034-2","title":"A curated dataset for data-driven turbulence modelling","year":2021,"lang":"en","type":"preprint","venue":"Scientific Data","topic":"Model Reduction and Neural Networks","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Reynolds-averaged Navier–Stokes equations; Turbulence; Benchmark (surveying); Computer science; Feature (linguistics); Large eddy simulation; Data set; Algorithm; Direct numerical simulation; Parametric statistics; Reynolds number; Data mining; Physics; Statistical physics; Artificial intelligence; Mathematics; Mechanics; Geology","score_opus":0.2410856084800902,"score_gpt":0.36543114375714847,"score_spread":0.12434553527705827,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3139258240","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019672925,0.00077682786,0.024307253,0.00059889915,0.00069830974,0.00036832714,0.9277463,0.020400632,0.0054305363],"genre_scores_gemma":[0.014295276,0.00013193309,0.015399702,0.00010355774,0.000042186002,0.00033972555,0.9681501,0.0005657267,0.00097181165],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99815637,0.00028856535,0.00019188544,0.0004554996,0.0007099583,0.0001975912],"domain_scores_gemma":[0.99700695,0.0005893843,0.00021158588,0.0011187783,0.00085062446,0.00022270204],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017784348,0.0020117192,0.0011902433,0.002930125,0.00089562935,0.0020219455,0.0038893616,0.0024983445,0.008363795],"category_scores_gemma":[0.0055387327,0.0005039575,0.0025495528,0.0027748754,0.0008048621,0.0015056997,0.0026034093,0.0026024946,0.013185059],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004985916,0.0006814835,0.008202276,0.0014454892,0.0002464566,0.00042720637,0.00012522445,0.0550511,0.004730018,0.0055237496,0.8607874,0.062281147],"study_design_scores_gemma":[0.00049922935,0.00028380513,0.020724997,0.0004835332,0.0000962183,0.00045910315,0.0003023354,0.16943254,0.014807299,0.0109459,0.78168553,0.00027958266],"about_ca_topic_score_codex":0.011455187,"about_ca_topic_score_gemma":0.019805437,"teacher_disagreement_score":0.011455187,"about_ca_system_score_codex":0.0011093091,"about_ca_system_score_gemma":0.0017971247,"threshold_uncertainty_score":0.027979672},"labels":[],"label_agreement":null},{"id":"W3159019058","doi":"10.1038/s41597-021-00900-3","title":"COVID-CT-MD, COVID-19 computed tomography scan dataset applicable in machine learning and deep learning","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 diagnosis using AI","field":"Medicine","cited_by":204,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Health Sciences Centre; University of Toronto; Sunnybrook Health Science Centre; McGill University Health Centre; Concordia University","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Coronavirus disease 2019 (COVID-19); Computed tomography; Artificial intelligence; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Chest radiograph; Computer science; Pneumonia; Medicine; Tomography; Radiology; Machine learning; Medical physics; Radiography; Infectious disease (medical specialty); Pathology; Disease; Internal medicine","score_opus":0.052622199486004814,"score_gpt":0.3532226569827437,"score_spread":0.30060045749673886,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3159019058","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04996346,0.0024861088,0.010870828,0.0013969886,0.00066561706,0.00068826973,0.92183506,0.0055632526,0.006530423],"genre_scores_gemma":[0.029479017,0.0004368362,0.009516599,0.00026098682,0.00006611504,0.0003292688,0.958207,0.00012288621,0.0015813211],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993119,0.00009066243,0.000081866616,0.00021002149,0.00019286004,0.00011271065],"domain_scores_gemma":[0.99944156,0.00009313969,0.000065702174,0.000121637975,0.00018781393,0.00009017609],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00066917186,0.002004988,0.000856475,0.0020508724,0.00082205416,0.0011610902,0.0024798764,0.001696672,0.005928984],"category_scores_gemma":[0.0023211022,0.00039861546,0.0011309723,0.0019532114,0.000529594,0.0007102875,0.0013027838,0.0016132364,0.0050109522],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011025516,0.0007643662,0.031729173,0.0017882292,0.0003320743,0.0013278483,0.000111691064,0.014259951,0.008907855,0.0021904975,0.8722036,0.06528223],"study_design_scores_gemma":[0.001288185,0.0008734085,0.087810695,0.0009256286,0.00043311308,0.0061242487,0.00072957715,0.14906305,0.026092533,0.008961991,0.7173226,0.00037497096],"about_ca_topic_score_codex":0.025810575,"about_ca_topic_score_gemma":0.050906517,"teacher_disagreement_score":0.025810575,"about_ca_system_score_codex":0.0012471694,"about_ca_system_score_gemma":0.002275128,"threshold_uncertainty_score":0.051320672},"labels":[],"label_agreement":null},{"id":"W3162096660","doi":"10.1038/s41597-021-00912-z","title":"Global data on earthworm abundance, biomass, diversity and corresponding environmental properties","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Invertebrate Taxonomy and Ecology","field":"Agricultural and Biological Sciences","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada Research Chairs; Natural Resources Canada; University of Toronto; Université de Sherbrooke; McGill University; Saint Mary's University","funders":"Biotechnology and Biological Sciences Research Council; Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Bundesministerium für Bildung und Forschung; Academy of Finland; Russian Foundation for Basic Research; Agence Nationale de la Recherche; Deutsche Forschungsgemeinschaft; Natural Sciences and Engineering Research Council of Canada; European Commission","keywords":"Earthworm; Abundance (ecology); Biomass (ecology); Diversity (politics); Ecology; Environmental science; Macroecology; Biodiversity; Biology","score_opus":0.0991375172588764,"score_gpt":0.22149763468346598,"score_spread":0.12236011742458958,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3162096660","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0067370376,0.0015811223,0.000390527,0.000085476044,0.00003360468,0.00003261662,0.98764205,0.00010170474,0.0033958447],"genre_scores_gemma":[0.017085575,0.0026129426,0.0019022926,0.00007375089,0.000045976052,0.0002267765,0.97658986,0.00005493495,0.0014079215],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9985454,0.00014678265,0.0004156104,0.00030695176,0.0004510995,0.0001341766],"domain_scores_gemma":[0.9906888,0.0026778711,0.0026745645,0.0009184878,0.0025468043,0.0004934418],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015793834,0.0011294585,0.0011778902,0.018980274,0.00032742662,0.0012209144,0.00080192194,0.0006042312,0.018216407],"category_scores_gemma":[0.008003294,0.00042975156,0.0010246031,0.03171022,0.0004884241,0.0011775885,0.0012828906,0.00074264343,0.01008236],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006383972,0.00018705659,0.2360074,0.037476927,0.0016675016,0.00076912186,0.0014762289,0.003978115,0.0065477067,0.0060286373,0.5405175,0.16470547],"study_design_scores_gemma":[0.00007333888,0.00005541286,0.30964744,0.002010876,0.00037482491,0.0003887219,0.00068664446,0.0002757605,0.0012451799,0.0015968876,0.68356955,0.00007541699],"about_ca_topic_score_codex":0.011752207,"about_ca_topic_score_gemma":0.016956171,"teacher_disagreement_score":0.018980274,"about_ca_system_score_codex":0.0006763741,"about_ca_system_score_gemma":0.001835909,"threshold_uncertainty_score":0.060939908},"labels":[],"label_agreement":null},{"id":"W3164073423","doi":"10.1038/s41597-021-00922-x","title":"Regional zircon U-Pb geochronology for the Maniitsoq region, southwest Greenland","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geological and Geochemical Analysis","field":"Earth and Planetary Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Geochronology; Zircon; Geology; Geochemistry","score_opus":0.07962031749569724,"score_gpt":0.24219144157598796,"score_spread":0.16257112408029073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3164073423","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.84919137,0.00059210806,0.0018161945,0.00015697698,0.000014716339,0.000072596944,0.1426051,0.00031216486,0.0052387174],"genre_scores_gemma":[0.75596386,0.0006266505,0.00793251,0.00018883518,0.000017541708,0.00018105922,0.22755851,0.00018184641,0.007349187],"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99985576,0.000008265794,0.000012304657,0.00005800085,0.000027680226,0.000038014172],"domain_scores_gemma":[0.9997454,0.000009501986,0.00005944198,0.000043997534,0.00010635719,0.000035285713],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041258862,0.0003775305,0.00025682003,0.0025308994,0.00048672978,0.0004769685,0.00041080313,0.00018837776,0.0016576808],"category_scores_gemma":[0.00038504842,0.00021821169,0.00026527303,0.0037299872,0.00019351848,0.0002990664,0.0008253279,0.00015192115,0.0007051892],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025674916,0.00007634921,0.8789955,0.00023231881,0.00023904213,0.00045343392,0.0018725842,0.0035451883,0.02123639,0.00094116636,0.017674394,0.07447693],"study_design_scores_gemma":[0.00001845715,0.0000125725865,0.9731122,0.000021196554,0.000035395155,0.00004905314,0.00036587636,0.00069699396,0.001222565,0.000092663926,0.024365973,0.0000070637493],"about_ca_topic_score_codex":0.34035528,"about_ca_topic_score_gemma":0.55111444,"teacher_disagreement_score":0.34035528,"about_ca_system_score_codex":0.0015881616,"about_ca_system_score_gemma":0.0028711024,"threshold_uncertainty_score":0.6767484},"labels":[],"label_agreement":null},{"id":"W3165720173","doi":"10.1038/s41597-021-00916-9","title":"A 120,000-year long climate record from a NW-Greenland deep ice core at ultra-high resolution","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"FP7 Environment; FP7 Ideas: European Research Council; Natural Environment Research Council; National Institute of Polar Research; Institut Polaire Français Paul Emile Victor; Centre National de la Recherche Scientifique; Korea Polar Research Institute; Agence Nationale de la Recherche; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Natural Resources Canada; Fonds Wetenschappelijk Onderzoek; Office of Polar Programs; Fonds De La Recherche Scientifique - FNRS; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Research Foundation; European Commission; Danmarks Grundforskningsfond; Villum Fonden; National Science Foundation","keywords":"Stadial; Ice core; Eemian; Geology; Chronology; Glacial period; Interglacial; Paleoclimatology; Abrupt climate change; Oceanography; Physical geography; Climatology; Climate change; Paleontology; Global warming; Geography; Effects of global warming","score_opus":0.059446238654265494,"score_gpt":0.27403551176324836,"score_spread":0.21458927310898288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3165720173","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9803587,0.00018947975,0.00035502255,0.00006559295,0.0000134288675,0.000014633756,0.01718992,0.000040495983,0.0017727265],"genre_scores_gemma":[0.94978315,0.00016125134,0.00239563,0.0001590484,0.000019366138,0.000028836866,0.045591798,0.00003268181,0.0018282727],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99989915,0.00000792442,0.000008992004,0.00003103672,0.000029445233,0.000023500472],"domain_scores_gemma":[0.9996778,0.000015172768,0.00006492556,0.000049558756,0.00013280696,0.000059801623],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00035145768,0.0002130167,0.00019841638,0.00092196406,0.00036425897,0.0004413903,0.00027469263,0.00018006765,0.0006166767],"category_scores_gemma":[0.00029052817,0.000109088345,0.00013760857,0.0012482384,0.00017962421,0.00029690302,0.00049293827,0.00021333012,0.0002836964],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021644322,0.00016651588,0.92345685,0.000093760726,0.00021760078,0.00030178312,0.00044328583,0.0017678423,0.030992921,0.00024248997,0.0066746185,0.035425868],"study_design_scores_gemma":[0.0000038584453,0.000007774579,0.99581367,0.0000071800996,0.000009847972,0.000025452775,0.000042586333,0.00031881532,0.00065136305,0.000021890208,0.0030941882,0.0000033317563],"about_ca_topic_score_codex":0.085712776,"about_ca_topic_score_gemma":0.22961512,"teacher_disagreement_score":0.085712776,"about_ca_system_score_codex":0.0008241507,"about_ca_system_score_gemma":0.00072311435,"threshold_uncertainty_score":0.1704278},"labels":[],"label_agreement":null},{"id":"W3173654308","doi":"10.1038/s41597-021-00948-1","title":"Profiling the small non-coding RNA transcriptome of the human placenta","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cancer-related molecular mechanisms research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Children's Hospital; University of British Columbia; Canadian Centre for Applied Research in Cancer Control; Izaak Walton Killam Health Centre","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; Canadian Institutes of Health Research; National Institutes of Health","keywords":"Transcriptome; Biology; microRNA; Placenta; Gene expression profiling; RNA; Computational biology; Genetics; Gene expression; Gene; Small nucleolar RNA; Bioinformatics; Fetus; Pregnancy; Non-coding RNA","score_opus":0.0531514596618521,"score_gpt":0.31158717941894265,"score_spread":0.25843571975709057,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3173654308","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9773932,0.0034927563,0.011947464,0.00009869338,0.000025364981,0.000058864545,0.005091759,0.00011463021,0.0017773282],"genre_scores_gemma":[0.96689034,0.0027860538,0.019016478,0.0001951351,0.000039955943,0.00010960878,0.009020013,0.000098600074,0.0018437511],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9998497,0.000024524012,0.000010552838,0.00005310525,0.000043314565,0.000018800847],"domain_scores_gemma":[0.99987054,0.00004151727,0.000027563277,0.000017012771,0.000030984877,0.000012333022],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00017539051,0.00013793318,0.00022932953,0.0005773564,0.00020557221,0.00042543781,0.00009573628,0.00016004845,0.00060617045],"category_scores_gemma":[0.0005361954,0.00012086174,0.0002242615,0.0005644342,0.00014305457,0.00010766767,0.0002108877,0.0002095148,0.0003862508],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019467222,0.0000103292705,0.016491031,0.00011795874,0.000037033744,0.0001530695,0.00021842192,0.00023458785,0.9722819,0.00010455121,0.00016725602,0.009989095],"study_design_scores_gemma":[0.000027352693,0.00062053977,0.5402793,0.000084712345,0.00027183237,0.0031077699,0.00076386053,0.0073505864,0.41704813,0.0009251343,0.029474584,0.00004610657],"about_ca_topic_score_codex":0.0008142584,"about_ca_topic_score_gemma":0.0012063454,"teacher_disagreement_score":0.0008142584,"about_ca_system_score_codex":0.00009332161,"about_ca_system_score_gemma":0.000231882,"threshold_uncertainty_score":0.0020278692},"labels":[],"label_agreement":null},{"id":"W3174112108","doi":"10.1038/s41597-021-00947-2","title":"Eight soybean reference genome resources from varying latitudes and agronomic traits","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soybean genetics and cultivation","field":"Agricultural and Biological Sciences","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Chinese Academy of Sciences Key Project; Ministry of Science and Technology of the People's Republic of China; Chinese Academy of Sciences","keywords":"Biology; Genome; Latitude; Agronomy; Biotechnology; Geography; Gene; Genetics","score_opus":0.0739302925013132,"score_gpt":0.2376375944388343,"score_spread":0.16370730193752112,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3174112108","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.60159475,0.0046739667,0.09865915,0.0005552535,0.00023900701,0.0010474356,0.2723209,0.00511245,0.015797015],"genre_scores_gemma":[0.08310624,0.0012973188,0.08477467,0.00012363725,0.00002284001,0.0005489018,0.8273332,0.0008276085,0.0019655642],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993795,0.00007124011,0.0001164566,0.00021028242,0.0001493367,0.00007302361],"domain_scores_gemma":[0.9986259,0.00026939606,0.00020864976,0.00031526663,0.0003700563,0.00021063205],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014122584,0.0011016395,0.0008505482,0.0035285535,0.00086433144,0.0007764354,0.0011969074,0.0005466559,0.003504103],"category_scores_gemma":[0.0018486563,0.0005595029,0.00084334356,0.00434278,0.00021871243,0.0006054628,0.0012908243,0.00094660925,0.0017683366],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025436128,0.00021348457,0.016261423,0.0016387939,0.00037419342,0.00095373415,0.0008907772,0.0012636636,0.90013295,0.0012957864,0.0059047663,0.06852675],"study_design_scores_gemma":[0.0011534838,0.0016076126,0.34328103,0.00071117637,0.002409094,0.0052419556,0.001104591,0.0067932797,0.30000108,0.0026654047,0.33470598,0.00032535192],"about_ca_topic_score_codex":0.002972683,"about_ca_topic_score_gemma":0.011769232,"teacher_disagreement_score":0.0035285535,"about_ca_system_score_codex":0.0006069434,"about_ca_system_score_gemma":0.0011948217,"threshold_uncertainty_score":0.011722386},"labels":[],"label_agreement":null},{"id":"W3176041963","doi":"10.1038/s41597-021-00939-2","title":"GlobSnow v3.0 Northern Hemisphere snow water equivalent dataset","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cryospheric studies and observations","field":"Earth and Planetary Sciences","cited_by":221,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"Academy of Finland; European Space Agency","keywords":"Snow; Water equivalent; Northern Hemisphere; Environmental science; Snowpack; Radiometer; Climatology; Defense Meteorological Satellite Program; Satellite; Meteorology; Remote sensing; Geography; Geology","score_opus":0.07065568232724688,"score_gpt":0.26332201574937014,"score_spread":0.19266633342212325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3176041963","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042475066,0.0000805725,0.00049609866,0.0000566399,0.00002367454,0.0000385092,0.9934587,0.00032248034,0.0012757914],"genre_scores_gemma":[0.004099139,0.00005711387,0.0011539245,0.000029762572,0.000012024318,0.00013513367,0.9939639,0.000052194755,0.00049678923],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9998148,0.000026018371,0.000028575945,0.000050832306,0.000045730256,0.000034039887],"domain_scores_gemma":[0.9996804,0.00003264179,0.000052989977,0.00006115452,0.00014388093,0.000028934142],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039531128,0.0006467359,0.0005153531,0.0013406344,0.00022928734,0.00044981085,0.0008537934,0.00033708164,0.0089831315],"category_scores_gemma":[0.0008448694,0.00018100272,0.00045011195,0.0031607389,0.000092595976,0.00046615617,0.0005943006,0.00043544435,0.006826117],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001691912,0.000068039175,0.02315357,0.0006857788,0.0001811109,0.00023383416,0.00016055592,0.004032551,0.0016687803,0.0018808447,0.9454219,0.022343874],"study_design_scores_gemma":[0.00024626605,0.00002381232,0.061849665,0.00014861456,0.000078977835,0.00011617773,0.00016518752,0.005010741,0.0011716404,0.0016080416,0.92954075,0.000040020466],"about_ca_topic_score_codex":0.04237707,"about_ca_topic_score_gemma":0.045759313,"teacher_disagreement_score":0.04237707,"about_ca_system_score_codex":0.00051287224,"about_ca_system_score_gemma":0.0010949804,"threshold_uncertainty_score":0.08426076},"labels":[],"label_agreement":null},{"id":"W3180664812","doi":"10.1038/s41597-021-00968-x","title":"A comprehensive annotation dataset of intact LTR retrotransposons of 300 plant genomes","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Chromosomal and Genetic Variations","field":"Agricultural and Biological Sciences","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Université Laval","funders":"Fundamental Research Funds for the Central Universities; Beijing Forestry University; National Natural Science Foundation of China","keywords":"Retrotransposon; Genome; Annotation; Biology; Long terminal repeat; Computational biology; Genetics; Transposable element; Evolutionary biology; Gene","score_opus":0.0803311561539355,"score_gpt":0.27020608102150895,"score_spread":0.18987492486757346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3180664812","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.031060275,0.0019827979,0.0036527608,0.00011083188,0.000041360458,0.00009145039,0.95889676,0.0018231105,0.002340762],"genre_scores_gemma":[0.0064798645,0.0003019265,0.00483382,0.000035627563,0.0000051973757,0.000119816636,0.9878698,0.00008460308,0.0002693654],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99858993,0.00016974502,0.00018059755,0.00061037735,0.00028995716,0.00015933756],"domain_scores_gemma":[0.99883336,0.00032054973,0.00018656763,0.00019691867,0.0002805842,0.00018211585],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008855733,0.0017177614,0.0016892996,0.004940181,0.0016379678,0.0012118459,0.0015979032,0.0016841544,0.0048572],"category_scores_gemma":[0.002236913,0.00052500726,0.0016147492,0.0072863214,0.00041007553,0.0011857592,0.0017946715,0.001522739,0.006409245],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003174311,0.0009910398,0.060464084,0.028181601,0.0012156586,0.0026462039,0.002149014,0.014295273,0.291388,0.006932476,0.42663324,0.16192901],"study_design_scores_gemma":[0.00039680576,0.00029779598,0.117846504,0.0008908436,0.0006081216,0.0012938407,0.0007176509,0.008222641,0.017901933,0.0031798182,0.8483973,0.00024681183],"about_ca_topic_score_codex":0.0064496547,"about_ca_topic_score_gemma":0.011827836,"teacher_disagreement_score":0.0064496547,"about_ca_system_score_codex":0.001015823,"about_ca_system_score_gemma":0.0025852087,"threshold_uncertainty_score":0.016249001},"labels":[],"label_agreement":null},{"id":"W3181216097","doi":"10.1038/s41597-021-00955-2","title":"A sub-national real-time epidemiological and vaccination database for the COVID-19 pandemic in Canada","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Data-Driven Disease Surveillance","field":"Medicine","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Guelph; SickKids Foundation; Hospital for Sick Children; Public Health Ontario; University of Toronto","funders":"Medical Research Council; National Institute for Health and Care Research; Wellcome Trust","keywords":"Pandemic; Coronavirus disease 2019 (COVID-19); Dashboard; Public health; Epidemiology; Outbreak; Vaccination; Open data; Geography; Political science; Medicine; Database; Computer science; World Wide Web; Nursing; Virology; Infectious disease (medical specialty)","score_opus":0.13854344895396647,"score_gpt":0.3767589396202392,"score_spread":0.23821549066627273,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3181216097","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009095409,0.0005220379,0.000738642,0.00064059737,0.00004062066,0.00019901012,0.9844514,0.0003557785,0.0039564893],"genre_scores_gemma":[0.041837122,0.0009373331,0.0050869933,0.00042041478,0.000028628965,0.00032790436,0.9484171,0.000110577996,0.0028339247],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99807966,0.000113875816,0.00023697683,0.00036052283,0.0007872395,0.00042161165],"domain_scores_gemma":[0.98861325,0.0007160615,0.000781998,0.00069796067,0.007537954,0.0016527954],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014379936,0.00079609244,0.0009339458,0.006566461,0.0029523168,0.0023771815,0.0019759678,0.000770513,0.009421432],"category_scores_gemma":[0.009546505,0.0005687428,0.00089348655,0.013180251,0.0005366816,0.00073595846,0.001615409,0.0013297944,0.003187858],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005591601,0.00012489357,0.11544984,0.0015354096,0.00033693202,0.000346081,0.00084018597,0.003050226,0.0007509175,0.0032900572,0.8274021,0.046314232],"study_design_scores_gemma":[0.00027749027,0.000061390245,0.39610264,0.0017434017,0.00024777683,0.0002497421,0.0017962919,0.007248303,0.0013701105,0.0013127441,0.58928746,0.00030257076],"about_ca_topic_score_codex":0.9944325,"about_ca_topic_score_gemma":0.9952362,"teacher_disagreement_score":0.9647921,"about_ca_system_score_codex":0.035207946,"about_ca_system_score_gemma":0.10168437,"threshold_uncertainty_score":0.2554527},"labels":[],"label_agreement":null},{"id":"W3188227111","doi":"10.1038/s41597-021-00983-y","title":"Global data set of long-term summertime vertical temperature profiles in 153 lakes","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and coastal ecosystems","field":"Earth and Planetary Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of the Environment, Conservation and Parks; Toronto Metropolitan University; University of Alberta; Université Laval; Laurentian University; University of Regina; International Institute for Sustainable Development","funders":"Natural Sciences and Engineering Research Council of Canada; Analyses et Expérimentations pour les Ecosystèmes; Suomen Ympäristökeskus; Russian Science Foundation; Institut National de la Recherche Agronomique; Ministry of Environment; Ministry of Business, Innovation and Employment; Vetenskapsrådet; University of California, Davis; University of Regina; Österreichischen Akademie der Wissenschaften; Ministère de l’Environnement, de la Protection de la nature et des Parcs; National Science Foundation; Belgian Federal Science Policy Office; Sight Research UK; Global Lake Ecological Observatory Network; Agence Nationale de la Recherche; Natural Environment Research Council; Florida International University; Klima- und Energiefonds; Russian Foundation for Basic Research; Queen's University; South Florida Water Management District; Canada Research Chairs","keywords":"Expansive; Environmental science; Climate change; Term (time); Global warming; Global change; Climatology; Geology; Atmospheric sciences; Physical geography; Hydrology (agriculture); Oceanography; Geography","score_opus":0.04959490324252962,"score_gpt":0.2750340758207046,"score_spread":0.22543917257817497,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3188227111","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5622228,0.00032872832,0.0018528878,0.00008029188,0.000024883087,0.00008027312,0.43324682,0.00037261043,0.0017906848],"genre_scores_gemma":[0.33539033,0.00020276765,0.005072968,0.0000595797,0.000024658446,0.00033232855,0.6579915,0.00008004956,0.00084585923],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99970156,0.000031788244,0.000058336867,0.000100644065,0.000066979424,0.00004060354],"domain_scores_gemma":[0.9992694,0.00006514733,0.00017941018,0.00013895646,0.00027228703,0.00007472271],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00033571382,0.0006647624,0.0004760494,0.0013869571,0.0004267268,0.00038429027,0.00047089532,0.00044691094,0.0012826945],"category_scores_gemma":[0.00064763305,0.00023738165,0.000385611,0.0033777568,0.00023754196,0.00037705913,0.0007034014,0.0002552027,0.0008871222],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00085461175,0.00046885724,0.8315079,0.00089795823,0.00082741765,0.00072408566,0.0013118464,0.01221453,0.03320735,0.0003673136,0.054421045,0.06319703],"study_design_scores_gemma":[0.000057066834,0.00009137867,0.9702955,0.00003704281,0.00010318815,0.00016126547,0.00035545765,0.001845834,0.00306902,0.000086059285,0.02385602,0.00004225478],"about_ca_topic_score_codex":0.041232005,"about_ca_topic_score_gemma":0.065845914,"teacher_disagreement_score":0.041232005,"about_ca_system_score_codex":0.0005045456,"about_ca_system_score_gemma":0.0010103605,"threshold_uncertainty_score":0.08198398},"labels":[],"label_agreement":null},{"id":"W3194100603","doi":"10.1038/s41597-021-01005-7","title":"Community composition of aquatic fungi across the thawing Arctic","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate change and permafrost","field":"Earth and Planetary Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; Center for Northern Studies; Université du Québec à Chicoutimi","funders":"Uppsala Multidisciplinary Center for Advanced Computational Science; Science for Life Laboratory; Sveriges Lantbruksuniversitet","keywords":"Permafrost; Thermokarst; Arctic; Environmental science; Aquatic ecosystem; Total organic carbon; Dissolved organic carbon; Environmental chemistry; Detritus; Ecology; Carbon fibers; Sediment; Biogeochemistry; Biology; Chemistry","score_opus":0.15627784578772005,"score_gpt":0.318352947738224,"score_spread":0.16207510195050392,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3194100603","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99923575,0.00014816737,0.00006432088,0.0000055268847,0.0000014463736,0.000004321008,0.0002961332,0.0000020341117,0.00024226848],"genre_scores_gemma":[0.9980977,0.00025822938,0.00049319444,0.000023025417,0.0000053491835,0.000010840567,0.0007405706,0.0000027397484,0.000368351],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99977666,0.000024820938,0.000016191607,0.0000843732,0.00004319603,0.000054799584],"domain_scores_gemma":[0.9997538,0.000026803558,0.000055290675,0.000008822301,0.000099129065,0.000056146273],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00020773806,0.00025711354,0.00035448503,0.0017841554,0.00085191464,0.00077556656,0.00013877051,0.00024623543,0.00067158625],"category_scores_gemma":[0.0002954708,0.000134582,0.0001904135,0.0010938778,0.00033286028,0.00025505165,0.00042823475,0.00019764299,0.0001302095],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036514195,0.00005786594,0.867923,0.00010558556,0.00007989825,0.00030557762,0.0028542895,0.00042305797,0.10951531,0.000081465485,0.00010774581,0.018181019],"study_design_scores_gemma":[0.0000021788587,0.000051169645,0.99722433,0.000011902128,0.000014617211,0.000112027185,0.0010560546,0.0001697381,0.0009429952,0.000014344662,0.0003958472,0.000004872926],"about_ca_topic_score_codex":0.017376747,"about_ca_topic_score_gemma":0.030540409,"teacher_disagreement_score":0.017376747,"about_ca_system_score_codex":0.00032293578,"about_ca_system_score_gemma":0.00030519263,"threshold_uncertainty_score":0.034551203},"labels":[],"label_agreement":null},{"id":"W3194353469","doi":"10.1038/s41597-022-01389-0","title":"Inter-species cell detection - datasets on pulmonary hemosiderophages in equine, human and feline specimens","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"AI in cancer detection","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Annotation; Computer science; Pipeline (software); Artificial intelligence; Cytology; Hemosiderin; Pathology; Medicine","score_opus":0.053244949372914664,"score_gpt":0.2823903223557649,"score_spread":0.22914537298285023,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3194353469","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6044987,0.008445599,0.020243606,0.0013213276,0.00043852185,0.0007057707,0.3441051,0.009313015,0.010928396],"genre_scores_gemma":[0.2639115,0.0013208,0.051531643,0.0005117768,0.000090649046,0.00058802316,0.6776241,0.00042429767,0.0039973105],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.998844,0.00012637604,0.00013234852,0.0004419741,0.00030290778,0.00015227673],"domain_scores_gemma":[0.9981748,0.00047564803,0.00019364002,0.0004677478,0.0005134406,0.00017474592],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014167615,0.0010800275,0.00058569515,0.0034463333,0.0009576926,0.000845276,0.0013127876,0.0020675126,0.0026588496],"category_scores_gemma":[0.0027438982,0.0002850814,0.0008605992,0.0015521869,0.0007025827,0.0005826581,0.0015913448,0.00079704996,0.00257377],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026822384,0.0011991172,0.21900506,0.008368753,0.00095942186,0.0070188968,0.0020869612,0.020565566,0.20373516,0.0031739327,0.25029975,0.28090516],"study_design_scores_gemma":[0.00022558411,0.00067559065,0.39617747,0.0011332578,0.00039354767,0.014096892,0.00257077,0.06680029,0.15235211,0.0046304623,0.36069375,0.00025021684],"about_ca_topic_score_codex":0.006895392,"about_ca_topic_score_gemma":0.015657518,"teacher_disagreement_score":0.006895392,"about_ca_system_score_codex":0.000845065,"about_ca_system_score_gemma":0.0007955157,"threshold_uncertainty_score":0.013710499},"labels":[],"label_agreement":null},{"id":"W3194760139","doi":"10.1038/s41597-021-00941-8","title":"Open-access quantitative MRI data of the spinal cord and reproducibility across participants, sites and manufacturers","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":77,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Philips (Canada); McGill University; Centre Hospitalier Universitaire de Sherbrooke; International Collaboration On Repair Discoveries; University of British Columbia; Université de Montréal; Université de Sherbrooke; Centre Hospitalier Universitaire Sainte-Justine; Polytechnique Montréal; Montreal Neurological Institute and Hospital; Mila - Quebec Artificial Intelligence Institute","funders":"National Institute of Neurological Disorders and Stroke; Staatssekretariat für Bildung, Forschung und Innovation; Economic and Social Research Council; Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; CRIS Cancer Foundation; Regione Puglia; University College London Hospitals NHS Foundation Trust; Ministero dell’Istruzione, dell’Università e della Ricerca; Ministero della Salute; Concordia University; Agentura Pro Zdravotnický Výzkum České Republiky; National Institutes of Health; Rosetrees Trust; European Commission; Multiple Sclerosis Society; Bundesministerium für Bildung und Forschung; National Imaging Facility; National Institute for Health and Care Research; University of Pennsylvania; SpinalCure Australia; University of Minnesota; National Science Foundation; Compute Canada; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; Polytechnique Montréal; Wellcome Trust; Institut de Valorisation des Données; AstraZeneca; Craig H. Neilsen Foundation; McGill University; Canada First Research Excellence Fund; Max-Planck-Gesellschaft","keywords":"Protocol (science); Reproducibility; Computer science; Spinal cord; Documentation; Data mining; Medicine; Statistics; Mathematics; Pathology","score_opus":0.6245720496163727,"score_gpt":0.5767339061992035,"score_spread":0.04783814341716919,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3194760139","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07016648,0.0068310634,0.39781493,0.005560696,0.0033227778,0.0073589515,0.45325306,0.018452251,0.037239723],"genre_scores_gemma":[0.20424879,0.003007717,0.32081735,0.0035610315,0.0011805277,0.03874187,0.39438957,0.016666891,0.017386245],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9727237,0.008445717,0.0047874767,0.0047021736,0.00871706,0.0006239324],"domain_scores_gemma":[0.8460825,0.05748343,0.012050132,0.046009578,0.036949597,0.0014247403],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.045483172,0.0017158035,0.0014566878,0.004205148,0.0014444222,0.0041127563,0.0033708054,0.002448414,0.045224566],"category_scores_gemma":[0.1963131,0.0012397546,0.00200293,0.004819531,0.0018156222,0.0030422937,0.004322847,0.0017663267,0.018530851],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0044510895,0.0006130082,0.060122952,0.012328694,0.0034578694,0.0008517175,0.0034683074,0.003684344,0.017278546,0.013361889,0.47890958,0.401472],"study_design_scores_gemma":[0.0017068062,0.0010428225,0.17642672,0.005143375,0.002153533,0.0038703308,0.0011578329,0.0054794275,0.020277044,0.043906722,0.73797184,0.00086354883],"about_ca_topic_score_codex":0.0030438055,"about_ca_topic_score_gemma":0.006465805,"teacher_disagreement_score":0.9966292,"about_ca_system_score_codex":0.0009324423,"about_ca_system_score_gemma":0.0039415765,"threshold_uncertainty_score":0.24054086},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3194991922","doi":"10.1038/s41597-021-01001-x","title":"Monitoring non-pharmaceutical public health interventions during the COVID-19 pandemic","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Food Security and Health in Diverse Populations","field":"Health Professions","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University Health Centre; McGill University","funders":"Economic and Social Research Council; Medical Research Council; Canadian Institutes of Health Research; Government of Canada","keywords":"Psychological intervention; Mandate; Pandemic; Public health; Coronavirus disease 2019 (COVID-19); Business; Public relations; Data science; Political science; Medicine; Computer science; Nursing; Infectious disease (medical specialty)","score_opus":0.7925991294067783,"score_gpt":0.6259501303240302,"score_spread":0.1666489990827481,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3194991922","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7442091,0.015609256,0.02808282,0.03689716,0.0009173498,0.004609644,0.079995245,0.00055356754,0.0891259],"genre_scores_gemma":[0.9253322,0.009176553,0.034097057,0.0065648668,0.00036016523,0.003606012,0.017758977,0.000077732766,0.0030265462],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9841867,0.008957934,0.0017805304,0.0013631389,0.0026713435,0.0010403243],"domain_scores_gemma":[0.9667157,0.012495419,0.010846373,0.0020817874,0.006190084,0.0016705744],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02033878,0.0004492694,0.0006111035,0.0028350896,0.00087575766,0.0018725744,0.00148482,0.0012406524,0.0022517145],"category_scores_gemma":[0.047193214,0.00032412546,0.0005293821,0.0031545463,0.0006677652,0.0020687487,0.0022119673,0.0016698637,0.0005612476],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007370296,0.00069351716,0.6697507,0.002921429,0.00070675666,0.00019678943,0.0029396983,0.003925428,0.0010559555,0.005025487,0.029200112,0.28284708],"study_design_scores_gemma":[0.00012860134,0.0011916291,0.91798896,0.004310266,0.0003398378,0.00023638003,0.005802927,0.005204747,0.003214068,0.0063237078,0.055134263,0.00012461783],"about_ca_topic_score_codex":0.038088437,"about_ca_topic_score_gemma":0.05295397,"teacher_disagreement_score":0.038088437,"about_ca_system_score_codex":0.0031283142,"about_ca_system_score_gemma":0.007515498,"threshold_uncertainty_score":0.10756302},"labels":[],"label_agreement":null},{"id":"W3195389227","doi":"10.1038/s41597-021-01007-5","title":"MNI-FTD templates, unbiased average templates of frontotemporal dementia variants","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Dementia and Cognitive Impairment Research","field":"Medicine","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Douglas Mental Health University Institute; McGill University; Centres Intégré Universitaires de Santé et de Services Sociaux; Montreal Neurological Institute and Hospital","funders":"National Institute on Aging; Alzheimer Society Research Program; University of California, San Francisco; Canadian Institutes of Health Research; University of Southern California; Alzheimer Society; National Institutes of Health; Consortium canadien en neurodégénérescence associée au vieillissement; Sanofi","keywords":"Frontotemporal dementia; Template; Computer science; Atrophy; Semantic dementia; Neuroimaging; Dementia; Probabilistic logic; Artificial intelligence; Medicine; Biology; Neuroscience; Pathology; Disease","score_opus":0.07653904424421291,"score_gpt":0.3469934429536982,"score_spread":0.2704543987094853,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3195389227","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.077415735,0.0022236453,0.83660525,0.00054151943,0.00037514354,0.0011105112,0.051012013,0.013702951,0.01701328],"genre_scores_gemma":[0.21831582,0.0010224716,0.70864207,0.00030843768,0.00010849756,0.0030946953,0.05370722,0.005040801,0.009759972],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9991462,0.00016000208,0.00013913316,0.0003017151,0.00019232888,0.000060605256],"domain_scores_gemma":[0.9988154,0.00024672065,0.0002168309,0.0003852897,0.00028898398,0.00004678327],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020736726,0.0012943372,0.00070464937,0.0024528215,0.0009444629,0.0017587392,0.0014942938,0.0009312943,0.012636212],"category_scores_gemma":[0.009552224,0.00067908684,0.0010896102,0.0023926857,0.00051071116,0.001051944,0.0013575637,0.00088316796,0.004210009],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00095421803,0.0001815618,0.044154078,0.0012718862,0.0013259406,0.001838941,0.0017640059,0.018158987,0.06078208,0.040006038,0.18130393,0.64825845],"study_design_scores_gemma":[0.0003881445,0.0005437941,0.23465158,0.0005957824,0.0011012672,0.015252633,0.00081055885,0.08955916,0.09470806,0.115375474,0.4466147,0.00039881942],"about_ca_topic_score_codex":0.015458915,"about_ca_topic_score_gemma":0.033298515,"teacher_disagreement_score":0.015458915,"about_ca_system_score_codex":0.0009332449,"about_ca_system_score_gemma":0.0019529243,"threshold_uncertainty_score":0.04227233},"labels":[],"label_agreement":null},{"id":"W3196272632","doi":"10.1038/s41597-021-01002-w","title":"Inter-laboratory mass spectrometry dataset based on passive sampling of drinking water for non-target analysis","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Isotope Analysis in Ecology","field":"Environmental Science","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Rijkswaterstaat; RECETOX Přírodovědecké Fakulty Masarykovy Univerzity; Masarykova Univerzita; Centre National de la Recherche Scientifique; Vlaamse Instelling voor Technologisch Onderzoek; Universitat de Girona; Örebro Universitet; Consiglio Nazionale delle Ricerche; Université Claude Bernard Lyon 1; Eidgenössische Anstalt für Wasserversorgung Abwasserreinigung und Gewässerschutz; Universität für Bodenkultur Wien; Bureau de Recherches Géologiques et Minières; National and Kapodistrian University of Athens; Euskal Herriko Unibertsitatea; Sveriges Lantbruksuniversitet; Aarhus Universitets Forskningsfond; University of Queensland; Vrije Universiteit Amsterdam; Institut National de Recherche pour l'Agriculture, l'Alimentation et l'Environnement; Norsk Institutt for Vannforskning; Jihočeská Univerzita v Českých Budějovicích; Universiteit van Amsterdam; Agence Nationale de la Recherche; Aarhus Universitet; Canadian Institute for Advanced Research; Ministry of Infrastructure and Water Management; Colorado State University","keywords":"Sampling (signal processing); Mass spectrometry; Environmental science; Passive sampling; Environmental chemistry; Chemistry; Computer science; Chromatography; Statistics; Mathematics; Calibration","score_opus":0.0235785649677717,"score_gpt":0.28273208864293287,"score_spread":0.25915352367516115,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3196272632","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33995706,0.0008240725,0.0387535,0.0005177419,0.00022332452,0.001539413,0.6070067,0.0020634132,0.009114704],"genre_scores_gemma":[0.21276301,0.00030994357,0.04249473,0.00033136384,0.000074359144,0.0025966112,0.73873854,0.000397182,0.0022941951],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99253464,0.0013898726,0.00065757206,0.0020961927,0.002928682,0.00039304065],"domain_scores_gemma":[0.9906866,0.0014702731,0.0013455851,0.0024572546,0.003694099,0.00034606986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005438385,0.0010597793,0.0013330202,0.00304738,0.0011111505,0.0012843319,0.0018971087,0.0012863948,0.0028297214],"category_scores_gemma":[0.0074253846,0.00035241543,0.0010599762,0.005245227,0.0008390793,0.00070440525,0.0022540027,0.0010683537,0.0025879077],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0075402674,0.0040498176,0.4455777,0.0075542526,0.0028354744,0.0019060128,0.0021017129,0.025661027,0.18282124,0.0042827744,0.130008,0.18566172],"study_design_scores_gemma":[0.00039341563,0.0012753232,0.5880317,0.00033297742,0.0006470745,0.00105462,0.0012152669,0.012709036,0.091044925,0.0046967575,0.29828855,0.00031043598],"about_ca_topic_score_codex":0.009277055,"about_ca_topic_score_gemma":0.012078055,"teacher_disagreement_score":0.009277055,"about_ca_system_score_codex":0.0014667127,"about_ca_system_score_gemma":0.0036181116,"threshold_uncertainty_score":0.028761268},"labels":[],"label_agreement":null},{"id":"W3197085852","doi":"10.1038/s41597-021-01026-2","title":"Author Correction: Open-access quantitative MRI data of the spinal cord and reproducibility across participants, sites and manufacturers","year":2021,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Medical Imaging and Analysis","field":"Engineering","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Philips (Canada); McGill University; Centre Hospitalier Universitaire de Sherbrooke; International Collaboration On Repair Discoveries; University of British Columbia; Université de Montréal; Université de Sherbrooke; Centre Hospitalier Universitaire Sainte-Justine; Polytechnique Montréal; Montreal Neurological Institute and Hospital; Mila - Quebec Artificial Intelligence Institute","funders":"National Center for Advancing Translational Sciences; National Institute of Biomedical Imaging and Bioengineering","keywords":"Reproducibility; Data science; Medicine; Computer science; Information retrieval; Statistics; Mathematics","score_opus":0.2917936382257923,"score_gpt":0.4651536342517678,"score_spread":0.17335999602597552,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3197085852","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00085661607,0.00078716886,0.004072758,0.052957576,0.92596114,0.00010806188,0.0053543793,0.0015330411,0.008369191],"genre_scores_gemma":[0.062476367,0.0061246566,0.03627155,0.09594704,0.19661233,0.0008826292,0.015164577,0.009450144,0.57707065],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98704404,0.0017742094,0.002418539,0.0015371006,0.006566232,0.0006598425],"domain_scores_gemma":[0.8762112,0.033272892,0.005601062,0.013119498,0.069618925,0.0021763847],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.008637753,0.0018981529,0.0016327221,0.004514147,0.003574055,0.003704142,0.0029250528,0.0054193,0.07979782],"category_scores_gemma":[0.17115384,0.0013087197,0.0012218603,0.0036045003,0.0026616936,0.0021765686,0.0024860485,0.007884913,0.04432876],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000060878258,0.000009451611,0.0001957651,0.00017400901,0.000014484697,0.00030781934,0.00006904459,0.00004188018,0.000081029204,0.0010961536,0.98862225,0.009327232],"study_design_scores_gemma":[0.000054125703,0.00002776058,0.0011942968,0.0006509207,0.0000621298,0.0011295088,0.00020788253,0.00039974946,0.0010448779,0.003530205,0.99163646,0.00006207916],"about_ca_topic_score_codex":0.013734671,"about_ca_topic_score_gemma":0.019290581,"teacher_disagreement_score":0.9913623,"about_ca_system_score_codex":0.0031697808,"about_ca_system_score_gemma":0.006574812,"threshold_uncertainty_score":0.2669503},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3202927809","doi":"10.1038/s41597-021-01006-6","title":"AusTraits, a curated plant trait database for the Australian flora","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":199,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université de Montréal","funders":"Australian Research Data Commons; NSW Department of Planning,Industry and Environment; Australian Research Council; Soochow University; Department of Biodiversity, Conservation and Attractions; Centre for Australian National Biodiversity Research; Department of Education and Training; Department of Environment, Land, Water and Planning, State Government of Victoria","keywords":"Taxon; Trait; Scope (computer science); Biology; Database; Ecology; Flora (microbiology); Field (mathematics); Taxonomic rank; Geography; Computer science","score_opus":0.16331794623119344,"score_gpt":0.3183892448547478,"score_spread":0.15507129862355434,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3202927809","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.051769495,0.0019128007,0.026385602,0.00029996422,0.00013604654,0.0002975548,0.8881979,0.0151036745,0.015896885],"genre_scores_gemma":[0.05750557,0.0010566101,0.051946342,0.00017232576,0.0000450122,0.000619907,0.87854135,0.0027282073,0.0073847077],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99902153,0.000112328766,0.00020060531,0.00029892696,0.00029341932,0.00007324493],"domain_scores_gemma":[0.99748933,0.000479675,0.0005753356,0.00060455356,0.0005789484,0.00027220533],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012962266,0.00093161553,0.001051143,0.008406194,0.00093968684,0.0017889225,0.0013423951,0.0005297326,0.018538203],"category_scores_gemma":[0.005349631,0.00083639915,0.0007307941,0.010632569,0.00042521404,0.0022719728,0.002345104,0.001028039,0.008653001],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013721638,0.00023922278,0.050526675,0.015169007,0.00083595526,0.0014485782,0.009950758,0.0034141873,0.06582838,0.012479322,0.4249998,0.41373596],"study_design_scores_gemma":[0.00007066874,0.00007639736,0.21432775,0.0004846079,0.0001803834,0.00062359683,0.00037846097,0.002071437,0.002996015,0.004394832,0.7742777,0.00011818669],"about_ca_topic_score_codex":0.023003627,"about_ca_topic_score_gemma":0.03699773,"teacher_disagreement_score":0.023003627,"about_ca_system_score_codex":0.0012661623,"about_ca_system_score_gemma":0.0030587632,"threshold_uncertainty_score":0.062016487},"labels":[],"label_agreement":null},{"id":"W3203260001","doi":"10.1038/s41597-021-01031-5","title":"COVID Border Accountability Project, a hand-coded global database of border closures introduced during 2020","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"Nanovic Institute for European Studies; University of Notre Dame","keywords":"Timeline; Accountability; Audit; Categorization; Social media; Pandemic; Political science; Immigration; Coronavirus disease 2019 (COVID-19); Public relations; Database; Data science; Business; Geography; World Wide Web; Computer science; Accounting","score_opus":0.24236091395207404,"score_gpt":0.48592408357320527,"score_spread":0.24356316962113123,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3203260001","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017797092,0.00012581615,0.00055405486,0.00019420609,0.00005393598,0.00010414453,0.99438184,0.00019312969,0.0026131258],"genre_scores_gemma":[0.0055986233,0.00021078264,0.002980956,0.00010226267,0.000025113684,0.00057628675,0.9894008,0.000097269956,0.0010079291],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99711573,0.0005951634,0.00075031194,0.0006084898,0.00069368957,0.00023659156],"domain_scores_gemma":[0.9815059,0.0051761935,0.005264038,0.0028160973,0.004168502,0.0010692002],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003699348,0.00088178087,0.0008460637,0.008187434,0.0007699057,0.002205393,0.0017814084,0.0011063031,0.022616232],"category_scores_gemma":[0.020141983,0.00064845727,0.00079486065,0.011194936,0.0004223023,0.0018592167,0.0027075808,0.0017050772,0.009988383],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000235297,0.00006356113,0.033028994,0.0020212855,0.0001163276,0.000100821206,0.0003615375,0.00092100893,0.0002781568,0.0037992017,0.9392859,0.019787908],"study_design_scores_gemma":[0.00018828266,0.00006644076,0.09614593,0.0016241332,0.00009962922,0.00020090984,0.0011138398,0.0018811765,0.0009081813,0.0026673353,0.89501303,0.000091058755],"about_ca_topic_score_codex":0.045269568,"about_ca_topic_score_gemma":0.05004641,"teacher_disagreement_score":0.045269568,"about_ca_system_score_codex":0.0019747815,"about_ca_system_score_gemma":0.0059558693,"threshold_uncertainty_score":0.09001213},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3203937348","doi":"10.1038/s41597-021-01033-3","title":"The “Narratives” fMRI dataset for evaluating models of naturalistic language comprehension","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neurobiology of Language and Bilingualism","field":"Neuroscience","cited_by":153,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Children's Hospital; University of British Columbia; McGill University; Montreal Neurological Institute and Hospital","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute of Mental Health; Advanced Research Projects Agency; U.S. Department of Health and Human Services; National Institutes of Health; National Institute of Biomedical Imaging and Bioengineering; Intel Corporation; Defense Advanced Research Projects Agency; U.S. Department of Defense","keywords":"Narrative; Computer science; Metadata; Comprehension; Neuroimaging; Variety (cybernetics); Benchmark (surveying); Data collection; Natural language processing; Psychology; Artificial intelligence; Linguistics; World Wide Web; Neuroscience","score_opus":0.24730925129721867,"score_gpt":0.4295171565604403,"score_spread":0.18220790526322164,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3203937348","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2624947,0.002619112,0.038512554,0.0019890817,0.00043990012,0.0010303962,0.66741455,0.010412272,0.015087459],"genre_scores_gemma":[0.14709142,0.00049607176,0.03914541,0.00043608955,0.00012886198,0.0016786548,0.805869,0.0011181182,0.004036319],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994617,0.0002097153,0.00005574099,0.00013652045,0.00009499938,0.000041398387],"domain_scores_gemma":[0.998226,0.00082476443,0.00013507437,0.0004540312,0.00022519127,0.00013506123],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015927152,0.0012593889,0.0004213242,0.0009778963,0.0006203449,0.0006794954,0.0014264841,0.0010908216,0.0072914935],"category_scores_gemma":[0.007017462,0.00030936452,0.000874189,0.0006747083,0.0005960677,0.00069937087,0.0012559646,0.0010448218,0.0051292717],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002284026,0.001272724,0.03560604,0.0039312975,0.0011894471,0.0026496898,0.0024461749,0.019270465,0.029743807,0.006697807,0.7829411,0.11196747],"study_design_scores_gemma":[0.0017514193,0.0017865353,0.14940299,0.0009144583,0.0008821567,0.009057369,0.002635229,0.08200864,0.03404682,0.03611228,0.68096036,0.00044175464],"about_ca_topic_score_codex":0.005903044,"about_ca_topic_score_gemma":0.017700695,"teacher_disagreement_score":0.0072914935,"about_ca_system_score_codex":0.00048802156,"about_ca_system_score_gemma":0.0009722618,"threshold_uncertainty_score":0.024392486},"labels":[],"label_agreement":null},{"id":"W3205684707","doi":"10.1038/s41597-021-01050-2","title":"SMAP-HydroBlocks, a 30-m satellite-based soil moisture dataset for the conterminous US","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Moisture and Remote Sensing","field":"Environmental Science","cited_by":91,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Global Institute for Water Security; University of Saskatchewan","funders":"National Oceanic and Atmospheric Administration; U.S. Department of Commerce; National Aeronautics and Space Administration","keywords":"Environmental science; Satellite; Water content; Remote sensing; Spatial variability; Spatial ecology; Atmospheric sciences; Climatology; Geography; Geology","score_opus":0.03861151678310767,"score_gpt":0.26925541692380023,"score_spread":0.23064390014069255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3205684707","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07671563,0.0002728305,0.002740231,0.00025419955,0.00008537215,0.00010752622,0.91272,0.0035075643,0.0035966402],"genre_scores_gemma":[0.08090778,0.00010714126,0.0062358375,0.00008774354,0.00003522311,0.00017282878,0.9114042,0.0002572436,0.0007920519],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9997733,0.00003245398,0.000020866919,0.000069875074,0.000074146905,0.00002934149],"domain_scores_gemma":[0.99951553,0.00005739055,0.000081384496,0.000113045506,0.0001658612,0.0000668428],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036495764,0.0006642143,0.00045011236,0.0011252108,0.0003406972,0.00045813338,0.0010133055,0.0006006099,0.0047409004],"category_scores_gemma":[0.0011804383,0.00028321779,0.0004855396,0.0021518664,0.00026746854,0.0007540657,0.0007578373,0.00067033235,0.002694933],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006940174,0.0004658268,0.13654988,0.0011522779,0.00067349593,0.00045913874,0.0004488774,0.029505536,0.015890133,0.0029238567,0.7361752,0.07506178],"study_design_scores_gemma":[0.0011162088,0.00015253855,0.4109744,0.00024231762,0.00019332119,0.00025203355,0.0004850188,0.112458594,0.010859356,0.0040818886,0.45899954,0.00018469915],"about_ca_topic_score_codex":0.044587888,"about_ca_topic_score_gemma":0.0689412,"teacher_disagreement_score":0.044587888,"about_ca_system_score_codex":0.00044800073,"about_ca_system_score_gemma":0.0010520127,"threshold_uncertainty_score":0.08865672},"labels":[],"label_agreement":null},{"id":"W3208039415","doi":"10.1038/s41597-021-01069-5","title":"A curated diverse molecular database of blood-brain barrier permeability with chemical descriptors","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":121,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Blood–brain barrier; Computer science; Categorical variable; Permeability (electromagnetism); Benchmark (surveying); Computational biology; Artificial intelligence; Chemistry; Neuroscience; Central nervous system; Biology; Machine learning; Membrane; Biochemistry","score_opus":0.04931558276298247,"score_gpt":0.3073137946991274,"score_spread":0.2579982119361449,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3208039415","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0072322576,0.0070992727,0.0027776996,0.0002782919,0.000078985206,0.0001455695,0.97749305,0.001979124,0.0029156897],"genre_scores_gemma":[0.011373044,0.0022721316,0.003939848,0.0001739266,0.000021389,0.00024468964,0.9813996,0.00009289027,0.0004825651],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986181,0.00018975828,0.00019663326,0.0004231656,0.00044919318,0.00012314493],"domain_scores_gemma":[0.9983364,0.00049922906,0.00029140385,0.00036340058,0.00032387546,0.00018571965],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011190163,0.0020908266,0.0034174044,0.0076428605,0.0008969923,0.0022127125,0.0034875283,0.0029362596,0.017407982],"category_scores_gemma":[0.004727483,0.000668523,0.0020044232,0.011478164,0.0006428509,0.0018247275,0.0018884481,0.0021465989,0.011130942],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023561837,0.00085526856,0.016309083,0.03577653,0.001689781,0.0016005348,0.00016796582,0.022136834,0.02758638,0.010143803,0.7894083,0.09196937],"study_design_scores_gemma":[0.0009997635,0.0006484027,0.019300751,0.0016115612,0.0010555937,0.0017633846,0.00014910381,0.016720174,0.01706696,0.008382553,0.9320542,0.0002475733],"about_ca_topic_score_codex":0.009920225,"about_ca_topic_score_gemma":0.016113954,"teacher_disagreement_score":0.017407982,"about_ca_system_score_codex":0.0014854472,"about_ca_system_score_gemma":0.004244089,"threshold_uncertainty_score":0.058235526},"labels":[],"label_agreement":null},{"id":"W3208632558","doi":"10.1038/s41597-021-01067-7","title":"Geolocation of unpublished archaeological sites in the Peruvian Amazon","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Amazonian Archaeology and Ethnohistory","field":"Arts and Humanities","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; McGill University","funders":"University of Toronto; Japan Society for the Promotion of Science; Ministry of Education, Culture, Sports, Science and Technology; Social Sciences and Humanities Research Council of Canada; Government of Canada; McGill University","keywords":"Geolocation; Amazon rainforest; Geography; Archaeology; World Wide Web; Biology; Ecology; Computer science","score_opus":0.14368336207317597,"score_gpt":0.2884888730588309,"score_spread":0.14480551098565494,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3208632558","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9331581,0.0037847648,0.0015993762,0.00040981217,0.00003141526,0.00017966825,0.030258859,0.00011543486,0.030462604],"genre_scores_gemma":[0.95196867,0.0041739433,0.0057998034,0.0000813028,0.000048867474,0.000548789,0.031156627,0.000063268766,0.006158652],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9988538,0.00039316821,0.00015837571,0.00017607903,0.0003212522,0.00009732209],"domain_scores_gemma":[0.99533045,0.001374905,0.0011321196,0.0008473448,0.0011267674,0.000188499],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011406809,0.0002483655,0.00024880777,0.011305176,0.0014358822,0.0011890099,0.00062753994,0.00032160236,0.005193683],"category_scores_gemma":[0.0077259657,0.00028547438,0.00013192657,0.019387584,0.00084328686,0.0006277385,0.0019653414,0.00021398565,0.0011820175],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002983646,0.000118901364,0.5852389,0.0023961423,0.000112392576,0.0023003526,0.11117016,0.000488532,0.00774808,0.0034235427,0.014932772,0.27177188],"study_design_scores_gemma":[0.000015434804,0.000073736526,0.8726869,0.00043506938,0.000033279106,0.0014601192,0.022319868,0.00016136024,0.0006920418,0.0005448085,0.1015421,0.000035406716],"about_ca_topic_score_codex":0.016366636,"about_ca_topic_score_gemma":0.04520202,"teacher_disagreement_score":0.016366636,"about_ca_system_score_codex":0.0004641728,"about_ca_system_score_gemma":0.00084657135,"threshold_uncertainty_score":0.032542706},"labels":[],"label_agreement":null},{"id":"W3210441536","doi":"10.1038/s41597-021-01073-9","title":"Attributes of the food and physical activity built environments from the Southern Cone of Latin America","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Obesity, Physical Activity, Diet","field":"Medicine","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"International Development Research Centre","keywords":"Latin Americans; Limiting; Geography; Built environment; Incentive; Physical activity; Healthy food; Environmental health; Walkability; Active living; Environmental quality; Regional science; Environmental planning; Business; Political science; Ecology; Medicine; Economics; Engineering","score_opus":0.060731334286667396,"score_gpt":0.28170590909923926,"score_spread":0.22097457481257188,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3210441536","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8482577,0.0003789356,0.0009027316,0.00023261251,0.000015190156,0.00011781776,0.14299193,0.00015212692,0.00695101],"genre_scores_gemma":[0.7407207,0.00048044717,0.004121868,0.00010446122,0.00001790765,0.00065386033,0.25169492,0.00006948167,0.0021363045],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996681,0.00007498128,0.00003036474,0.00008585143,0.00006849675,0.00007214442],"domain_scores_gemma":[0.9993463,0.00010177104,0.0001350557,0.00009737602,0.00023208873,0.000087394495],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00033166385,0.0004454485,0.0004006071,0.0022242228,0.00050259655,0.0008851548,0.00047934128,0.0003085801,0.0023024334],"category_scores_gemma":[0.0013803008,0.00014489208,0.00043416198,0.0051148334,0.00025409847,0.00031649452,0.001144885,0.00028762553,0.0007816774],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022013042,0.00017770196,0.94662756,0.00026395303,0.00008016274,0.000386244,0.0019137779,0.0024602672,0.0011940569,0.000572394,0.022097887,0.024005938],"study_design_scores_gemma":[0.000033758395,0.00003754334,0.9529966,0.00013095792,0.000027017893,0.0002156269,0.00657067,0.0036921075,0.00028782035,0.00042625697,0.03555848,0.000023226046],"about_ca_topic_score_codex":0.14612573,"about_ca_topic_score_gemma":0.21102826,"teacher_disagreement_score":0.14612573,"about_ca_system_score_codex":0.000717352,"about_ca_system_score_gemma":0.0007272838,"threshold_uncertainty_score":0.29055035},"labels":[],"label_agreement":null},{"id":"W3215310298","doi":"10.1038/s41597-022-01638-2","title":"A multi-sensor human gait dataset captured through an optical system and inertial measurement units","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Gait Recognition and Analysis","field":"Engineering","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre for Interdisciplinary Research in Music Media and Technology","funders":"Fundação de Amparo à Pesquisa do Estado de São Paulo; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior; Centre for Interdisciplinary Research in Music Media and Technology","keywords":"Gait; Inertial frame of reference; Inertial measurement unit; Computer science; Units of measurement; Computer vision; Artificial intelligence; Remote sensing; Geography; Physical medicine and rehabilitation; Physics; Medicine","score_opus":0.17279967380175246,"score_gpt":0.29239765996487965,"score_spread":0.1195979861631272,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3215310298","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45673874,0.0012626597,0.08271959,0.00035270036,0.00038694005,0.002102848,0.44472066,0.002973406,0.008742471],"genre_scores_gemma":[0.468928,0.0006397384,0.06286533,0.00015758748,0.00011692243,0.0027393478,0.45808527,0.00018496232,0.0062828297],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9997278,0.000030990028,0.000051928633,0.00007066868,0.00008853548,0.000030079362],"domain_scores_gemma":[0.99957865,0.00003962043,0.000051872583,0.000082624436,0.00020907882,0.00003816668],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023625133,0.0007088394,0.00054188684,0.0017806479,0.00030761547,0.0002856778,0.00047134937,0.0005977013,0.004795381],"category_scores_gemma":[0.0006626564,0.00014894496,0.00037566057,0.0014921218,0.00015369277,0.00025640804,0.0005159857,0.00024279424,0.0020448316],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004174747,0.0026764043,0.09566047,0.0071968082,0.000918557,0.0025082678,0.00074667146,0.011681991,0.18571259,0.0017276787,0.18037874,0.5066172],"study_design_scores_gemma":[0.0003960343,0.0023320466,0.8295143,0.00052388874,0.00030610428,0.0044972235,0.0007760167,0.028075391,0.034716614,0.0019846635,0.09666056,0.00021707594],"about_ca_topic_score_codex":0.0023561297,"about_ca_topic_score_gemma":0.008007876,"teacher_disagreement_score":0.004795381,"about_ca_system_score_codex":0.0001465176,"about_ca_system_score_gemma":0.000412636,"threshold_uncertainty_score":0.016042113},"labels":[],"label_agreement":null},{"id":"W3215526847","doi":"10.1038/s41597-022-01507-y","title":"The RETA Benchmark for Retinal Vascular Tree Analysis","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Retinal Imaging and Analysis","field":"Medicine","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Key Research and Development Program of China; Capital Medical University","keywords":"Segmentation; Annotation; Computer science; Benchmark (surveying); Tree (set theory); Artificial intelligence; Workflow; Software; Visualization; Task (project management); Pattern recognition (psychology); Database","score_opus":0.045044382642986484,"score_gpt":0.32508216363956427,"score_spread":0.2800377809965778,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3215526847","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3510357,0.010881114,0.3984558,0.0013264586,0.0012766942,0.001668439,0.06488731,0.14891407,0.02155434],"genre_scores_gemma":[0.41805947,0.001499476,0.44434157,0.0005125117,0.00023316222,0.0010958088,0.11567381,0.012264392,0.006319734],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99648076,0.00090820494,0.0004384941,0.0010128483,0.0009080092,0.00025171594],"domain_scores_gemma":[0.9948716,0.002031876,0.000441941,0.0012381594,0.0010986087,0.00031786432],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004853001,0.0025877196,0.001182147,0.00596047,0.0011982996,0.0035426868,0.0032974947,0.0025223203,0.0034842256],"category_scores_gemma":[0.012098556,0.00071032706,0.0021641613,0.0029521582,0.00073188654,0.001374608,0.0022457638,0.001520473,0.0022789899],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003192322,0.0012169362,0.023696698,0.0047435584,0.00222333,0.0019698963,0.0009513241,0.18790732,0.08063393,0.010982504,0.18796325,0.49451888],"study_design_scores_gemma":[0.00035890762,0.0009925443,0.028080866,0.00043692323,0.00045560044,0.0035976467,0.0004118013,0.7794384,0.086337745,0.014344758,0.085275896,0.0002689343],"about_ca_topic_score_codex":0.0076289643,"about_ca_topic_score_gemma":0.009547159,"teacher_disagreement_score":0.0076289643,"about_ca_system_score_codex":0.0011134975,"about_ca_system_score_gemma":0.0013191138,"threshold_uncertainty_score":0.025665402},"labels":[],"label_agreement":null},{"id":"W3216017632","doi":"10.1038/s41597-021-01088-2","title":"BSE49, a diverse, high-quality benchmark dataset of separation energies of chemical bonds","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Free Radicals and Antioxidants","field":"Chemistry","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia, Okanagan Campus; University of British Columbia","funders":"Agencia Estatal de Investigación; British Columbia Knowledge Development Fund; Natural Sciences and Engineering Research Council of Canada; University of British Columbia; Ministerio de Ciencia e Innovación; Western Canada Research Grid; Compute Canada","keywords":"Molecule; Covalent bond; Homolysis; Chemistry; Bond energy; Sextuple bond; Bond cleavage; Computational chemistry; Bond length; Radical; Bond order; Organic chemistry","score_opus":0.062073017165149375,"score_gpt":0.3402202235745776,"score_spread":0.27814720640942825,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3216017632","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.101677686,0.0066746306,0.012240663,0.0008884378,0.00034083505,0.00023930642,0.8624453,0.0040114042,0.011481774],"genre_scores_gemma":[0.07570528,0.0011032022,0.017395366,0.00027590583,0.000060597482,0.0003922379,0.9024851,0.00037423745,0.0022080971],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986744,0.00028904207,0.0000903877,0.0003368181,0.00048974255,0.00011966491],"domain_scores_gemma":[0.9981824,0.0007180763,0.00017183505,0.00039626058,0.00040894878,0.00012246652],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013013218,0.0016114333,0.0011581734,0.002422809,0.000950117,0.0011785052,0.0036645825,0.0028627045,0.0094742775],"category_scores_gemma":[0.0050911712,0.00040431245,0.0018547226,0.0041512377,0.0007698452,0.0011947466,0.0010713051,0.0017860978,0.004536168],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013196542,0.0012588038,0.01430623,0.007852686,0.0009997332,0.0005464416,0.00013837978,0.19083565,0.010258505,0.018840369,0.6933717,0.06027186],"study_design_scores_gemma":[0.0014400509,0.0009126686,0.030628804,0.0006609291,0.0002814725,0.0006949596,0.00033502694,0.2370795,0.024770875,0.035266865,0.6677116,0.00021722831],"about_ca_topic_score_codex":0.007528036,"about_ca_topic_score_gemma":0.011686289,"teacher_disagreement_score":0.0094742775,"about_ca_system_score_codex":0.0010924777,"about_ca_system_score_gemma":0.0014651987,"threshold_uncertainty_score":0.03169459},"labels":[],"label_agreement":null},{"id":"W3216508885","doi":"10.1038/s41597-021-01083-7","title":"Crash and disengagement data of autonomous vehicles on public roads in California","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Traffic and Road Safety","field":"Engineering","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Crash; Software deployment; Disengagement theory; Public domain; Computer science; Publication; Transport engineering; Suite; Computer security; Data science; Business; Engineering; Geography; Advertising; Political science; Law","score_opus":0.08127586703752618,"score_gpt":0.26722206499781825,"score_spread":0.18594619796029208,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3216508885","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5218272,0.0005952185,0.00073786924,0.00040786216,0.00010809045,0.000114662624,0.4629944,0.0008625883,0.01235209],"genre_scores_gemma":[0.33361697,0.00042343032,0.0017392759,0.00006117424,0.000035887413,0.00015113836,0.6604523,0.00009151334,0.0034283681],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9985386,0.000093813665,0.00014845241,0.00040892643,0.0006295514,0.00018062166],"domain_scores_gemma":[0.9965733,0.0004678907,0.000413407,0.0004998944,0.0017961521,0.00024935306],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000808498,0.00041095508,0.0003645181,0.0034724555,0.00078904856,0.0010340282,0.0009159641,0.0005510986,0.003365349],"category_scores_gemma":[0.0032517724,0.00022155407,0.0004203135,0.004417548,0.00037260694,0.00073019747,0.0009897641,0.00083042367,0.0015030646],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062562956,0.00049673376,0.6582258,0.0010924105,0.00030353115,0.0006432459,0.0017873506,0.013819738,0.0012126457,0.0024329165,0.26063463,0.058725383],"study_design_scores_gemma":[0.00004107213,0.0000923781,0.880198,0.00024302275,0.000088250075,0.00012603226,0.0029723498,0.008237671,0.0011671726,0.00044159102,0.10628937,0.00010316516],"about_ca_topic_score_codex":0.3969415,"about_ca_topic_score_gemma":0.37663165,"teacher_disagreement_score":0.3969415,"about_ca_system_score_codex":0.0016260132,"about_ca_system_score_gemma":0.0028291338,"threshold_uncertainty_score":0.7892621},"labels":[],"label_agreement":null},{"id":"W35241693","doi":"10.1038/s41597-022-01154-3","title":"Role Representations at the Core of School Principals' Practices in the Midst of School Reform in Quebec.","year":2002,"lang":"en","type":"article","venue":"Scientific Data","topic":"Education, sociology, and vocational training","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Core (optical fiber); Political science; Pedagogy; School choice; Mathematics education; Sociology; Public administration; Psychology; Law","score_opus":0.5458783606244283,"score_gpt":0.5288809495672291,"score_spread":0.016997411057199208,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W35241693","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92726225,0.0011730851,0.0045098104,0.016563784,0.00010225898,0.00017861414,0.004108404,0.00026972502,0.045831993],"genre_scores_gemma":[0.97793156,0.00015057763,0.0015214831,0.00023355395,0.0000050392787,0.000026263797,0.0006048908,0.00002202864,0.019504583],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.9982987,0.00060674525,0.000043033273,0.0002457408,0.00028225637,0.0005234855],"domain_scores_gemma":[0.99541205,0.0005951462,0.000552806,0.00028589676,0.0014544745,0.0016996035],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018751998,0.00019247069,0.00023434554,0.0012523488,0.0054194354,0.002368062,0.0014316598,0.0007264426,0.011379118],"category_scores_gemma":[0.005239351,0.00020017946,0.00019190225,0.0021996233,0.0019211601,0.0013364963,0.001602388,0.00088779646,0.0010862056],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027531615,0.0003349174,0.5123436,0.00020758917,0.000032155735,0.0007279693,0.10302609,0.0024715078,0.0019956157,0.03857442,0.0812949,0.25871596],"study_design_scores_gemma":[0.000018973637,0.000085755324,0.6504291,0.0002185194,0.000017386134,0.0001343516,0.13611616,0.0045066406,0.0010348922,0.0030533832,0.2043215,0.00006330508],"about_ca_topic_score_codex":0.91720414,"about_ca_topic_score_gemma":0.97468096,"teacher_disagreement_score":0.08279586,"about_ca_system_score_codex":0.043343402,"about_ca_system_score_gemma":0.034906067,"threshold_uncertainty_score":0.3144799},"labels":[],"label_agreement":null},{"id":"W4214595825","doi":"10.1038/s41597-022-01162-3","title":"Worldwide diversity of endophytic fungi and insects associated with dormant tree twigs","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Pathogens and Fungal Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Research, Development and Innovation Office; Bundesamt für Umwelt; Australian Centre for International Agricultural Research; Agentúra na Podporu Výskumu a Vývoja; Ministry of Agriculture of the People's Republic of China; Ministarstvo Prosvete, Nauke i Tehnološkog Razvoja; Department of Forestry, Fisheries and the Environment; Eidgenössische Technische Hochschule Zürich; Russian Foundation for Basic Research; Eesti Teadusfondi; Siberian Branch, Russian Academy of Sciences; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; McGill University; Génome Québec; Javna Agencija za Raziskovalno Dejavnost RS; Hungarian Scientific Research Fund; Uniwersytet Rolniczy im. Hugona Kołłątaja w Krakowie; Department of Agriculture, Environment and Rural Affairs, UK Government; Instituto Nacional de investigação Agrária e Veterinária; Agriculture and Agri-Food Canada; National Science Foundation","keywords":"Biology; Gymnosperm; Plant use of endophytic fungi in defense; Taxon; Ecology; Herbivore; Marine fungi; Biodiversity; Botany","score_opus":0.024484448356290343,"score_gpt":0.215701642575229,"score_spread":0.19121719421893865,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4214595825","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9922902,0.0011300498,0.00044077812,0.000014035384,0.0000022731556,0.000006197354,0.0055534695,0.000013130539,0.00054989156],"genre_scores_gemma":[0.98704195,0.0009294564,0.0013883772,0.000026640662,0.000005247659,0.00001311086,0.010238803,0.000007956873,0.00034843796],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99972016,0.00003556733,0.000033277644,0.00011989148,0.00004829929,0.000042842494],"domain_scores_gemma":[0.99916685,0.00012224942,0.00046071067,0.000073764335,0.00008330344,0.00009323939],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00025468812,0.00024552067,0.00023027718,0.0019227389,0.0002032267,0.0004783401,0.00013123274,0.00020061586,0.0010111795],"category_scores_gemma":[0.00040710365,0.00010476163,0.00025162785,0.0019358147,0.00020245096,0.00027832986,0.00036077452,0.0001804716,0.00027425378],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008975342,0.000018594264,0.9523238,0.00018607931,0.00012692447,0.00016419547,0.00049552013,0.0002186916,0.025833016,0.00008698411,0.000386227,0.020070283],"study_design_scores_gemma":[9.565049e-7,0.000019839941,0.9977241,0.000017164703,0.000016911325,0.00025170622,0.00021866176,0.000047548994,0.00064917654,0.00001813706,0.0010333826,0.0000025907113],"about_ca_topic_score_codex":0.0014883435,"about_ca_topic_score_gemma":0.0030636308,"teacher_disagreement_score":0.0019227389,"about_ca_system_score_codex":0.00008965591,"about_ca_system_score_gemma":0.000112271024,"threshold_uncertainty_score":0.003382802},"labels":[],"label_agreement":null},{"id":"W4214842975","doi":"10.1038/s41597-022-01152-5","title":"Data on the diets of Salish Sea harbour seals from DNA metabarcoding","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine animal studies overview","field":"Environmental Science","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; University of British Columbia; Vancouver Island University; Fisheries and Oceans Canada","funders":"Pacific Salmon Foundation; Washington State University","keywords":"Phoca; Harbour; Predation; Fishery; Harbor seal; Marine ecosystem; Trophic level; Biology; Apex predator; Ecosystem; Ecology; DNA barcoding; Environmental DNA; Marine conservation; Biodiversity; Geography","score_opus":0.15018950995247163,"score_gpt":0.2973172623125636,"score_spread":0.14712775236009198,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4214842975","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9161354,0.00040440174,0.0014375696,0.00003839547,0.000013691331,0.00006552263,0.079733975,0.0000448954,0.0021262746],"genre_scores_gemma":[0.7116993,0.001362713,0.013779931,0.00008387945,0.00005149701,0.0003438021,0.26871657,0.000046408768,0.003915938],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996369,0.00003961118,0.00007864765,0.00011092756,0.00009440463,0.00003950645],"domain_scores_gemma":[0.9983518,0.00019721268,0.0007773121,0.00021688947,0.00033222418,0.00012470271],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006128381,0.00029507358,0.0002481823,0.0039372263,0.00035134464,0.00033994418,0.00022890093,0.00019743296,0.0030211892],"category_scores_gemma":[0.0010600302,0.00019420926,0.00029547387,0.0033640473,0.00018765012,0.00024317374,0.0005539337,0.00016348934,0.0008753559],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003105465,0.00004548123,0.9091248,0.00045885332,0.00012682336,0.00036078066,0.0013358182,0.00027851836,0.039707426,0.00016402351,0.0026967458,0.045390196],"study_design_scores_gemma":[0.000004004486,0.00006516814,0.9920161,0.00002464636,0.000028093098,0.00015270246,0.0002455989,0.00006370669,0.0011353324,0.00003542607,0.006222226,0.0000070157503],"about_ca_topic_score_codex":0.010604145,"about_ca_topic_score_gemma":0.030615265,"teacher_disagreement_score":0.010604145,"about_ca_system_score_codex":0.00023574551,"about_ca_system_score_gemma":0.0005538368,"threshold_uncertainty_score":0.021084845},"labels":[],"label_agreement":null},{"id":"W4220673714","doi":"10.1038/s41597-022-01210-y","title":"Open access dataset of task-free hemodynamic activity in 4-month-old infants during sleep using fNIRS","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Non-Invasive Vital Sign Monitoring","field":"Engineering","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Ministerio de Economía y Competitividad; Eusko Jaurlaritza; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Task (project management); Hemodynamics; Sleep (system call); Computer science; Psychology; Medicine; Internal medicine","score_opus":0.06863746598377127,"score_gpt":0.3324490765443133,"score_spread":0.26381161056054203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220673714","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03804043,0.0006230177,0.0028026365,0.0002458008,0.00017410995,0.00017107111,0.954307,0.0016770264,0.0019588612],"genre_scores_gemma":[0.036754582,0.00020839873,0.003374476,0.00008863875,0.00006969454,0.00051068823,0.9576635,0.000121830395,0.0012082579],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9995454,0.00006492687,0.00006754224,0.00015054703,0.00009924867,0.00007228335],"domain_scores_gemma":[0.99859005,0.00033509504,0.0002078685,0.00033274427,0.00036229685,0.00017202477],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00079094234,0.00091984327,0.0008406282,0.001379182,0.00041990363,0.00070051313,0.0013535145,0.0013993131,0.0059315814],"category_scores_gemma":[0.0031343095,0.0002549663,0.0006937343,0.0012069741,0.0002991619,0.00031996684,0.0012619123,0.0006693374,0.006199344],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004378266,0.0009706519,0.10480108,0.005938217,0.0013253468,0.0033529468,0.00077060645,0.0042251935,0.030170837,0.0015438338,0.7394744,0.10304859],"study_design_scores_gemma":[0.0009006795,0.0007714867,0.48117408,0.0010597932,0.00045077826,0.0030739994,0.0007901271,0.008891321,0.012929432,0.0033090569,0.48635054,0.00029869901],"about_ca_topic_score_codex":0.0108747445,"about_ca_topic_score_gemma":0.028365789,"teacher_disagreement_score":0.0108747445,"about_ca_system_score_codex":0.00038998568,"about_ca_system_score_gemma":0.00087455433,"threshold_uncertainty_score":0.021622896},"labels":[],"label_agreement":null},{"id":"W4220828579","doi":"10.1038/s41597-022-01197-6","title":"An expert-curated global database of online newspaper articles on spiders and spider bites","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Animal and Plant Science Education","field":"Psychology","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital; McGill University; University of Toronto; Espace pour la vie","funders":"Science and Engineering Research Board; Helsingin Yliopisto","keywords":"Spider; Newspaper; Metadata; Social media; Data science; Database; World Wide Web; Computer science; Ecology; Biology; Sociology; Media studies","score_opus":0.12992224496705482,"score_gpt":0.39505248393556575,"score_spread":0.26513023896851096,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220828579","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01687385,0.0018781783,0.0014161335,0.00021785316,0.00009870007,0.00064025354,0.9720659,0.0006824504,0.0061265687],"genre_scores_gemma":[0.018996548,0.0018303314,0.01079441,0.00020085383,0.00016689714,0.0019333941,0.9628733,0.00028598314,0.0029181994],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9959235,0.0005798987,0.0012948901,0.00084365066,0.0010872738,0.00027070966],"domain_scores_gemma":[0.96060896,0.019539263,0.0067125866,0.002633601,0.008838284,0.0016673626],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032147325,0.0012897084,0.0014386086,0.0447074,0.0009773616,0.0028407478,0.0011085506,0.001194396,0.022732142],"category_scores_gemma":[0.021086253,0.0005590562,0.00071015884,0.033472173,0.0007383259,0.002110491,0.0023639784,0.0010264497,0.017471299],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011008964,0.00057739805,0.047792375,0.05039156,0.00050106255,0.0030936028,0.0065605002,0.00094934064,0.01306866,0.0032108475,0.64803815,0.22471562],"study_design_scores_gemma":[0.00013254462,0.00014093822,0.1788633,0.00406148,0.00043351238,0.0011934798,0.0025125556,0.00075124844,0.0032947673,0.0008156357,0.8076679,0.00013262834],"about_ca_topic_score_codex":0.005931899,"about_ca_topic_score_gemma":0.013431477,"teacher_disagreement_score":0.0447074,"about_ca_system_score_codex":0.0010000627,"about_ca_system_score_gemma":0.0039731413,"threshold_uncertainty_score":0.076046586},"labels":[],"label_agreement":null},{"id":"W4220979273","doi":"10.1038/s41597-022-01231-7","title":"Neurocognitive aging data release with behavioral, structural and multi-echo functional MRI measures","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":58,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; McGill University; Douglas Mental Health University Institute; McGill Genome Centre; University of Toronto; Toronto Metropolitan University; Montreal Neurological Institute and Hospital","funders":"National Center for Research Resources; Canadian Institutes of Health Research; U.S. Department of Health and Human Services; Government of Canada; National Institutes of Health; Natural Sciences and Engineering Research Council of Canada","keywords":"Socioemotional selectivity theory; Neurocognitive; Neuroimaging; Context (archaeology); Psychology; Cognition; Cognitive psychology; Developmental psychology; Neuroscience; Biology","score_opus":0.2409381386783846,"score_gpt":0.3359719223071308,"score_spread":0.0950337836287462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220979273","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008556765,0.00013068847,0.0035061212,0.0003218616,0.00009287028,0.00046319125,0.981106,0.0016586932,0.0041638464],"genre_scores_gemma":[0.008897182,0.00012649503,0.006588736,0.00019881588,0.000068392255,0.0028101876,0.97696555,0.0005229989,0.0038217765],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9986099,0.00019453645,0.00030222943,0.00035013317,0.00041542482,0.00012769623],"domain_scores_gemma":[0.9921961,0.0011605135,0.00086027925,0.0020922758,0.0031931535,0.0004976617],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033131451,0.0014284749,0.0011947533,0.002585767,0.00082090386,0.0012601143,0.0019113709,0.0013250735,0.027265042],"category_scores_gemma":[0.01300447,0.00068770006,0.0009205381,0.0028598842,0.00045257548,0.0010941108,0.0022074108,0.0011593562,0.026783029],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016716114,0.0002522365,0.03334453,0.0012785748,0.00038152267,0.0004653652,0.00043997623,0.00082731905,0.0045653423,0.0020293389,0.91384697,0.04089717],"study_design_scores_gemma":[0.0005903572,0.00024136926,0.10008592,0.00032738977,0.00020536716,0.0006051414,0.00022583453,0.00077315693,0.004418207,0.0036072584,0.8888081,0.00011198497],"about_ca_topic_score_codex":0.011995934,"about_ca_topic_score_gemma":0.023051353,"teacher_disagreement_score":0.027265042,"about_ca_system_score_codex":0.000756364,"about_ca_system_score_gemma":0.0020736668,"threshold_uncertainty_score":0.09121066},"labels":[],"label_agreement":null},{"id":"W4221074757","doi":"10.1038/s41597-022-01249-x","title":"‘Citizens’ Attitudes Under Covid19’, a cross-country panel survey of public opinion in 11 advanced democracies","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Région Occitanie Pyrénées-Méditerranée; European University Institute; Centre National de la Recherche Scientifique; Agence Française de Développement; UniCredit and Universities Foundation; Agence Nationale de la Recherche; Harvard Business School; Conseil Régional Aquitaine; Temple University; McGill University; Università Bocconi; World Bank Group","keywords":"Public opinion; Coronavirus disease 2019 (COVID-19); Pandemic; Panel survey; Political science; Survey research; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); 2019-20 coronavirus outbreak; Corporate governance; Public health; Survey data collection; Opinion survey; Cross country; Public relations; Economic growth; Demographic economics; Sociology; Opinion leadership; Socioeconomics; Politics; Economics; Medicine; Law; Virology","score_opus":0.6626786409163391,"score_gpt":0.5076800047267307,"score_spread":0.1549986361896084,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4221074757","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7377723,0.00030168178,0.0024192566,0.0013661874,0.00010688654,0.00040725607,0.24628879,0.000076047174,0.011261683],"genre_scores_gemma":[0.7028697,0.00030310615,0.0040902495,0.0009244727,0.0001480984,0.0016039048,0.28619725,0.00003658383,0.0038265216],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9987245,0.0006289166,0.00010395742,0.00017762906,0.00018661456,0.00017842924],"domain_scores_gemma":[0.99426466,0.0013155249,0.0021491244,0.0008130121,0.0006941392,0.00076363696],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018162497,0.00029991212,0.0002903742,0.0014528448,0.0008219378,0.0010497276,0.0005674114,0.00089143036,0.004044699],"category_scores_gemma":[0.0060138055,0.00024407778,0.00033492869,0.0028794652,0.00043346558,0.0006711707,0.0015823276,0.0012489061,0.0016717487],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021774729,0.00040746265,0.9001012,0.00016159413,0.0001902612,0.00016756452,0.002635397,0.00064473954,0.0005414623,0.002260547,0.07673061,0.015941365],"study_design_scores_gemma":[0.000044729615,0.00014001013,0.9595421,0.00006163903,0.00004303591,0.00014103197,0.0024522198,0.00051762635,0.00038647177,0.00045026402,0.03618908,0.000031709398],"about_ca_topic_score_codex":0.028131774,"about_ca_topic_score_gemma":0.039480776,"teacher_disagreement_score":0.028131774,"about_ca_system_score_codex":0.00065572356,"about_ca_system_score_gemma":0.0008353971,"threshold_uncertainty_score":0.055936098},"labels":[],"label_agreement":null},{"id":"W4221157599","doi":"10.1038/s41597-022-01836-y","title":"Multi-day dataset of forearm and wrist electromyogram for hand gesture recognition and biometrics","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Muscle activation and electromyography studies","field":"Engineering","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Biometrics; Computer science; Wrist; Electromyography; Usability; Artificial intelligence; Gesture recognition; Gesture; Pattern recognition (psychology); Wearable computer; Forearm; Speech recognition; Physical medicine and rehabilitation; Human–computer interaction; Medicine","score_opus":0.07645241778837612,"score_gpt":0.27372020277188547,"score_spread":0.19726778498350933,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4221157599","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.115470655,0.0038727599,0.023679007,0.000634528,0.0008309361,0.00097476837,0.8409423,0.005881743,0.0077132406],"genre_scores_gemma":[0.059367847,0.0006057228,0.012789829,0.00019191604,0.00014759025,0.001086788,0.92219436,0.0001507536,0.0034651991],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99876535,0.0001640267,0.00018273546,0.00041199988,0.00034411225,0.00013178925],"domain_scores_gemma":[0.99863166,0.00019957445,0.00015277087,0.0004127017,0.00042658017,0.00017680979],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010021711,0.0019200186,0.0014829554,0.0022359171,0.00063688675,0.000866461,0.0023780717,0.0019209498,0.0047692345],"category_scores_gemma":[0.0021524797,0.00030166196,0.001406996,0.0019948578,0.0004855426,0.00058963435,0.0017574121,0.0010313758,0.00885581],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00200105,0.0017988242,0.061386853,0.005396908,0.0011183916,0.002484809,0.00046298318,0.010398294,0.0343271,0.0014731239,0.631676,0.24747568],"study_design_scores_gemma":[0.00057369174,0.0013960107,0.45274246,0.00085205823,0.0005563899,0.0070555704,0.0009721312,0.037044406,0.02693789,0.003392945,0.46801803,0.00045838585],"about_ca_topic_score_codex":0.005713254,"about_ca_topic_score_gemma":0.016152881,"teacher_disagreement_score":0.005713254,"about_ca_system_score_codex":0.0004675461,"about_ca_system_score_gemma":0.0011282777,"threshold_uncertainty_score":0.015954673},"labels":[],"label_agreement":null},{"id":"W4225545880","doi":"10.1038/s41597-022-01216-6","title":"A comprehensive LFQ benchmark dataset on modern day acquisition strategies in proteomics","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Proteomics Techniques and Applications","field":"Chemistry","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Spinal Cord Injury BC","funders":"Vlaamse regering; Fonds Wetenschappelijk Onderzoek; Agence Nationale de la Recherche","keywords":"Computer science; Data acquisition; Benchmark (surveying); Data mining; Proteomics; Data science; Biology","score_opus":0.045270968687654976,"score_gpt":0.32205059355483034,"score_spread":0.2767796248671754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4225545880","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.042977743,0.0048180437,0.01636336,0.0008848231,0.00039748824,0.00048228638,0.9100295,0.017108222,0.0069385753],"genre_scores_gemma":[0.009684305,0.0004406646,0.012271734,0.0002270364,0.000031199008,0.00037053274,0.9758479,0.00050194195,0.00062462466],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99219257,0.0011770498,0.00084233464,0.0023889043,0.0027449408,0.00065425085],"domain_scores_gemma":[0.9937057,0.0015138595,0.00043322568,0.0019181723,0.0020921894,0.00033682634],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006867699,0.0033778946,0.001691936,0.004998242,0.002260172,0.0029493945,0.0042686784,0.0029717474,0.006534886],"category_scores_gemma":[0.009823054,0.0007219848,0.0019431618,0.00891515,0.0011636507,0.002561006,0.003570923,0.0021699085,0.01119644],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014842547,0.0008480361,0.01566339,0.004057079,0.00073073566,0.0006275797,0.00020780851,0.015202138,0.025764512,0.0030141226,0.85258025,0.07982001],"study_design_scores_gemma":[0.00084000785,0.00077425677,0.07588575,0.00075547124,0.00032577617,0.0019142462,0.0004400397,0.051478688,0.050284423,0.012244408,0.80464196,0.00041498747],"about_ca_topic_score_codex":0.009731618,"about_ca_topic_score_gemma":0.015266511,"teacher_disagreement_score":0.009731618,"about_ca_system_score_codex":0.0023458716,"about_ca_system_score_gemma":0.0033331648,"threshold_uncertainty_score":0.03632027},"labels":[],"label_agreement":null},{"id":"W4226138129","doi":"10.1038/s41597-022-01141-8","title":"A construction classification system database for understanding resource use in building construction","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Sustainable Building Design and Assessment","field":"Engineering","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; University of Toronto","keywords":"Interoperability; Computer science; Ontology; Field (mathematics); Resource (disambiguation); Database; Building material; Materials management; Data science; Construction engineering; Architectural engineering; Civil engineering; World Wide Web; Engineering; Operations management","score_opus":0.12537023880506995,"score_gpt":0.2868134783604212,"score_spread":0.16144323955535125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4226138129","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015577976,0.00029123228,0.031702098,0.00029682822,0.00009937628,0.0007676257,0.92361444,0.013423824,0.014226621],"genre_scores_gemma":[0.024093483,0.000275144,0.0489526,0.0001075461,0.00002504788,0.0011561535,0.92148703,0.0005337202,0.003369325],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99846774,0.00013804683,0.0004814482,0.00038135424,0.00040482308,0.00012658483],"domain_scores_gemma":[0.996956,0.0005363572,0.0004186447,0.0009091078,0.0009626775,0.00021723434],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011408003,0.0010225533,0.0008332708,0.007314849,0.00093395205,0.0021538844,0.0019347012,0.0012572363,0.017483097],"category_scores_gemma":[0.004468484,0.00048819187,0.0013420467,0.010141454,0.00030419856,0.0026912624,0.0017664495,0.0010658782,0.012465468],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006955334,0.00073528686,0.07420932,0.003234307,0.00030940227,0.0007841709,0.0014554389,0.0105358735,0.010704811,0.022204485,0.5997598,0.27537155],"study_design_scores_gemma":[0.00009890246,0.00008819356,0.077606864,0.00039583197,0.00009715106,0.0003436574,0.000976448,0.016772551,0.0064815977,0.007677158,0.8893354,0.00012624556],"about_ca_topic_score_codex":0.02489508,"about_ca_topic_score_gemma":0.033125762,"teacher_disagreement_score":0.02489508,"about_ca_system_score_codex":0.0018778256,"about_ca_system_score_gemma":0.0030363912,"threshold_uncertainty_score":0.05848676},"labels":[],"label_agreement":null},{"id":"W4235512955","doi":"10.1038/s41597-021-01044-0","title":"Author Correction: Open-access quantitative MRI data of the spinal cord and reproducibility across participants, sites and manufacturers","year":2021,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Medical Imaging and Analysis","field":"Engineering","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Philips (Canada); McGill University; Centre Hospitalier Universitaire de Sherbrooke; International Collaboration On Repair Discoveries; University of British Columbia; Université de Montréal; Université de Sherbrooke; Centre Hospitalier Universitaire Sainte-Justine; Polytechnique Montréal; Montreal Neurological Institute and Hospital; Mila - Quebec Artificial Intelligence Institute","funders":"National Center for Advancing Translational Sciences; National Institute of Biomedical Imaging and Bioengineering","keywords":"Reproducibility; Computer science; Data science; Medicine; Information retrieval; Statistics; Mathematics","score_opus":0.2917936382257923,"score_gpt":0.4651536342517678,"score_spread":0.17335999602597552,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4235512955","genre_codex":"editorial","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000426419,0.0007161611,0.002756506,0.05045336,0.93629736,0.0000706333,0.004058352,0.0012360774,0.003985053],"genre_scores_gemma":[0.044443898,0.006878805,0.03785105,0.1712564,0.2782177,0.001091596,0.014138253,0.010362875,0.43575948],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9801175,0.0030878696,0.0039472687,0.0022991346,0.009436373,0.0011117898],"domain_scores_gemma":[0.80903554,0.057362393,0.008669958,0.01717229,0.104592234,0.003167606],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.012411991,0.0022987735,0.0024240273,0.005515548,0.004316368,0.0047081243,0.003924303,0.008137273,0.070399545],"category_scores_gemma":[0.20592773,0.001862088,0.0015018228,0.0041808835,0.0032217598,0.0027221995,0.0033911401,0.010242581,0.03993739],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005868653,0.0000098985465,0.00015499607,0.00017700769,0.000017832155,0.00024295104,0.00006450331,0.000035836027,0.00006657404,0.0007588459,0.9914624,0.006950395],"study_design_scores_gemma":[0.00010063172,0.0000315321,0.0015567236,0.00082954654,0.000090567984,0.00088631746,0.00025833014,0.00044609894,0.0010346608,0.0037840956,0.990886,0.000095457544],"about_ca_topic_score_codex":0.01553612,"about_ca_topic_score_gemma":0.02429911,"teacher_disagreement_score":0.987588,"about_ca_system_score_codex":0.0037431493,"about_ca_system_score_gemma":0.007964319,"threshold_uncertainty_score":0.23550999},"labels":[],"label_agreement":null},{"id":"W4250125441","doi":"10.1038/s41597-020-0515-6","title":"Publisher Correction: A global database of Holocene paleotemperature records","year":2020,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Okanagan University College; University of British Columbia, Okanagan Campus; Bishop's University; Okanagan College; Geological Survey of Canada; University of Ottawa; University of New Brunswick; Center for Northern Studies; Concordia University; Université du Québec à Montréal; University of British Columbia; Fisheries and Oceans Canada; Mount Royal University; Université Laval; Natural Resources Canada","funders":"","keywords":"Holocene; Database; Geology; Data science; Oceanography; Computer science","score_opus":0.04864310150019533,"score_gpt":0.282948224771884,"score_spread":0.2343051232716887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4250125441","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00036733327,0.00086761377,0.00092718704,0.02976696,0.9244474,0.00005216909,0.025613226,0.00088220666,0.017075831],"genre_scores_gemma":[0.008894466,0.0057203867,0.008369798,0.03532883,0.1164309,0.0003721505,0.064524464,0.0039741504,0.7563849],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99659026,0.0002874994,0.0006370605,0.000382161,0.001881256,0.00022178698],"domain_scores_gemma":[0.9647996,0.005294263,0.0014266884,0.0030942715,0.024008892,0.0013763616],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032949317,0.0016240621,0.0013445595,0.0063191685,0.002674593,0.0054439926,0.002610643,0.002431823,0.14031246],"category_scores_gemma":[0.039095648,0.0008798858,0.0010525399,0.006324608,0.001018598,0.002571197,0.0020399643,0.005636609,0.09670686],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010932885,0.0000032349706,0.000060180148,0.00004353031,0.0000026546215,0.000043781154,0.0000081379285,0.000018433635,0.000018513372,0.00030125535,0.9958509,0.0036383914],"study_design_scores_gemma":[0.000016206539,0.0000066336474,0.00088010303,0.00017553865,0.00001192543,0.00009646669,0.00004585925,0.00006372271,0.00013106626,0.00048283854,0.9980749,0.000014846067],"about_ca_topic_score_codex":0.03681057,"about_ca_topic_score_gemma":0.053026076,"teacher_disagreement_score":0.14031246,"about_ca_system_score_codex":0.002612527,"about_ca_system_score_gemma":0.0068860534,"threshold_uncertainty_score":0.46939194},"labels":[],"label_agreement":null},{"id":"W4280533865","doi":"10.1038/s41597-022-01330-5","title":"A benchmark dataset for Hydrogen Combustion","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"Division of Chemistry; Natural Sciences and Engineering Research Council of Canada; Office of Science; National Energy Research Scientific Computing Center; China Scholarship Council; U.S. Department of Energy; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Potential energy surface; Combustion; Ab initio; Benchmark (surveying); Hydrogen; Reaction coordinate; Chemistry; Potential energy; Density functional theory; Computer science; Statistical physics; Computational chemistry; Physics; Atomic physics; Physical chemistry","score_opus":0.0528722733954892,"score_gpt":0.3192898238311557,"score_spread":0.26641755043566645,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4280533865","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030001769,0.0026209846,0.002721865,0.00083661,0.00063933036,0.0002036159,0.9516748,0.0033020522,0.007999008],"genre_scores_gemma":[0.017908819,0.00045423603,0.0032476983,0.00016868589,0.000051943523,0.00012263922,0.9758473,0.00010775345,0.002090989],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99892163,0.000134931,0.00009776575,0.00030082595,0.00038165005,0.00016325142],"domain_scores_gemma":[0.9990325,0.00016119528,0.00007125055,0.0002466498,0.0003967677,0.000091640344],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009450439,0.0019876012,0.0012976559,0.0024646183,0.0009915021,0.0011809089,0.0028440251,0.0025856786,0.0093373945],"category_scores_gemma":[0.0025723958,0.00027078646,0.0015210703,0.0039036514,0.0004946374,0.0011667286,0.0012050844,0.0015700595,0.010496444],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00069779705,0.00059087115,0.011146581,0.0014924617,0.00033480828,0.00039966236,0.00004853548,0.022051292,0.0031132253,0.0028995739,0.9143667,0.042858485],"study_design_scores_gemma":[0.0006229905,0.00047191183,0.034713257,0.0004928343,0.00019222795,0.00077322824,0.00041421747,0.10664952,0.015036138,0.009332354,0.83110064,0.00020064709],"about_ca_topic_score_codex":0.024208669,"about_ca_topic_score_gemma":0.04512797,"teacher_disagreement_score":0.024208669,"about_ca_system_score_codex":0.001588976,"about_ca_system_score_gemma":0.001605738,"threshold_uncertainty_score":0.04813552},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4280538994","doi":"10.1038/s41597-022-01326-1","title":"A global inventory of animal diversity measured in different grazing treatments","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Rangeland Management and Livestock Ecology","field":"Environmental Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; University of Toronto; University of Alberta","funders":"Killam Trusts; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Biodiversity; Grazing; Ecosystem; Ecology; Domestication; Conservation grazing; Geography; Terrestrial ecosystem; Biology; Environmental resource management; Agroforestry; Environmental science","score_opus":0.05652990885548431,"score_gpt":0.25401390766263576,"score_spread":0.19748399880715145,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4280538994","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20485105,0.005306542,0.0025813296,0.00013651396,0.000062221014,0.00019243386,0.7769429,0.00035909217,0.009567963],"genre_scores_gemma":[0.26402935,0.003913809,0.012248056,0.00018569353,0.00013386067,0.0009149653,0.71598214,0.0001511676,0.0024409015],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99835724,0.0002490221,0.00049032294,0.00047634763,0.00030764967,0.00011938312],"domain_scores_gemma":[0.9910416,0.0014722834,0.0041593635,0.0007400219,0.0020028008,0.0005838872],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013190937,0.00075926055,0.00076269754,0.02123954,0.0003165923,0.00088997866,0.0006448615,0.00038073873,0.00466344],"category_scores_gemma":[0.0034125615,0.00032719134,0.0006924823,0.02455182,0.00029921412,0.001182651,0.0011610609,0.0003948806,0.0019167867],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025701598,0.000101291174,0.8565865,0.0046888473,0.0015314748,0.00023602167,0.0015308578,0.0008405277,0.0051465444,0.0008719563,0.03571623,0.09249277],"study_design_scores_gemma":[0.000013357583,0.00006251764,0.9648448,0.0002353117,0.00017768054,0.00021187073,0.00046303423,0.00017987359,0.0004924428,0.00018432418,0.033114575,0.00002008788],"about_ca_topic_score_codex":0.0075887274,"about_ca_topic_score_gemma":0.011220952,"teacher_disagreement_score":0.02123954,"about_ca_system_score_codex":0.0004860517,"about_ca_system_score_gemma":0.0007562703,"threshold_uncertainty_score":0.015600741},"labels":[],"label_agreement":null},{"id":"W4280604425","doi":"10.1038/s41597-022-01334-1","title":"Recommendations for repositories and scientific gateways from a neuroscience perspective","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Research Data Management Practices","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"Karolinska Institutet","keywords":"Perspective (graphical); Gateway (web page); Set (abstract data type); Service (business); World Wide Web; Key (lock); Data science; Computer science; Neuroscience; Psychology; Business","score_opus":0.25733234120299003,"score_gpt":0.4056090017322596,"score_spread":0.14827666052926958,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4280604425","genre_codex":"commentary","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0073643397,0.015279792,0.110911116,0.64403564,0.01870172,0.0073872916,0.0126970075,0.013979843,0.16964334],"genre_scores_gemma":[0.020909963,0.029143747,0.64445263,0.11841707,0.0042146295,0.0061687753,0.029016824,0.0037861303,0.14389014],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.96514344,0.01051926,0.0071413047,0.0014461466,0.012775139,0.002974704],"domain_scores_gemma":[0.7758641,0.059670996,0.015472923,0.014711296,0.10391056,0.030370098],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.043803338,0.0016116097,0.0013272713,0.010785239,0.0040902896,0.015637534,0.006161854,0.020200424,0.040376984],"category_scores_gemma":[0.15738948,0.0016610428,0.0022973095,0.015443347,0.0025966947,0.024213575,0.006000734,0.00809559,0.033515133],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007553681,0.00027259058,0.001870923,0.0017343066,0.000031370273,0.0003873346,0.0007257494,0.0006994456,0.001056551,0.017353725,0.8621346,0.11365798],"study_design_scores_gemma":[0.00007872957,0.00004773769,0.001292279,0.0032245633,0.000049569615,0.00019815199,0.0014015528,0.0005835336,0.00056857785,0.010013625,0.9824516,0.0000901218],"about_ca_topic_score_codex":0.023579875,"about_ca_topic_score_gemma":0.052190404,"teacher_disagreement_score":0.9843625,"about_ca_system_score_codex":0.005049588,"about_ca_system_score_gemma":0.031081332,"threshold_uncertainty_score":0.23165697},"labels":[],"label_agreement":null},{"id":"W4281486635","doi":"10.1038/s41597-022-01338-x","title":"MusMorph, a database of standardized mouse morphology data for morphometric meta-analyses","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Morphological variations and asymmetry","field":"Mathematics","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Alberta Children's Hospital; University of Alberta; Alberta Bone and Joint Health Institute; University of Calgary","funders":"National Institute of Dental and Craniofacial Research; National Institute of Diabetes and Digestive and Kidney Diseases; National Heart, Lung, and Blood Institute; Alberta Children's Hospital Foundation; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; National Human Genome Research Institute; Alberta Children's Hospital Research Institute; Alberta Innovates - Health Solutions; National Institutes of Health; U.S. Department of Health and Human Services; Children's Hospital Foundation; Government of Canada; Alberta Innovates; University of Calgary","keywords":"Workflow; Computer science; Context (archaeology); Scripting language; Morphometrics; Pipeline (software); File format; Biology; Database; Artificial intelligence; Information retrieval","score_opus":0.5703114693802749,"score_gpt":0.45797957275730233,"score_spread":0.11233189662297255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281486635","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009266194,0.0014180378,0.07936478,0.0002340366,0.00020631356,0.0002830307,0.83502954,0.0679505,0.0062476913],"genre_scores_gemma":[0.015863363,0.0009927901,0.08581578,0.0002637233,0.000051773834,0.0019028592,0.87881166,0.013574872,0.0027231607],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977659,0.0002556354,0.0003440987,0.00067367684,0.000842994,0.00011771081],"domain_scores_gemma":[0.99689925,0.0009098011,0.00055734307,0.000981784,0.00040835087,0.00024345648],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034098413,0.002456136,0.0018340409,0.007811878,0.00092653796,0.0024390751,0.003543443,0.0014819545,0.025815846],"category_scores_gemma":[0.007888543,0.0014975279,0.0020175127,0.0044722,0.00061769295,0.0019514719,0.0039189593,0.0020887712,0.019627651],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010890984,0.00028931824,0.029394224,0.008668322,0.0023886107,0.0011335555,0.0008313292,0.009125194,0.065430716,0.018040359,0.7113538,0.15225552],"study_design_scores_gemma":[0.0004494104,0.0002539159,0.04078067,0.001025628,0.0011240254,0.001873474,0.00022477815,0.009578931,0.031561464,0.023072867,0.88969827,0.00035656616],"about_ca_topic_score_codex":0.0024583277,"about_ca_topic_score_gemma":0.008533181,"teacher_disagreement_score":0.025815846,"about_ca_system_score_codex":0.0007913634,"about_ca_system_score_gemma":0.002422673,"threshold_uncertainty_score":0.0863626},"labels":[],"label_agreement":null},{"id":"W4281659215","doi":"10.1038/s41597-022-01314-5","title":"Indoor heat measurement data from low-income households in rural and urban South Asia","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate Change and Health Impacts","field":"Environmental Science","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Government College University, Lahore; Department for International Development; International Development Research Centre","keywords":"Geography; Socioeconomic status; Urban heat island; Heat stress; South asia; Socioeconomics; Rural area; Thermal comfort; Environmental science; Environmental health; Meteorology; Population; Medicine","score_opus":0.1739709056250032,"score_gpt":0.30670960178359713,"score_spread":0.13273869615859393,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281659215","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7345684,0.00017785565,0.0011497837,0.00023642741,0.000027809952,0.00010875136,0.25839186,0.00020224888,0.0051367916],"genre_scores_gemma":[0.646937,0.00031317392,0.0026638764,0.00014267204,0.000023637542,0.0004445643,0.347228,0.000042406202,0.0022046887],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996457,0.00006757485,0.000051614912,0.00009120348,0.00007799656,0.000065877546],"domain_scores_gemma":[0.9989177,0.00016316689,0.00021114518,0.00022171762,0.0003828136,0.00010345261],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00032878137,0.00034182923,0.0003859707,0.00081582874,0.00030931705,0.00045338145,0.0005647883,0.00035219343,0.0024904986],"category_scores_gemma":[0.00085862976,0.00017638788,0.00038666726,0.0027663761,0.00022125452,0.00035150294,0.00073902233,0.00034579524,0.0014561143],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043465663,0.00032848972,0.9181044,0.00092206197,0.00021680241,0.00033852417,0.001953426,0.0040525775,0.0033262142,0.0005531396,0.040915683,0.028853983],"study_design_scores_gemma":[0.000028544038,0.00005369589,0.9785715,0.00006838373,0.00004353866,0.00010540361,0.0023918636,0.0020693883,0.0012087355,0.0001384107,0.0152966315,0.000023920848],"about_ca_topic_score_codex":0.04746629,"about_ca_topic_score_gemma":0.060459644,"teacher_disagreement_score":0.04746629,"about_ca_system_score_codex":0.00039538328,"about_ca_system_score_gemma":0.0006382016,"threshold_uncertainty_score":0.09437996},"labels":[],"label_agreement":null},{"id":"W4281704256","doi":"10.1038/s41597-022-01342-1","title":"Downburst-like experimental impinging jet measurements at the WindEEE Dome","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Wind and Air Flow Studies","field":"Environmental Science","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ontario Tech University; McGill University; Wind Energy Institute of Canada; Western University","funders":"European Commission","keywords":"Thunderstorm; Jet (fluid); Meteorology; Context (archaeology); Geology; Dome (geology); Outflow; Wind tunnel; Environmental science; Physics; Mechanics","score_opus":0.07372112800546714,"score_gpt":0.2776835011676654,"score_spread":0.20396237316219823,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281704256","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9932199,0.000101038866,0.0012095144,0.000014909261,0.000017782433,0.00005564024,0.00418128,0.00013405737,0.0010659982],"genre_scores_gemma":[0.98263323,0.0001374091,0.0033038717,0.000031112126,0.00002021464,0.00009234681,0.013125141,0.000054749482,0.000601934],"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9995447,0.000031645162,0.000022343767,0.00012454417,0.0001840856,0.00009259769],"domain_scores_gemma":[0.99930775,0.00012435149,0.000057328518,0.00011190198,0.00023974999,0.00015893622],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051212095,0.00040985394,0.0004368084,0.00088623376,0.000671472,0.0004405559,0.00036204467,0.00041291502,0.00070673996],"category_scores_gemma":[0.00079443166,0.00018410273,0.0002250652,0.0007224256,0.00033772466,0.00028663292,0.00049431244,0.00042595138,0.00029784229],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018945886,0.0016160612,0.5586938,0.00055827165,0.00028873366,0.0015844743,0.0021393576,0.013673477,0.3575907,0.00048376073,0.004722032,0.05675463],"study_design_scores_gemma":[0.00008239263,0.00054073095,0.96406907,0.00003372923,0.00005240072,0.00021367802,0.00049130764,0.0061816466,0.023374846,0.000073064635,0.004841477,0.000045663768],"about_ca_topic_score_codex":0.014754405,"about_ca_topic_score_gemma":0.053357948,"teacher_disagreement_score":0.014754405,"about_ca_system_score_codex":0.00031719956,"about_ca_system_score_gemma":0.0004892456,"threshold_uncertainty_score":0.029337049},"labels":[],"label_agreement":null},{"id":"W4281713424","doi":"10.1038/s41597-022-01371-w","title":"Voluntary consensus based geospatial data standards for the global illegal trade in wild fauna and flora","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Wildlife Ecology and Conservation","field":"Environmental Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Esri (Canada)","funders":"Division of Civil, Mechanical and Manufacturing Innovation; College of Engineering, Michigan State University; Michigan State University; National Science Foundation","keywords":"Geospatial analysis; Wildlife; Citizen journalism; Data discovery; Data science; Citizen science; Environmental resource management; Geography; Metadata; Computer science; Ecology; World Wide Web; Biology; Remote sensing; Environmental science","score_opus":0.0375187251057688,"score_gpt":0.28029885888343137,"score_spread":0.24278013377766255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281713424","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023614194,0.0013927119,0.66323173,0.037170142,0.0056201485,0.013653444,0.039476626,0.0050630034,0.21077798],"genre_scores_gemma":[0.1956981,0.0016183093,0.65237844,0.009718434,0.00075342075,0.030148974,0.080613114,0.0027458072,0.026325395],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.8060789,0.08481475,0.04210346,0.009642491,0.052434716,0.0049256333],"domain_scores_gemma":[0.5627776,0.11141388,0.020210009,0.12700106,0.17153661,0.0070607904],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.17469387,0.0011068175,0.0012776538,0.011169696,0.004728201,0.013018941,0.0067706583,0.005616464,0.008836795],"category_scores_gemma":[0.2827657,0.0014409092,0.0018313653,0.013078316,0.006727427,0.009039528,0.012559441,0.008153996,0.0064358427],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040867177,0.00045131712,0.01667291,0.0026755305,0.00019253028,0.00048913545,0.013538656,0.006750027,0.0069811395,0.50250864,0.24000823,0.20932327],"study_design_scores_gemma":[0.00009927537,0.000114062204,0.011531262,0.0024647552,0.00006570412,0.0002700644,0.0046385,0.0044208225,0.0039104945,0.11409672,0.8581898,0.00019855294],"about_ca_topic_score_codex":0.022067362,"about_ca_topic_score_gemma":0.016649816,"teacher_disagreement_score":0.17469387,"about_ca_system_score_codex":0.008004724,"about_ca_system_score_gemma":0.042635076,"threshold_uncertainty_score":0.9238805},"labels":[],"label_agreement":null},{"id":"W4281730041","doi":"10.1038/s41597-022-01396-1","title":"A global database of woody tissue carbon concentrations","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Forest Ecology and Biodiversity Studies","field":"Agricultural and Biological Sciences","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital; University of Toronto","funders":"Centre for Global Change Science, University of Toronto; U.S. Forest Service; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Northern Research Station; University of Toronto; University of Toronto Scarborough; Impact Fund; U.S. Department of Agriculture","keywords":"Woody plant; Biome; Database; Environmental science; Coarse woody debris; Biology; Ecology; Ecosystem; Computer science","score_opus":0.05071179355147193,"score_gpt":0.2554810614561086,"score_spread":0.20476926790463668,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281730041","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030449504,0.0024266057,0.0045893355,0.00005542049,0.00003371652,0.00006408371,0.95729375,0.0009459242,0.004141659],"genre_scores_gemma":[0.05013526,0.0013357257,0.010090051,0.000060254948,0.000047705562,0.00021405265,0.93711066,0.00019169247,0.0008145652],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986848,0.00010628337,0.0003054614,0.00046677017,0.00034335957,0.00009332898],"domain_scores_gemma":[0.99151987,0.001499497,0.0022507068,0.0013497063,0.0028850138,0.0004952585],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014647961,0.0009716715,0.0009211382,0.01704888,0.00036443718,0.0015905345,0.0011816274,0.0006835414,0.0050130235],"category_scores_gemma":[0.004746596,0.00041918512,0.00048413465,0.02253809,0.00026865696,0.0018078767,0.0009837417,0.00070367666,0.005148213],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010523323,0.00022442084,0.38299918,0.008205193,0.0010521964,0.00071600545,0.00073709706,0.0112724425,0.024782222,0.007367555,0.21505037,0.34654105],"study_design_scores_gemma":[0.000107318476,0.000117877746,0.37476626,0.0007723446,0.0003291451,0.00072210253,0.0002732055,0.0036307536,0.009795951,0.002511505,0.60681,0.00016355797],"about_ca_topic_score_codex":0.007468313,"about_ca_topic_score_gemma":0.0053503616,"teacher_disagreement_score":0.01704888,"about_ca_system_score_codex":0.0005582152,"about_ca_system_score_gemma":0.0015657069,"threshold_uncertainty_score":0.016770244},"labels":[],"label_agreement":null},{"id":"W4281760743","doi":"10.1038/s41597-022-01381-8","title":"CAN-SAR: A database of Canadian species at risk information","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Guelph; Carleton University; Environment and Climate Change Canada","funders":"Environment and Climate Change Canada","keywords":"Listing (finance); Threatened species; Database; Computer science; Environmental resource management; Business; Ecology; Biology; Habitat; Environmental science","score_opus":0.05533889582137173,"score_gpt":0.23479111536652028,"score_spread":0.17945221954514856,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281760743","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002452229,0.00050992507,0.0025331154,0.00037170842,0.000049063572,0.0002374815,0.9647423,0.0018303181,0.027273826],"genre_scores_gemma":[0.010415216,0.000992708,0.01734811,0.00025990268,0.000026638305,0.00036703344,0.962577,0.0006827251,0.007330703],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981741,0.00013028577,0.00029556899,0.00023584507,0.0009230881,0.00024104089],"domain_scores_gemma":[0.97598183,0.0026112234,0.0018224254,0.0019271518,0.015444121,0.002213149],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002222138,0.0011850569,0.00078036083,0.022897713,0.0023801348,0.0034184274,0.0030106225,0.00097327994,0.05204585],"category_scores_gemma":[0.016286831,0.00067004806,0.0005633128,0.03353651,0.0007228856,0.0025767144,0.0021230602,0.0012145876,0.01670023],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016374476,0.000043900407,0.007248544,0.0018625762,0.00006405431,0.00019617147,0.0010148486,0.0009396459,0.0010231604,0.0058010723,0.90047574,0.08116641],"study_design_scores_gemma":[0.000030004112,0.000007772347,0.016393311,0.0005467471,0.00004483644,0.00007461997,0.00043467229,0.0006569367,0.000457763,0.0012036555,0.98004407,0.00010557632],"about_ca_topic_score_codex":0.92001474,"about_ca_topic_score_gemma":0.95600826,"teacher_disagreement_score":0.07998526,"about_ca_system_score_codex":0.019073829,"about_ca_system_score_gemma":0.06183346,"threshold_uncertainty_score":0.17411071},"labels":[],"label_agreement":null},{"id":"W4281769686","doi":"10.1038/s41597-022-01341-2","title":"Multi sequence average templates for aging and neurodegenerative disease populations","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Dementia and Cognitive Impairment Research","field":"Medicine","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; University of Alberta; McGill University","funders":"Consortium canadien en neurodégénérescence associée au vieillissement","keywords":"Dementia; Cognitive impairment; Template; Dementia with Lewy bodies; Atrophy; Vascular dementia; Cognition; Magnetic resonance imaging; Disease; Medicine; Neuroscience; Audiology; Psychology; Pathology; Computer science; Radiology","score_opus":0.23881487453547012,"score_gpt":0.423258379006209,"score_spread":0.18444350447073887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281769686","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039868988,0.0008258498,0.94904524,0.00017797106,0.00017627413,0.00020354088,0.0031739501,0.003818247,0.002709968],"genre_scores_gemma":[0.18767357,0.00071843877,0.79482126,0.00025822833,0.00012312144,0.0008309259,0.009877814,0.0019539362,0.003742765],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99937683,0.000147245,0.000097698234,0.00018868322,0.00013395296,0.000055678247],"domain_scores_gemma":[0.9981668,0.0006740623,0.00018166951,0.00039775137,0.00049490854,0.000084788946],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002060948,0.000720124,0.0006596379,0.0015487256,0.0005467974,0.0011619527,0.0009877707,0.00094021845,0.0070449174],"category_scores_gemma":[0.007380043,0.00046002408,0.0013550331,0.0014691135,0.00031940817,0.0009993879,0.0011539112,0.0007803329,0.0023514112],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011714367,0.00021399408,0.02252988,0.00042800957,0.00046635215,0.00081891654,0.00062820705,0.08962722,0.054616284,0.016136535,0.023764115,0.78959906],"study_design_scores_gemma":[0.00013221503,0.0006424909,0.04020528,0.00014782576,0.00045468364,0.004615493,0.00033451727,0.72697794,0.10048332,0.04779583,0.07799498,0.00021538079],"about_ca_topic_score_codex":0.003214453,"about_ca_topic_score_gemma":0.0055847135,"teacher_disagreement_score":0.0070449174,"about_ca_system_score_codex":0.00050890277,"about_ca_system_score_gemma":0.0011081605,"threshold_uncertainty_score":0.023567557},"labels":[],"label_agreement":null},{"id":"W4282836069","doi":"10.1038/s41597-022-01398-z","title":"Optimising the classification of feature-based attention in frequency-tagged electroencephalography data","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"EEG and Brain-Computer Interfaces","field":"Neuroscience","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research","funders":"Centre of Excellence for Integrative Brain Function, Australian Research Council; Medical Research Council; Department of Education and Training; National Health and Medical Research Council; Canadian Institute for Advanced Research","keywords":"Electroencephalography; Feature (linguistics); Computer science; Pattern recognition (psychology); Artificial intelligence; Psychology; Neuroscience; Linguistics","score_opus":0.1058437764238055,"score_gpt":0.3185147871910828,"score_spread":0.2126710107672773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282836069","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5849331,0.0024484042,0.3862338,0.0008009751,0.0003354704,0.0007056934,0.013752313,0.008030149,0.0027601128],"genre_scores_gemma":[0.70704305,0.00053285365,0.26100212,0.00020339419,0.00015939675,0.00052619266,0.028595252,0.00022787653,0.0017098988],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99909544,0.00017229708,0.00009960433,0.0003161461,0.00019275906,0.00012380647],"domain_scores_gemma":[0.99788684,0.000976363,0.0001796064,0.0003082246,0.00058168615,0.00006723637],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014981889,0.0010558678,0.0010209846,0.0019451234,0.0003468301,0.0010129532,0.0011224605,0.0010993943,0.0008422065],"category_scores_gemma":[0.0078084045,0.0001883115,0.0008863144,0.0012850674,0.00035343962,0.00079716695,0.00084515923,0.0007803493,0.0010204702],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008633508,0.0006759991,0.024806788,0.0006163139,0.00025990652,0.00045857963,0.0002916108,0.058549788,0.076989114,0.0011320941,0.019123781,0.8162327],"study_design_scores_gemma":[0.00013519733,0.00043482208,0.044999495,0.00007955993,0.0001229001,0.0007115779,0.00027398163,0.8837605,0.05302072,0.0061475197,0.010249551,0.00006428623],"about_ca_topic_score_codex":0.0050529004,"about_ca_topic_score_gemma":0.005753737,"teacher_disagreement_score":0.0050529004,"about_ca_system_score_codex":0.0005052824,"about_ca_system_score_gemma":0.00072976924,"threshold_uncertainty_score":0.010046959},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4282928084","doi":"10.1038/s41597-022-01386-3","title":"A longitudinal multi-scanner multimodal human neuroimaging dataset","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Centre for Addiction and Mental Health","funders":"National Institute of Mental Health; U.S. Department of Health and Human Services","keywords":"Neuroimaging; Scanner; Modalities; Human brain; Computer science; Diffusion MRI; Consistency (knowledge bases); Artificial intelligence; Medical physics; Psychology; Medicine; Neuroscience; Magnetic resonance imaging; Radiology","score_opus":0.21471255176186121,"score_gpt":0.36186154707270757,"score_spread":0.14714899531084635,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282928084","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03126082,0.0009025889,0.005825671,0.0010401185,0.00018091849,0.0002949454,0.9562364,0.0015977736,0.0026607944],"genre_scores_gemma":[0.025473736,0.0002892486,0.0053487658,0.00029445466,0.00007856966,0.0008364926,0.96502185,0.00018377643,0.0024731085],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99916804,0.00019264976,0.00009622291,0.0002862477,0.00015949749,0.000097393],"domain_scores_gemma":[0.99779654,0.00046151233,0.00023776178,0.00082919694,0.000505669,0.00016939291],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016601533,0.0011886859,0.0010345401,0.0015289532,0.0008440181,0.0011941832,0.002113836,0.0022019646,0.01337542],"category_scores_gemma":[0.0042305067,0.000560085,0.0012210017,0.0018640708,0.0005143944,0.00068488077,0.0015570378,0.0015244067,0.012884526],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016233757,0.00079466816,0.03292893,0.0015510551,0.0010695297,0.001667213,0.00033521056,0.005401814,0.005635179,0.0020758065,0.8898722,0.057044983],"study_design_scores_gemma":[0.0011645051,0.00080983597,0.17796001,0.0006812223,0.0006333554,0.0057723876,0.0005828993,0.010352464,0.005834279,0.010707263,0.785065,0.00043687646],"about_ca_topic_score_codex":0.012924743,"about_ca_topic_score_gemma":0.029689496,"teacher_disagreement_score":0.01337542,"about_ca_system_score_codex":0.0009489575,"about_ca_system_score_gemma":0.0014093562,"threshold_uncertainty_score":0.044745266},"labels":[],"label_agreement":null},{"id":"W4282932896","doi":"10.1038/s41597-022-01463-7","title":"Datasets for benchmarking antimicrobial resistance genes in bacterial metagenomic and whole genome sequencing","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University; Sunnybrook Health Science Centre; Public Health Agency of Canada; Sunnybrook Hospital; McMaster University","funders":"Biotechnology and Biological Sciences Research Council; Medical Research Council; Joint Programming Initiative on Antimicrobial Resistance; Canadian Institutes of Health Research; Quadram Institute Bioscience; Universiteit van Amsterdam; Government of Canada","keywords":"Metagenomics; Computational biology; Biology; Genome; Gene prediction; Genomics; Whole genome sequencing; Identification (biology); DNA sequencing; Gene; Genetics","score_opus":0.03591850040073827,"score_gpt":0.26020224917409196,"score_spread":0.2242837487733537,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282932896","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09655106,0.001467967,0.035131786,0.00096949324,0.00049036124,0.0010170358,0.84789574,0.009637502,0.0068390975],"genre_scores_gemma":[0.025220333,0.0002980177,0.042153854,0.0002000506,0.000020959294,0.0009822919,0.9299804,0.00058541016,0.00055872666],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9937032,0.0016759833,0.00095624384,0.0015638861,0.001641424,0.00045928272],"domain_scores_gemma":[0.9931768,0.0019119559,0.00069307967,0.0020903023,0.0014857149,0.00064206764],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008189781,0.0016167869,0.0013232897,0.004679214,0.0018581138,0.0023046788,0.0037358408,0.0025419875,0.0042283274],"category_scores_gemma":[0.016564522,0.0007433136,0.002503987,0.007318336,0.00090936833,0.001883297,0.004015908,0.0027942064,0.003437934],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0061051375,0.0035026318,0.09153033,0.014848913,0.0029330142,0.0016179818,0.0023684571,0.16201751,0.12321313,0.032351736,0.3942734,0.16523777],"study_design_scores_gemma":[0.0012104356,0.0017425087,0.117103904,0.0015611587,0.0006521117,0.0010528489,0.001670645,0.12716357,0.065383635,0.039231773,0.64260054,0.00062684744],"about_ca_topic_score_codex":0.0051359576,"about_ca_topic_score_gemma":0.008298604,"teacher_disagreement_score":0.008189781,"about_ca_system_score_codex":0.0016084495,"about_ca_system_score_gemma":0.002799718,"threshold_uncertainty_score":0.04331225},"labels":[],"label_agreement":null},{"id":"W4282962995","doi":"10.1038/s41597-022-01401-7","title":"A large, curated, open-source stroke neuroimaging dataset to improve lesion segmentation algorithms","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Acute Ischemic Stroke Management","field":"Medicine","cited_by":149,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Research Institute for Aging; Heart and Stroke Foundation; University of Waterloo; Sunnybrook Health Science Centre; Ontario Brain Institute; University of Toronto; University of British Columbia","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute of Neurological Disorders and Stroke; Medical Research Council; National Institute of General Medical Sciences; Wellcome Trust; U.S. Department of Veterans Affairs","keywords":"Generalizability theory; Segmentation; Computer science; Artificial intelligence; Neuroimaging; Atlas (anatomy); Pattern recognition (psychology); Stroke (engine); Machine learning; Medicine; Mathematics; Statistics","score_opus":0.06612173964472086,"score_gpt":0.35520756388031677,"score_spread":0.2890858242355959,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282962995","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06313531,0.0045764693,0.12696743,0.0029031541,0.0017017265,0.0022896035,0.7443144,0.04143211,0.012679912],"genre_scores_gemma":[0.03402878,0.00073566684,0.08418107,0.0006717822,0.00024626154,0.0021101413,0.87154615,0.0035725834,0.0029076603],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99564505,0.00082094263,0.0007849007,0.0014533984,0.0010115695,0.00028415048],"domain_scores_gemma":[0.98929733,0.0031473471,0.0008642033,0.0032621736,0.0029533973,0.00047558302],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0061153774,0.0024252199,0.0018835632,0.005143421,0.0017793855,0.0029839466,0.0035447532,0.002502169,0.013295459],"category_scores_gemma":[0.02557747,0.00079299876,0.0024840438,0.0046691364,0.001018811,0.0022918372,0.003557612,0.0025666188,0.018072834],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00094938895,0.0006041335,0.019608336,0.003539988,0.0009490201,0.0006642142,0.0005762145,0.010119229,0.009748481,0.0036093676,0.7664796,0.18315212],"study_design_scores_gemma":[0.00123103,0.0006009301,0.08778549,0.0020132337,0.0010832532,0.003393831,0.00084198755,0.059634298,0.026733536,0.02815184,0.78796077,0.00056986016],"about_ca_topic_score_codex":0.01693151,"about_ca_topic_score_gemma":0.03640953,"teacher_disagreement_score":0.99645525,"about_ca_system_score_codex":0.0019966804,"about_ca_system_score_gemma":0.0054630404,"threshold_uncertainty_score":0.04447776},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4282979725","doi":"10.1038/s41597-022-01391-6","title":"Long-term ice phenology records spanning up to 578 years for 78 lakes around the Northern Hemisphere","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Arctic and Antarctic ice dynamics","field":"Earth and Planetary Sciences","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Environment; York University","funders":"Division of Environmental Biology; Vetenskapsrådet; National Science Foundation; Siberian Branch, Russian Academy of Sciences; Svenska Forskningsrådet Formas; York University; Leibniz-Institut für Angewandte Geophysik; University of Wisconsin-Madison","keywords":"Phenology; Climate change; Physical geography; Cryosphere; Climatology; Ice core; Geography; Northern Hemisphere; Environmental science; Sea ice; Oceanography; Ecology; Geology; Biology","score_opus":0.03937940469166653,"score_gpt":0.26067399814258424,"score_spread":0.2212945934509177,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282979725","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97272444,0.0018219118,0.0006349695,0.00010391138,0.000011637084,0.000022724735,0.021261353,0.00007138592,0.003347655],"genre_scores_gemma":[0.96789217,0.0016469117,0.0018056642,0.00005070209,0.000038165806,0.000065982975,0.026858,0.00003244791,0.0016099612],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99985814,0.000014332708,0.000020522159,0.000033166907,0.00004188573,0.00003203519],"domain_scores_gemma":[0.9995468,0.00006288306,0.00016450392,0.00003320113,0.0001396622,0.000052815358],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003045587,0.00023319523,0.00022866404,0.0016999008,0.00051514833,0.00046378333,0.00017269189,0.00011454453,0.0011833891],"category_scores_gemma":[0.00058436743,0.000094991774,0.00018986997,0.0025474848,0.00021120723,0.00044847783,0.0005895784,0.00013400703,0.00031570948],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008975386,0.000016834781,0.9451545,0.000200906,0.00015240704,0.00021297146,0.0014154497,0.000524941,0.00326988,0.00010359918,0.004164293,0.04469438],"study_design_scores_gemma":[0.000002705886,0.000012831012,0.9894839,0.000035570392,0.000056797537,0.000085264255,0.00061413605,0.00016940772,0.00046802772,0.000037736903,0.009027508,0.000006223998],"about_ca_topic_score_codex":0.045402545,"about_ca_topic_score_gemma":0.13894181,"teacher_disagreement_score":0.045402545,"about_ca_system_score_codex":0.0004545169,"about_ca_system_score_gemma":0.00073090167,"threshold_uncertainty_score":0.09027654},"labels":[],"label_agreement":null},{"id":"W4283012421","doi":"10.1038/s41597-022-01423-1","title":"A dataset of simulated patient-physician medical interviews with a focus on respiratory cases","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Tellabs (Canada); University of Waterloo; Polytechnique Montréal; Western University","funders":"","keywords":"Computer science; Conversation; Artificial intelligence; Focus (optics); Natural language processing; Set (abstract data type); Converse; Machine learning; Psychology","score_opus":0.0875093153622667,"score_gpt":0.35201481907884236,"score_spread":0.26450550371657566,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283012421","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6078508,0.002081299,0.016185526,0.004269673,0.00091741106,0.0016051482,0.3506335,0.0014852297,0.014971373],"genre_scores_gemma":[0.5901259,0.00074607,0.023540622,0.0011153605,0.00030013995,0.0017492538,0.3763774,0.00009163376,0.0059536085],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981883,0.00095193705,0.0001582053,0.00030279785,0.00025680062,0.00014187195],"domain_scores_gemma":[0.9955349,0.0027963913,0.00025945823,0.0004803619,0.000515009,0.0004138274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012286092,0.00067752186,0.0005034858,0.00103983,0.0006825943,0.00065906916,0.0011976149,0.00225145,0.004838457],"category_scores_gemma":[0.0058692545,0.00018316353,0.00054211984,0.001149346,0.00044560057,0.0004671047,0.0009344594,0.00080274115,0.002620764],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0050455322,0.0065042926,0.1811896,0.00524164,0.0007305542,0.013030612,0.0080237705,0.059925023,0.009213215,0.005588054,0.5620433,0.14346436],"study_design_scores_gemma":[0.00095212413,0.0031816307,0.29457048,0.0014258258,0.00035205437,0.01056768,0.021248572,0.15785527,0.012565891,0.0091586,0.48766893,0.00045298322],"about_ca_topic_score_codex":0.009292213,"about_ca_topic_score_gemma":0.01866661,"teacher_disagreement_score":0.009292213,"about_ca_system_score_codex":0.0010768914,"about_ca_system_score_gemma":0.0010787849,"threshold_uncertainty_score":0.018476248},"labels":[],"label_agreement":null},{"id":"W4283020606","doi":"10.1038/s41597-022-01373-8","title":"PISCOeo_pm, a reference evapotranspiration gridded database based on FAO Penman-Monteith in Peru","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Water Relations and Carbon Dynamics","field":"Environmental Science","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Government of Canada; United States Agency for International Development","keywords":"Evapotranspiration; Cru; Wind speed; Environmental science; Penman–Monteith equation; Multivariate interpolation; Meteorology; Air temperature; Climatology; Dew point; Geography; Precipitation; Mathematics; Geology; Bilinear interpolation","score_opus":0.04865238788339328,"score_gpt":0.2543612250733049,"score_spread":0.20570883718991162,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283020606","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24518302,0.0014587621,0.011724581,0.00044376001,0.00004362758,0.00019970609,0.72227377,0.004163138,0.01450963],"genre_scores_gemma":[0.33155134,0.00061759393,0.016036738,0.000056952988,0.000029308163,0.0006337631,0.6483997,0.00040746192,0.0022670426],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99979013,0.000049607643,0.000023829694,0.0000619378,0.000051525818,0.000023046692],"domain_scores_gemma":[0.9994685,0.00006237001,0.000121930214,0.00009727245,0.00021811624,0.000031831794],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00035546438,0.00043826865,0.00031644324,0.0019133622,0.00021970291,0.0005721644,0.00049415627,0.0002450485,0.003685907],"category_scores_gemma":[0.0016763493,0.00017242452,0.00028408345,0.0049008667,0.00013106603,0.00054356444,0.0005985299,0.00028429966,0.0012061909],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001060433,0.00023838054,0.38211232,0.002846105,0.00082944916,0.00096306787,0.0012308578,0.05191141,0.01016017,0.007134586,0.3158477,0.22566552],"study_design_scores_gemma":[0.00029918159,0.00007362604,0.63912547,0.00025467726,0.00016009837,0.00029038687,0.00054689526,0.0540196,0.0036507756,0.0013697839,0.30011576,0.00009372189],"about_ca_topic_score_codex":0.06308057,"about_ca_topic_score_gemma":0.041160144,"teacher_disagreement_score":0.06308057,"about_ca_system_score_codex":0.00048281543,"about_ca_system_score_gemma":0.0007908337,"threshold_uncertainty_score":0.12542683},"labels":[],"label_agreement":null},{"id":"W4283206600","doi":"10.1038/s41597-022-01383-6","title":"COVIDiSTRESS diverse dataset on psychological and behavioural outcomes one year into the COVID-19 pandemic","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 and Mental Health","field":"Psychology","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Texas A and M University","keywords":"Pandemic; Coronavirus disease 2019 (COVID-19); 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Psychology; Data science; Geography; Biology; Virology; Medicine; Computer science; Pathology; Outbreak; Infectious disease (medical specialty)","score_opus":0.4563412743949642,"score_gpt":0.5145741972217943,"score_spread":0.058232922826830125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283206600","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0062701395,0.000087731,0.00022593704,0.00050426956,0.000062989915,0.00006205233,0.99140924,0.00012525285,0.0012523636],"genre_scores_gemma":[0.010790618,0.0000850917,0.0007740306,0.00032843856,0.00004446866,0.00051874795,0.98610455,0.00004043777,0.0013135149],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99829406,0.00054953416,0.00025059053,0.00033253367,0.00033217296,0.00024105652],"domain_scores_gemma":[0.99357563,0.00174585,0.0010873518,0.0010960011,0.0018846703,0.0006105962],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020968614,0.000592145,0.00071441056,0.002020147,0.0007841255,0.0012387554,0.0012621081,0.0014320053,0.0149561735],"category_scores_gemma":[0.015941227,0.000335986,0.00087394327,0.0037102567,0.00033194697,0.0006643688,0.0026414506,0.0015258742,0.0113439085],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030245044,0.00011996929,0.052521333,0.00057947007,0.00015483792,0.000081964856,0.00026264947,0.0006767356,0.00024775875,0.001134958,0.9338801,0.0100377565],"study_design_scores_gemma":[0.0005018863,0.0001849581,0.44486725,0.00096849585,0.00012570118,0.00023578572,0.0012836605,0.0014711823,0.00072058244,0.002942258,0.5465382,0.00016007219],"about_ca_topic_score_codex":0.035193946,"about_ca_topic_score_gemma":0.047133278,"teacher_disagreement_score":0.035193946,"about_ca_system_score_codex":0.0013384599,"about_ca_system_score_gemma":0.0017421095,"threshold_uncertainty_score":0.06997818},"labels":[],"label_agreement":null},{"id":"W4283330995","doi":"10.1038/s41597-022-01425-z","title":"Global hydro-environmental lake characteristics at high spatial resolution","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":103,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Australian Research Council; Natural Sciences and Engineering Research Council of Canada","keywords":"Structural basin; Land cover; Environmental science; Hydrology (agriculture); Grid; Range (aeronautics); Grid cell; Land use; Remote sensing; Geology; Ecology; Geomorphology","score_opus":0.016097980207951675,"score_gpt":0.21646955088171277,"score_spread":0.2003715706737611,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283330995","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05292196,0.00014145748,0.003295725,0.00017609303,0.000027039376,0.000047801972,0.93493253,0.003241691,0.0052156583],"genre_scores_gemma":[0.1154205,0.00009272935,0.007924061,0.00005145331,0.000021981547,0.00023544386,0.874166,0.00030934718,0.0017785715],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9996532,0.000043911827,0.00004198167,0.00009484267,0.00010570238,0.000060375598],"domain_scores_gemma":[0.99935025,0.0000645964,0.00010434038,0.00016590432,0.0002534252,0.000061529965],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036706045,0.0004295648,0.000426985,0.0031353722,0.00026629958,0.0011419469,0.0004967504,0.00031337645,0.009594751],"category_scores_gemma":[0.001487225,0.00020325015,0.00066678843,0.005954512,0.00019518276,0.00091521954,0.0012238905,0.0003970649,0.0047983993],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050899026,0.0001303105,0.21434619,0.0008738177,0.0006087793,0.00037845763,0.0005614605,0.03178971,0.0056427647,0.0057994043,0.6475037,0.09185651],"study_design_scores_gemma":[0.00026376348,0.000046481775,0.42890924,0.00020378135,0.00014192596,0.00019232198,0.0007700624,0.02999095,0.004974308,0.003932304,0.5303832,0.00019166448],"about_ca_topic_score_codex":0.039390054,"about_ca_topic_score_gemma":0.052689683,"teacher_disagreement_score":0.039390054,"about_ca_system_score_codex":0.0005790692,"about_ca_system_score_gemma":0.0009437959,"threshold_uncertainty_score":0.078321576},"labels":[],"label_agreement":null},{"id":"W4283367256","doi":"10.1038/s41597-022-01467-3","title":"A multi-city urban atmospheric greenhouse gas measurement data synthesis","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric and Environmental Gas Dynamics","field":"Environmental Science","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"Climate Program Office; U.S. Department of Commerce; Jet Propulsion Laboratory; National Aeronautics and Space Administration","keywords":"Greenhouse gas; Scope (computer science); Environmental science; Climate change; Baseline (sea); Environmental resource management; Methane; Environmental planning; Business; Natural resource economics; Environmental protection; Computer science; Political science; Ecology","score_opus":0.07164669648370428,"score_gpt":0.2414225334785098,"score_spread":0.16977583699480553,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283367256","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14352849,0.0050482247,0.18910667,0.005241337,0.0011526314,0.008700999,0.60876244,0.002336378,0.03612284],"genre_scores_gemma":[0.32760093,0.0020179409,0.33026537,0.0014794648,0.00015390104,0.011501196,0.31769243,0.0006336518,0.008655127],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9915392,0.0037628494,0.0012128766,0.001481265,0.001699292,0.00030458707],"domain_scores_gemma":[0.99206233,0.001969939,0.0007330161,0.0018496868,0.003134209,0.00025084827],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009925724,0.00077432214,0.00082448724,0.005141228,0.0013382094,0.0025817403,0.0012679591,0.00090595265,0.0066600447],"category_scores_gemma":[0.012560925,0.00081342977,0.0024695566,0.008383712,0.00039829264,0.0013470621,0.0036070975,0.0009838116,0.0016492923],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011426647,0.0011072155,0.22326826,0.012332245,0.0090111885,0.0009222143,0.0037857022,0.09234414,0.01890284,0.045114994,0.19611955,0.395949],"study_design_scores_gemma":[0.0005366597,0.00036286042,0.18056104,0.0020838662,0.0031994204,0.000179009,0.005758344,0.03663738,0.015286102,0.013911058,0.7412296,0.00025460488],"about_ca_topic_score_codex":0.05241937,"about_ca_topic_score_gemma":0.07907563,"teacher_disagreement_score":0.05241937,"about_ca_system_score_codex":0.0027883456,"about_ca_system_score_gemma":0.00897446,"threshold_uncertainty_score":0.1042285},"labels":[],"label_agreement":null},{"id":"W4283377869","doi":"10.1038/s41597-022-01459-3","title":"A global ensemble of ocean wave climate statistics from contemporary wave reanalysis and hindcasts","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ocean Waves and Remote Sensing","field":"Earth and Planetary Sciences","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"","keywords":"Downscaling; Hindcast; Metadata; Climatology; Significant wave height; Environmental science; Scale (ratio); Sea state; Meteorology; Wind wave; Computer science; Geography; Remote sensing; Geology; Precipitation; Cartography; Oceanography","score_opus":0.061396723068705084,"score_gpt":0.2422891398862953,"score_spread":0.18089241681759022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283377869","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3616376,0.00032671014,0.024221543,0.00023247745,0.00024280064,0.00016241804,0.60525143,0.0029244828,0.0050005144],"genre_scores_gemma":[0.17186162,0.00020439464,0.02524074,0.0000544123,0.00006337929,0.00027613656,0.8007449,0.0003269016,0.0012275522],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99947304,0.00006116767,0.0000553491,0.00018168481,0.00017565282,0.000053123636],"domain_scores_gemma":[0.99805486,0.00019754266,0.00020976132,0.0005979962,0.0008104688,0.00012933684],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012297145,0.00067040604,0.00069659436,0.002288353,0.00029035457,0.0008699537,0.0008589485,0.00048968,0.0018677055],"category_scores_gemma":[0.00243104,0.0003595275,0.00092510844,0.0040214485,0.00020720335,0.0009588485,0.00086376054,0.00084981794,0.0015987463],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00052712625,0.0004502927,0.30049464,0.0004857138,0.0023489767,0.00026818906,0.0003008543,0.2384013,0.011216812,0.004127445,0.23280409,0.20857467],"study_design_scores_gemma":[0.00027621596,0.00015842744,0.6374507,0.00016499082,0.00047785774,0.00013422234,0.00021199834,0.20237173,0.008838005,0.0026045502,0.14703213,0.00027919194],"about_ca_topic_score_codex":0.034528926,"about_ca_topic_score_gemma":0.036171112,"teacher_disagreement_score":0.034528926,"about_ca_system_score_codex":0.00047010084,"about_ca_system_score_gemma":0.0014632089,"threshold_uncertainty_score":0.06865585},"labels":[],"label_agreement":null},{"id":"W4283696359","doi":"10.1038/s41597-022-01475-3","title":"A Global Building Occupant Behavior Database","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Building Energy and Comfort Optimization","field":"Engineering","cited_by":89,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Carleton University","funders":"Dr B R Ambedkar National Institute of Technology Jalandhar; Università degli Studi di Perugia; Indian Institute of Technology Bombay; Università della Calabria; U.S. Department of Energy; Tongji University; Silesian University of Technology; Colorado School of Mines; RMIT University; Tsinghua University; Universidade de São Paulo; Technische Universität Wien; Aalborg Universitet; Universidade Federal de Santa Catarina; Center for the Built Environment; Swinburne University of Technology; University of Toronto; University of New South Wales; Khalifa University of Science, Technology and Research; Western Kentucky University; American Society of Heating, Refrigerating and Air-Conditioning Engineers; Beijing University of Technology; University of Texas at Austin; University of Texas at Arlington; Syddansk Universitet; National Science Foundation","keywords":"Database; Computer science; World Wide Web","score_opus":0.03695447322705658,"score_gpt":0.2720786608106723,"score_spread":0.2351241875836157,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283696359","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012251652,0.00043173134,0.01790399,0.00017305202,0.000055437486,0.0003046051,0.9557043,0.0086974995,0.0044777533],"genre_scores_gemma":[0.016063705,0.00019865656,0.011621783,0.000077407974,0.000012069918,0.00030021375,0.9708054,0.00018158178,0.0007391698],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984498,0.00017793248,0.00035357862,0.0004323693,0.00047784444,0.00010839277],"domain_scores_gemma":[0.9968412,0.00042984812,0.00032719993,0.0010990137,0.0010137375,0.00028901242],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013585032,0.0012847774,0.001415413,0.0046394365,0.00053566485,0.0019483161,0.0027270385,0.0012075388,0.008678684],"category_scores_gemma":[0.003973868,0.00064909295,0.0010566708,0.0071109273,0.00025079673,0.0022276626,0.0017675044,0.001011325,0.009098161],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008238216,0.0007521202,0.07351446,0.0027063384,0.00044094078,0.0005523971,0.00056458026,0.02926883,0.007013638,0.0120440675,0.68182164,0.19049716],"study_design_scores_gemma":[0.0002699027,0.00022729063,0.09047609,0.00047527847,0.00021901476,0.0007390107,0.0007212811,0.054349188,0.009798241,0.00835529,0.8341462,0.00022314009],"about_ca_topic_score_codex":0.0113943685,"about_ca_topic_score_gemma":0.011152356,"teacher_disagreement_score":0.0113943685,"about_ca_system_score_codex":0.00094431447,"about_ca_system_score_gemma":0.0018750215,"threshold_uncertainty_score":0.029033065},"labels":[],"label_agreement":null},{"id":"W4283792177","doi":"10.1038/s41597-022-01434-y","title":"An 8-year record of phytoplankton productivity and nutrient distributions from surface waters of Saanich Inlet","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Water Quality and Pollution Assessment","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Victoria","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada; University of Victoria","keywords":"Phytoplankton; Inlet; Nutrient; Productivity; Environmental science; Oceanography; Primary productivity; Ecology; Biology; Geology","score_opus":0.04037195568643193,"score_gpt":0.2811424661554095,"score_spread":0.24077051046897757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283792177","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9786024,0.00020488189,0.00017830929,0.000065422886,0.000017504959,0.0000305367,0.018101158,0.00003755566,0.0027622401],"genre_scores_gemma":[0.95560026,0.0003968469,0.002012993,0.000095949385,0.000022456881,0.00008314536,0.039346714,0.000021332224,0.0024202808],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998466,0.000008460855,0.00002152267,0.00004721994,0.000049281112,0.00002688844],"domain_scores_gemma":[0.999453,0.00003561002,0.000084803854,0.000046082085,0.00027695647,0.00010351358],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023072727,0.0002689209,0.00027825584,0.0011653379,0.000773766,0.00051033596,0.00037154977,0.00036822612,0.0007477981],"category_scores_gemma":[0.00043344076,0.0002040315,0.00019108572,0.0019361279,0.00033833453,0.0002475402,0.00066797336,0.0002791103,0.0002898443],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007441096,0.00003121627,0.9766634,0.00013774195,0.000078928104,0.00028252872,0.0012175967,0.0005713824,0.0053157234,0.000081082275,0.0027771012,0.01276885],"study_design_scores_gemma":[0.0000030143212,0.000007304281,0.99710256,0.000017808026,0.00001004952,0.000027633978,0.00036187415,0.0002196201,0.00017421594,0.000009218282,0.0020612215,0.0000054324],"about_ca_topic_score_codex":0.5679598,"about_ca_topic_score_gemma":0.82789475,"teacher_disagreement_score":0.4320402,"about_ca_system_score_codex":0.001737113,"about_ca_system_score_gemma":0.0028000746,"threshold_uncertainty_score":0.8691692},"labels":[],"label_agreement":null},{"id":"W4284673623","doi":"10.1038/s41597-022-01503-2","title":"Fisheries dataset on moulting patterns and shell quality of American lobsters H. americanus in Atlantic Canada","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Crustacean biology and ecology","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Canadian Council on Animal Care; University of Prince Edward Island","funders":"Fisheries and Oceans Canada; Atlantic Canada Opportunities Agency; Government of Canada","keywords":"American lobster; Homarus; Fishery; Biology; Moulting; Carapace; Fishing; Crustacean; Ecology; Decapoda; Geography; Larva","score_opus":0.04093232918795222,"score_gpt":0.2710283560594465,"score_spread":0.23009602687149425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4284673623","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033743504,0.0002973692,0.0001225535,0.00007250799,0.000012496962,0.000039919938,0.963816,0.00011527788,0.0017803832],"genre_scores_gemma":[0.037750278,0.00020540808,0.0005439133,0.00006619651,0.000006985729,0.00008737597,0.9597593,0.00001784956,0.001562637],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99956936,0.000013353872,0.000048686314,0.00010332747,0.00015659463,0.000108653796],"domain_scores_gemma":[0.99714786,0.00012805987,0.00035190547,0.0001617464,0.0018957326,0.00031478173],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039521052,0.00067215454,0.0005731458,0.0035686996,0.0011341799,0.0008327942,0.0012683675,0.0005112382,0.006028746],"category_scores_gemma":[0.0016000454,0.00029085437,0.00038830203,0.006546251,0.0002393822,0.0002757188,0.0007882415,0.00041950692,0.003148886],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027281474,0.0001135918,0.53678834,0.0005976154,0.00031052154,0.0003625116,0.0003564101,0.0013089301,0.0014818654,0.0004660467,0.43245864,0.025482696],"study_design_scores_gemma":[0.000050156363,0.000016750171,0.8979487,0.00024574136,0.00005799919,0.00014403867,0.00041562057,0.0012783282,0.00043745388,0.00013916839,0.09921944,0.00004658113],"about_ca_topic_score_codex":0.9453613,"about_ca_topic_score_gemma":0.97549146,"teacher_disagreement_score":0.054638684,"about_ca_system_score_codex":0.00515484,"about_ca_system_score_gemma":0.0073070205,"threshold_uncertainty_score":0.10992086},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4284958904","doi":"10.1038/s41597-022-01519-8","title":"Author Correction: A global ensemble of ocean wave climate statistics from contemporary wave reanalysis and hindcasts","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ocean Waves and Remote Sensing","field":"Earth and Planetary Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"","keywords":"Climatology; Statistics; Environmental science; Geology; Mathematics","score_opus":0.06996294645536501,"score_gpt":0.25679683253715496,"score_spread":0.18683388608178997,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4284958904","genre_codex":"editorial","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008986242,0.000870242,0.0060644057,0.02674661,0.9412164,0.000041036383,0.017215552,0.0021366812,0.0048104795],"genre_scores_gemma":[0.07733687,0.008118752,0.050580867,0.044382792,0.25580516,0.0007005506,0.08780176,0.023941573,0.45133165],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99440235,0.00075468357,0.001034829,0.00083000987,0.0026621951,0.00031583654],"domain_scores_gemma":[0.9313491,0.012634374,0.0027761543,0.005624447,0.046469163,0.0011467576],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044842786,0.0015728043,0.0013071377,0.004265149,0.0019415729,0.0036324298,0.0027396062,0.0024851651,0.09729376],"category_scores_gemma":[0.09476532,0.0011064137,0.0013911844,0.0056979456,0.0010619613,0.0024832261,0.0025218858,0.007479837,0.05184861],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026856274,0.0000029565056,0.00014764673,0.000093704184,0.000013881816,0.000068696216,0.000033521203,0.000099804776,0.00007678929,0.0007368124,0.9924603,0.006239099],"study_design_scores_gemma":[0.00006292542,0.000014351209,0.0016949696,0.000357678,0.000048607148,0.0003552909,0.00011916506,0.0009010913,0.0006654348,0.002225373,0.99350464,0.000050593495],"about_ca_topic_score_codex":0.016324576,"about_ca_topic_score_gemma":0.019490287,"teacher_disagreement_score":0.09729376,"about_ca_system_score_codex":0.0019603895,"about_ca_system_score_gemma":0.0051564616,"threshold_uncertainty_score":0.32548004},"labels":[],"label_agreement":null},{"id":"W4285030286","doi":"10.1038/s41597-022-01495-z","title":"The Toronto older adults gait archive: video and 3D inertial motion capture data of older adults’ walking","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Balance, Gait, and Falls Prevention","field":"Health Professions","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Vector Institute; University of Toronto; Toronto Rehabilitation Institute; University Health Network","funders":"Government of Canada; CIHR Skin Research Training Centre; AGE-WELL","keywords":"Gait; Motion capture; Physical medicine and rehabilitation; Motion (physics); Computer science; Medicine; Artificial intelligence","score_opus":0.029035931952375513,"score_gpt":0.3376554024780065,"score_spread":0.308619470525631,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285030286","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13002247,0.0018217259,0.010994568,0.00027102805,0.00025194915,0.00066307746,0.84811467,0.0022632785,0.0055973344],"genre_scores_gemma":[0.15359373,0.0009824008,0.015773755,0.00010972687,0.00012833632,0.0009229361,0.8240107,0.0001543687,0.0043240953],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99959654,0.000044073036,0.000056898603,0.00008705183,0.00016673474,0.00004875994],"domain_scores_gemma":[0.99889475,0.000072087576,0.00011868911,0.00020509574,0.0005844523,0.00012497952],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004228413,0.0009266231,0.00051173766,0.0023005917,0.00045786792,0.0005371166,0.00087892346,0.00073876115,0.00590999],"category_scores_gemma":[0.0018790639,0.0002485375,0.00036236565,0.0022558374,0.00027306742,0.00035002557,0.0007795744,0.0003770498,0.0037030848],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015082612,0.00064997224,0.12520859,0.0030271432,0.00045445314,0.0019193081,0.001462598,0.006502875,0.014845029,0.0015460708,0.59449834,0.24837737],"study_design_scores_gemma":[0.00032628837,0.0004636268,0.7145265,0.00075129105,0.00021501502,0.0017545793,0.0012619322,0.012635979,0.0047401437,0.0010242177,0.26211324,0.00018720295],"about_ca_topic_score_codex":0.13291365,"about_ca_topic_score_gemma":0.353276,"teacher_disagreement_score":0.13291365,"about_ca_system_score_codex":0.00073941756,"about_ca_system_score_gemma":0.0015454943,"threshold_uncertainty_score":0.26428002},"labels":[],"label_agreement":null},{"id":"W4285730943","doi":"10.1038/s41597-022-01531-y","title":"Author Correction: Long-term ice phenology records spanning up to 578 years for 78 lakes around the Northern Hemisphere","year":2022,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Cryospheric studies and observations","field":"Earth and Planetary Sciences","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Environment; York University","funders":"","keywords":"Phenology; Term (time); Northern Hemisphere; Environmental science; Southern Hemisphere; Climatology; Physical geography; Oceanography; Geography; Geology; Ecology; Biology; Physics; Astronomy","score_opus":0.0749726229622267,"score_gpt":0.29066089205426177,"score_spread":0.21568826909203506,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285730943","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00043312722,0.0005486511,0.0014387995,0.020187972,0.95166683,0.000063912834,0.01703408,0.0012143683,0.0074121957],"genre_scores_gemma":[0.023626992,0.004625283,0.01772703,0.04641384,0.15414713,0.00058891735,0.044421088,0.010444855,0.69800484],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99730057,0.00028367373,0.00052799063,0.0004426171,0.0012171025,0.00022808847],"domain_scores_gemma":[0.9636527,0.0048466884,0.0015874804,0.0034118115,0.025441585,0.0010598181],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026772807,0.002077842,0.0017735959,0.005342298,0.0035146652,0.0034555066,0.003227791,0.0027748554,0.13028532],"category_scores_gemma":[0.042249393,0.0014623794,0.00142165,0.004704193,0.0013680416,0.0026682701,0.0021359448,0.0063731684,0.061866812],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011696734,0.000002907768,0.00007007019,0.000052854793,0.0000049521514,0.000048633596,0.000013841632,0.000026627697,0.00001929523,0.00015952124,0.9972994,0.002290413],"study_design_scores_gemma":[0.00006769685,0.0000098940955,0.001982186,0.00028013816,0.000037013047,0.00020984722,0.00012669661,0.00020627156,0.0003676392,0.0009857826,0.9956937,0.000033148932],"about_ca_topic_score_codex":0.07238738,"about_ca_topic_score_gemma":0.093605034,"teacher_disagreement_score":0.13028532,"about_ca_system_score_codex":0.0028468973,"about_ca_system_score_gemma":0.00715475,"threshold_uncertainty_score":0.43584782},"labels":[],"label_agreement":null},{"id":"W4286494955","doi":"10.1038/s41597-022-01552-7","title":"The Multilingual Picture Database","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Categorization, perception, and language","field":"Psychology","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Centre for Research on Brain Language and Music","funders":"Department of Education and Training; Saint Petersburg State University; Israel Science Foundation; Academy of Finland; Narodowe Centrum Nauki; National Research Foundation of Korea; Comunidad de Madrid","keywords":"Multilingualism; Computer science; Psycholinguistics; Variety (cybernetics); Field (mathematics); Value (mathematics); Linguistics; Natural language processing; Artificial intelligence; Psychology; Mathematics; Cognition","score_opus":0.06283960389950143,"score_gpt":0.36443300166592907,"score_spread":0.30159339776642763,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4286494955","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0048797303,0.0006417665,0.0016077523,0.0002720972,0.00008986301,0.00019797022,0.98359317,0.0037033376,0.0050142724],"genre_scores_gemma":[0.0035986954,0.00011893393,0.002275194,0.00006399993,0.000010193366,0.00021495318,0.99272823,0.00017663601,0.00081313885],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980515,0.0003194593,0.000311158,0.00059168716,0.000535833,0.00019031012],"domain_scores_gemma":[0.9974274,0.0003752051,0.00020608501,0.0009184047,0.0007145139,0.00035837403],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013031021,0.0018623866,0.0013214805,0.0050344984,0.0012481974,0.0025273617,0.00417363,0.0022570333,0.03571274],"category_scores_gemma":[0.0070080645,0.0005948052,0.0011786614,0.006508165,0.0006190455,0.0038922818,0.004170771,0.0022354545,0.045471024],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048354355,0.00020008298,0.0042067543,0.0019451369,0.00012873182,0.00034525857,0.0002967748,0.00076438487,0.0019414991,0.0030700467,0.9545405,0.032077298],"study_design_scores_gemma":[0.00028971527,0.000086712564,0.013937853,0.0003071291,0.00007147203,0.00053685426,0.0006864253,0.0029959304,0.0025432922,0.0029430948,0.97549385,0.00010765906],"about_ca_topic_score_codex":0.018059362,"about_ca_topic_score_gemma":0.029404595,"teacher_disagreement_score":0.03571274,"about_ca_system_score_codex":0.0015437948,"about_ca_system_score_gemma":0.0025189596,"threshold_uncertainty_score":0.11947107},"labels":[],"label_agreement":null},{"id":"W4286750103","doi":"10.1038/s41597-022-01544-7","title":"Single-cell RNA-seq of primary bone marrow neutrophils from female and male adult mice","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; Rose Hills Foundation; Buck Institute for Research on Aging; Echo Foundation; McGill University; National Institutes of Health; Pew Charitable Trusts; U.S. Department of Health and Human Services","keywords":"Biology; Transcriptome; Innate immune system; Immune system; Sexual dimorphism; Bone marrow; Immunology; Cell; Cell type; RNA; Phenotype; Cell biology; Gene expression; Gene; Genetics; Endocrinology","score_opus":0.029329118770350472,"score_gpt":0.21773767739736172,"score_spread":0.18840855862701125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4286750103","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71584773,0.0046880487,0.08408518,0.0004013215,0.0003613005,0.00023912957,0.18365638,0.00142925,0.009291676],"genre_scores_gemma":[0.7577002,0.0033315236,0.093164735,0.0016721722,0.000114842296,0.0010072782,0.12874949,0.0015880167,0.012671783],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9997745,0.000010679189,0.000010711951,0.00012045775,0.0000575599,0.000026058719],"domain_scores_gemma":[0.9998654,0.000034697998,0.000020382227,0.000016977026,0.000039595605,0.00002296525],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00024059608,0.00026101831,0.00037945382,0.00032982585,0.00036863936,0.00033857976,0.00025539208,0.00035274046,0.0018671091],"category_scores_gemma":[0.00026966547,0.00015976607,0.00035104682,0.00028182523,0.00023729658,0.00012622506,0.00029656183,0.00060830894,0.0008106577],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018959379,0.000013139411,0.003103237,0.00019133041,0.000050137558,0.00008699629,0.00006739087,0.00043666226,0.9848839,0.00029145484,0.0011419225,0.009544095],"study_design_scores_gemma":[0.00009344768,0.0004031709,0.21658163,0.0001720708,0.00029824892,0.0010512925,0.0003339411,0.010314853,0.6908483,0.0026693565,0.077138074,0.00009565279],"about_ca_topic_score_codex":0.0011283138,"about_ca_topic_score_gemma":0.0040725023,"teacher_disagreement_score":0.0018671091,"about_ca_system_score_codex":0.00016566587,"about_ca_system_score_gemma":0.0002936216,"threshold_uncertainty_score":0.0062461495},"labels":[],"label_agreement":null},{"id":"W4288068880","doi":"10.1038/s41597-022-01515-y","title":"Community established best practice recommendations for tephra studies—from collection through analysis","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Environment Research Council; Research Councils UK; Sight Research UK","keywords":"Tephra; Metadata; Discoverability; Documentation; Standardization; Data science; Earth science; Interoperability; Computer science; Geography; Volcano; World Wide Web; Geology; Paleontology","score_opus":0.24386890989179388,"score_gpt":0.40820881212516447,"score_spread":0.16433990223337058,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288068880","genre_codex":"commentary","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0066931336,0.011616072,0.36369106,0.43104538,0.017469402,0.040905528,0.012054905,0.013004186,0.103520356],"genre_scores_gemma":[0.012261379,0.007986041,0.8825605,0.04019024,0.0016500313,0.02838513,0.006649033,0.0027550752,0.017562646],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.7121281,0.15073797,0.06393668,0.010086988,0.054582354,0.008527878],"domain_scores_gemma":[0.25152302,0.18354873,0.029411739,0.08720207,0.42114937,0.027165078],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.31197536,0.0027343025,0.0032762447,0.01958356,0.0070697274,0.020297078,0.017233076,0.016280225,0.020526087],"category_scores_gemma":[0.55016977,0.0035643226,0.005308707,0.017395003,0.007933185,0.023010526,0.016280293,0.017910985,0.031063426],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016909074,0.000602663,0.0033081667,0.005509501,0.00014922605,0.00058847474,0.007978143,0.001008472,0.002007291,0.0122489855,0.64915967,0.3172703],"study_design_scores_gemma":[0.00022616088,0.00019066918,0.004371795,0.025775936,0.00016481586,0.0003651352,0.010307385,0.0012796132,0.0020801,0.026197016,0.9286947,0.00034674146],"about_ca_topic_score_codex":0.03215409,"about_ca_topic_score_gemma":0.045883056,"teacher_disagreement_score":0.31197536,"about_ca_system_score_codex":0.010969929,"about_ca_system_score_gemma":0.101948746,"threshold_uncertainty_score":0.84845686},"labels":[],"label_agreement":null},{"id":"W4288439585","doi":"10.1038/s41597-022-01559-0","title":"The role of plant functional groups mediating climate impacts on carbon and biodiversity of alpine grasslands","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Peatlands and Wetlands Ecology","field":"Environmental Science","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Senter for Internasjonalisering av Utdanning; Norges Forskningsråd; Universitetet i Bergen","keywords":"Environmental science; Ecosystem; Plant community; Biomass (ecology); Ecology; Forb; Species richness; Graminoid; Biodiversity; Climate change; Growing season; Biology; Grassland","score_opus":0.015555839301691065,"score_gpt":0.20310089661832093,"score_spread":0.18754505731662985,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288439585","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9998272,0.000037088732,0.000026365067,0.000004479542,3.840135e-7,6.0704826e-7,0.000025081807,0.0000012665554,0.00007760469],"genre_scores_gemma":[0.99979514,0.000017267288,0.00005674808,0.000006314772,0.0000011421795,0.0000019856975,0.00008060822,0.0000011096322,0.000039666447],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99959844,0.0001910834,0.000020361244,0.00008225608,0.000030571882,0.00007723708],"domain_scores_gemma":[0.9990295,0.00030608394,0.00029078027,0.00009778917,0.000058937614,0.00021679302],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008206602,0.00028101465,0.00030074076,0.0005099472,0.0003674629,0.00048627093,0.00022901094,0.0002250652,0.000661531],"category_scores_gemma":[0.00086510106,0.00012517818,0.00020836173,0.00020218937,0.0006400894,0.00029222635,0.00039567004,0.0001982271,0.00007095638],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012621403,0.00017181362,0.8681587,0.00006965559,0.00023730594,0.00009882001,0.00051050814,0.00095295697,0.12041719,0.00013319618,0.000077016535,0.007910616],"study_design_scores_gemma":[0.0000041612943,0.000046015866,0.9991301,0.0000016445381,0.000011817267,0.000012636276,0.000056244753,0.0002810573,0.0003809033,0.000022333266,0.000051159885,0.0000019144081],"about_ca_topic_score_codex":0.007402937,"about_ca_topic_score_gemma":0.015253986,"teacher_disagreement_score":0.007402937,"about_ca_system_score_codex":0.00036002946,"about_ca_system_score_gemma":0.00021180235,"threshold_uncertainty_score":0.014719665},"labels":[],"label_agreement":null},{"id":"W4288684979","doi":"10.1038/s41597-022-01534-9","title":"ISARIC-COVID-19 dataset: A Prospective, Standardized, Global Dataset of Patients Hospitalized with COVID-19","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 Clinical Research Studies","field":"Medicine","cited_by":72,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Grand River Hospital; Misericordia Community Hospital; BC Children's Hospital; University of British Columbia; Sinai Health System; Sturgeon Community Hospital; London Health Sciences Centre; Mount Sinai Hospital; Royal Columbian Hospital; Kingston Health Sciences Centre; Joseph Brant Hospital; University of Alberta; University Health Network; St. Joseph’s Healthcare Hamilton; Alberta Children's Hospital; Red Deer Regional Hospital; Grey Nuns Community Hospital; SickKids Foundation; University of Manitoba; University of Northern British Columbia; Children's Hospital of Eastern Ontario; Vancouver General Hospital; North York General Hospital; St Joseph's Health Centre; Centre Hospitalier Universitaire de Sherbrooke; Toronto East General Hospital; Foothills Medical Centre; McMaster University; Mills Memorial Hospital; Lions Gate Hospital; Centre Hospitalier Universitaire Sainte-Justine; Ottawa Hospital; Royal Alexandra Hospital; Hôpital du Sacré-Cœur de Montréal; Centre intégré de santé et de services sociaux de Chaudière-Appalaches; Health Sciences Centre; McGill University Health Centre; Etobicoke General Hospital; Montreal Children's Hospital; Hôpital de l'Enfant-Jésus; Brantford Energy (Canada); Island Health; Sunnybrook Health Science Centre; Humber River Regional Hospital; St. Boniface Hospital; Institut universitaire de cardiologie et de pneumologie de Québec; Centre Hospitalier de l’Université de Montréal","funders":"Foreign, Commonwealth and Development Office; Medical Research Council; Kementerian Pendidikan Nasional; Public Health England; Minderoo Foundation; University of Oxford; Imperial College London; Canadian Institutes of Health Research; National Institute of General Medical Sciences; National Institute for Health and Care Research; National Heart, Lung, and Blood Institute; National Institute for Health Research Health Protection Research Unit; Wellcome Trust; Bill and Melinda Gates Foundation","keywords":"Coronavirus disease 2019 (COVID-19); Medicine; Pandemic; Emergency medicine; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); MEDLINE; Intensive care medicine; 2019-20 coronavirus outbreak; Medical emergency; Internal medicine; Outbreak; Pathology; Disease; Infectious disease (medical specialty)","score_opus":0.0842770208689757,"score_gpt":0.45181960599310755,"score_spread":0.36754258512413185,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4288684979","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022701465,0.000214725,0.0002793665,0.00019179862,0.00004022787,0.00007317337,0.9962119,0.00011772659,0.0006009149],"genre_scores_gemma":[0.0036495316,0.000104972816,0.0006527164,0.00012550812,0.000029472421,0.00024236798,0.99500567,0.000020900232,0.00016879238],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99767154,0.0005055077,0.0006600437,0.00059327786,0.0003099847,0.0002596672],"domain_scores_gemma":[0.99363494,0.0019365249,0.0014133155,0.0009846572,0.0013984668,0.00063212053],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023513609,0.0010791436,0.0014930328,0.0034317167,0.0006037949,0.0016453579,0.0020920357,0.0016790823,0.011952175],"category_scores_gemma":[0.013371036,0.0004314474,0.0013118284,0.0058513726,0.00033533128,0.0007083427,0.0017632163,0.0015896072,0.0079561705],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00097045576,0.00016843708,0.05716293,0.0022132879,0.0004454683,0.00029847803,0.00014576486,0.0017957577,0.0006130636,0.0019090746,0.92165923,0.012618033],"study_design_scores_gemma":[0.0017869575,0.00031288876,0.16859327,0.0022133456,0.00052788603,0.0011613942,0.00059495255,0.004947761,0.0013611146,0.0042792647,0.8140003,0.00022089164],"about_ca_topic_score_codex":0.016147954,"about_ca_topic_score_gemma":0.019848162,"teacher_disagreement_score":0.016147954,"about_ca_system_score_codex":0.001382914,"about_ca_system_score_gemma":0.0037145917,"threshold_uncertainty_score":0.039983988},"labels":[],"label_agreement":null},{"id":"W4289261700","doi":"10.1038/s41597-022-01517-w","title":"The United States COVID-19 Forecast Hub dataset","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":129,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University; University of British Columbia; University of Victoria; TRIUMF","funders":"Oak Ridge National Laboratory; Natural Sciences and Engineering Research Council of Canada; Quest for Intelligence, Massachusetts Institute of Technology; Plant Sciences Institute, Iowa State University; National Institutes of Health; Centers for Disease Control and Prevention; Klaus Tschira Stiftung; Center for Emerging Infectious Diseases, University of Iowa; Iowa State University; North Carolina State University; Bundesministerium für Bildung und Forschung; Council of State and Territorial Epidemiologists; Institute for Health Metrics and Evaluation; National Institute of General Medical Sciences; University of Massachusetts Amherst; Johns Hopkins Bloomberg School of Public Health; Wellcome Trust; U.S. Department of Energy; National Institute of Diabetes and Digestive and Kidney Diseases; California Institute of Technology; Eunice Kennedy Shriver National Institute of Child Health and Human Development; Bill and Melinda Gates Foundation; Los Alamos National Laboratory; Johns Hopkins University; Gordon and Betty Moore Foundation; Indiana University-Purdue University Indianapolis; U.S. Department of Homeland Security; National Nuclear Security Administration; Laboratory Directed Research and Development; National Science Foundation","keywords":"Leverage (statistics); Download; Coronavirus disease 2019 (COVID-19); Government (linguistics); Pandemic; Scale (ratio); Disease control; Computer science; Data science; Econometrics; Business; Geography; Infectious disease (medical specialty); Economics; World Wide Web; Environmental health; Medicine; Machine learning","score_opus":0.5946829767412236,"score_gpt":0.495350206423102,"score_spread":0.0993327703181216,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4289261700","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003931683,0.00025722824,0.000644494,0.00044395932,0.00012368894,0.00003567875,0.9922047,0.00063148496,0.0017269602],"genre_scores_gemma":[0.0070120785,0.00012357745,0.0012722688,0.00010476068,0.000037142527,0.000081927334,0.99087673,0.000056095847,0.00043528152],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992131,0.00019865051,0.00011445112,0.00020086873,0.00018359839,0.00008928197],"domain_scores_gemma":[0.9976447,0.0008374863,0.00028016826,0.00043138067,0.0006195895,0.00018652927],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013900719,0.000947383,0.00064783957,0.0022354766,0.0005575303,0.0011640171,0.0014802339,0.0012743233,0.012005834],"category_scores_gemma":[0.0072976258,0.00032874345,0.00089855684,0.0033367423,0.0002467873,0.0007896767,0.0010345273,0.0015391279,0.009510831],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013914621,0.00006994362,0.011195634,0.00032855704,0.00009443618,0.00008153044,0.00004430565,0.006081802,0.00021299167,0.0013278,0.97246134,0.007962618],"study_design_scores_gemma":[0.0005982881,0.00009908704,0.03878868,0.00053955294,0.00012979982,0.00023741809,0.0003525301,0.03396039,0.0014216247,0.006703849,0.9170411,0.00012769524],"about_ca_topic_score_codex":0.03186092,"about_ca_topic_score_gemma":0.03687044,"teacher_disagreement_score":0.03186092,"about_ca_system_score_codex":0.0009991718,"about_ca_system_score_gemma":0.0018114167,"threshold_uncertainty_score":0.063350916},"labels":[],"label_agreement":null},{"id":"W4289545216","doi":"10.1038/s41597-022-01568-z","title":"Developing a high-resolution gridded rainfall product for Bangladesh during 1901–2018","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Precipitation Measurement and Analysis","field":"Earth and Planetary Sciences","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"National Research Foundation of Korea","keywords":"Product (mathematics); Environmental science; Climatology; Geology; Mathematics","score_opus":0.08742541192371227,"score_gpt":0.2588237999639356,"score_spread":0.17139838804022334,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4289545216","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6502371,0.00021776378,0.025691397,0.00037668392,0.000083414474,0.00048427313,0.3133444,0.001990584,0.007574422],"genre_scores_gemma":[0.7150078,0.00033783767,0.05802427,0.00004806601,0.000020085374,0.0006954711,0.22339246,0.0002198031,0.0022541464],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99976,0.000037825885,0.000043633263,0.00006399036,0.00007268927,0.000021943175],"domain_scores_gemma":[0.9993206,0.00006978676,0.00012036392,0.00007318997,0.00038198222,0.000034112865],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00052733667,0.0005087099,0.0002505085,0.0013029869,0.00018393902,0.0004901381,0.00042581657,0.00022685576,0.0022461207],"category_scores_gemma":[0.001746684,0.00022185067,0.00026646466,0.0023359726,0.000119235214,0.0005092793,0.00035212812,0.00024938036,0.0015119812],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00073022366,0.00035177518,0.507342,0.0008841203,0.0003565685,0.0014690026,0.0010149828,0.09952532,0.036115743,0.0017313437,0.053811934,0.29666707],"study_design_scores_gemma":[0.00019729462,0.00018660285,0.770132,0.00013243852,0.000115705785,0.00026192886,0.0013355983,0.13922155,0.01994291,0.0006589684,0.06769934,0.00011567378],"about_ca_topic_score_codex":0.04200519,"about_ca_topic_score_gemma":0.036260292,"teacher_disagreement_score":0.04200519,"about_ca_system_score_codex":0.00072573917,"about_ca_system_score_gemma":0.0008363967,"threshold_uncertainty_score":0.083521366},"labels":[],"label_agreement":null},{"id":"W4290065549","doi":"10.1038/s41597-022-01554-5","title":"A global dataset of seaweed net primary productivity","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and fisheries research","field":"Environmental Science","cited_by":69,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bedford Institute of Oceanography; Fisheries and Oceans Canada","funders":"","keywords":"Primary productivity; Primary production; Productivity; Algae; Net (polyhedron); Primary (astronomy); Primary producers; Environmental science; Ecology; Biology; Mathematics; Phytoplankton; Ecosystem; Nutrient; Economics","score_opus":0.044058248072791545,"score_gpt":0.28634018294948155,"score_spread":0.24228193487669,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4290065549","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03300949,0.00014600587,0.00064460025,0.000052445845,0.000016472377,0.000026463991,0.96452576,0.00026604853,0.0013126895],"genre_scores_gemma":[0.018484809,0.00007524276,0.0014703822,0.000017932687,0.000007725083,0.00006081599,0.9794462,0.000027032036,0.00040992702],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99942386,0.000054580367,0.000087723616,0.00019777344,0.00016089359,0.000075232456],"domain_scores_gemma":[0.99819356,0.0002679034,0.00037922733,0.0004018053,0.0005489472,0.00020854293],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005310519,0.0006720466,0.0005686491,0.0031258734,0.00031139297,0.00079713884,0.0007166345,0.0006258637,0.004927067],"category_scores_gemma":[0.001905634,0.0002688694,0.00057924836,0.006989835,0.0002169878,0.0005101373,0.001084608,0.000595209,0.004730847],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00070786057,0.00031707535,0.40203068,0.0026097458,0.0009702371,0.00088098494,0.0006422818,0.018672029,0.020393327,0.0031421026,0.4579522,0.09168144],"study_design_scores_gemma":[0.00016713054,0.0000939973,0.63159984,0.0001740968,0.00009717849,0.00031736473,0.00044380673,0.004853108,0.0032023229,0.0009619772,0.35801116,0.00007798427],"about_ca_topic_score_codex":0.018709367,"about_ca_topic_score_gemma":0.01967899,"teacher_disagreement_score":0.018709367,"about_ca_system_score_codex":0.000544046,"about_ca_system_score_gemma":0.001007535,"threshold_uncertainty_score":0.037200928},"labels":[],"label_agreement":null},{"id":"W4291020824","doi":"10.1038/s41597-022-01604-y","title":"Global dataset of species-specific inland recreational fisheries harvest for consumption","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fish Ecology and Management Studies","field":"Environmental Science","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"U.S. Geological Survey; University of Missouri","keywords":"Recreation; Fishing; Fishery; Recreational fishing; Fisheries management; Livelihood; Food security; Consumption (sociology); Ecosystem; Geography; Landlocked country; Ecosystem services; Environmental resource management; Natural resource economics; Agriculture; Environmental science; Ecology; Biology; Economics","score_opus":0.08465527016480735,"score_gpt":0.27953319603743176,"score_spread":0.19487792587262442,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4291020824","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020728493,0.00067597034,0.0011071655,0.00008886251,0.000028017857,0.000037446367,0.97310233,0.00019850866,0.0040332903],"genre_scores_gemma":[0.03930353,0.00065600016,0.0030629952,0.00008471662,0.000014427125,0.00021909452,0.9553723,0.00006095971,0.0012259145],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99943537,0.0000796418,0.0001248109,0.00016318481,0.00012657356,0.00007045652],"domain_scores_gemma":[0.998047,0.0003478934,0.00060787663,0.0002237913,0.0006408088,0.00013267873],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007395138,0.000812194,0.0005113766,0.0054480266,0.00029847393,0.0006479563,0.0006522963,0.00043821204,0.0078080203],"category_scores_gemma":[0.0027544482,0.0002761969,0.00079997454,0.009740091,0.00023334601,0.0007877575,0.001140758,0.0006812257,0.0039621857],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003911779,0.00014148917,0.3215938,0.008108081,0.0011744825,0.00068655296,0.0010684713,0.0068648783,0.0035177567,0.0046449825,0.55053383,0.101274475],"study_design_scores_gemma":[0.000050748913,0.0000645254,0.5970135,0.0011622388,0.0002809251,0.00046110502,0.0010481649,0.0022059444,0.0010549747,0.0011517989,0.39541233,0.00009375013],"about_ca_topic_score_codex":0.025734114,"about_ca_topic_score_gemma":0.025895659,"teacher_disagreement_score":0.025734114,"about_ca_system_score_codex":0.00049489515,"about_ca_system_score_gemma":0.0011054255,"threshold_uncertainty_score":0.05116868},"labels":[],"label_agreement":null},{"id":"W4291021452","doi":"10.1038/s41597-022-01608-8","title":"BRAX, Brazilian labeled chest x-ray dataset","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 diagnosis using AI","field":"Medicine","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children","funders":"National Institutes of Health; Sociedade Beneficente Israelita Brasileira Albert Einstein; National Institute of Biomedical Imaging and Bioengineering; U.S. Department of Health and Human Services","keywords":"Computer science; Radiology; Radiography; Medical physics; Artificial intelligence; Medicine","score_opus":0.08167703993095068,"score_gpt":0.361811452426393,"score_spread":0.28013441249544235,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4291021452","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022491269,0.0022416264,0.0031483846,0.00083372026,0.00016529213,0.00040255414,0.96185017,0.00325195,0.0056150756],"genre_scores_gemma":[0.017552376,0.00059170864,0.004497832,0.00017302718,0.00005587024,0.00036415612,0.97558093,0.000105945364,0.0010781264],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990559,0.00015551047,0.00018043471,0.00024373663,0.00027368634,0.00009072265],"domain_scores_gemma":[0.9982712,0.0004602321,0.00023595645,0.00037133272,0.0005050794,0.00015615324],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009929395,0.0015509968,0.0009463292,0.0035529986,0.0006422146,0.0011440892,0.00216663,0.0015186527,0.008963859],"category_scores_gemma":[0.0055743847,0.00037413876,0.00077422545,0.0028274956,0.00045429252,0.00061084813,0.0013641579,0.00093334756,0.009926184],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012269935,0.0004734232,0.026741909,0.004036,0.00021431447,0.0015653397,0.00034339284,0.0024556692,0.006780045,0.0020588487,0.8700411,0.08406309],"study_design_scores_gemma":[0.00077263487,0.0002806088,0.10458781,0.0013708695,0.00027354757,0.0039798017,0.00072965777,0.013042281,0.008011466,0.0032340658,0.86356205,0.00015511905],"about_ca_topic_score_codex":0.022928517,"about_ca_topic_score_gemma":0.0339954,"teacher_disagreement_score":0.022928517,"about_ca_system_score_codex":0.0013399312,"about_ca_system_score_gemma":0.002282361,"threshold_uncertainty_score":0.045590103},"labels":[],"label_agreement":null},{"id":"W4292182021","doi":"10.1038/s41597-022-01579-w","title":"Fundamental social motives measured across forty-two cultures in two waves","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cultural Differences and Values","field":"Psychology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; University of Waterloo","funders":"Akademie Věd České Republiky; Universidad de La Frontera; Grantová Agentura České Republiky; Fundação de Amparo à Pesquisa do Estado de São Paulo; Aberystwyth University; National Science Foundation","keywords":"Feminism; Sociology; Geography; Gender studies","score_opus":0.2643450644218393,"score_gpt":0.49003977297467755,"score_spread":0.22569470855283824,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4292182021","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9979087,0.000067624715,0.0002654869,0.00004953767,0.000012482299,0.000052801915,0.00058464403,0.0000025517797,0.0010562721],"genre_scores_gemma":[0.99746823,0.000078568686,0.00066134,0.000058963084,0.00000764639,0.000264762,0.0010242448,0.0000050818808,0.00043116516],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9987832,0.00048332755,0.00015884385,0.00015589503,0.00022004188,0.00019870978],"domain_scores_gemma":[0.9947798,0.0009920875,0.0017826131,0.0006102848,0.0012363921,0.0005988034],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024621272,0.00030022569,0.0003467635,0.0014368746,0.0011275993,0.0014851558,0.00032378852,0.0004320057,0.002058482],"category_scores_gemma":[0.00676794,0.000368626,0.0005221151,0.0017863655,0.0010326962,0.0009893375,0.0019659258,0.00086791156,0.00036656598],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000099487894,0.00014004683,0.97732294,0.000057804922,0.00015282688,0.00005378575,0.011232274,0.000109702625,0.00056804466,0.0004904629,0.00083573285,0.008936981],"study_design_scores_gemma":[0.00000707674,0.00005397197,0.98861814,0.000031398547,0.000022954324,0.000039561746,0.009447914,0.00015312174,0.00022242652,0.00021089087,0.0011749661,0.000017538368],"about_ca_topic_score_codex":0.006889733,"about_ca_topic_score_gemma":0.010642369,"teacher_disagreement_score":0.006889733,"about_ca_system_score_codex":0.000640657,"about_ca_system_score_gemma":0.00051837735,"threshold_uncertainty_score":0.0136992335},"labels":[],"label_agreement":null},{"id":"W4293090388","doi":"10.1038/s41597-022-01624-8","title":"A multi-scale probabilistic atlas of the human connectome","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"Centre d'Imagerie BioMédicale; Université de Lausanne; Centre Hospitalier Universitaire Vaudois; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Atlas (anatomy); Connectome; Scale (ratio); Computer science; Probabilistic logic; Data science; Biology; Artificial intelligence; Geography; Neuroscience; Cartography; Functional connectivity; Paleontology","score_opus":0.14059844892235812,"score_gpt":0.31754636454830976,"score_spread":0.17694791562595164,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293090388","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034769863,0.00044805548,0.93328893,0.0003196482,0.00008787225,0.00036621487,0.015064471,0.0074523063,0.008202664],"genre_scores_gemma":[0.2458943,0.00093250314,0.7240414,0.00017869899,0.00009059895,0.0017097587,0.017504772,0.0031981883,0.006449824],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9996704,0.000097761425,0.000026307687,0.000093151786,0.000086971544,0.00002532209],"domain_scores_gemma":[0.99941194,0.00024476124,0.00007862478,0.00014129089,0.00008791603,0.000035415684],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00069796125,0.00048605914,0.00035902057,0.0020272455,0.0005103815,0.001200875,0.00065745314,0.00077593245,0.013928303],"category_scores_gemma":[0.0020468934,0.00043463733,0.0006871023,0.0020940804,0.000465728,0.00079774373,0.0010891815,0.0009487786,0.0026665144],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007904347,0.00027113344,0.014935698,0.0016804877,0.0004985387,0.0018635713,0.0018809193,0.17752786,0.12821907,0.093174726,0.11534314,0.46381444],"study_design_scores_gemma":[0.00020938407,0.00037545565,0.08389784,0.0003171317,0.0002350299,0.009518607,0.00032598383,0.45055264,0.033876017,0.13055277,0.28978002,0.00035919517],"about_ca_topic_score_codex":0.0026959905,"about_ca_topic_score_gemma":0.004958402,"teacher_disagreement_score":0.013928303,"about_ca_system_score_codex":0.00043487296,"about_ca_system_score_gemma":0.0012138219,"threshold_uncertainty_score":0.0465948},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"bench_or_experimental","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4293103396","doi":"10.1038/s41597-022-01571-4","title":"qMRI-BIDS: An extension to the brain imaging data structure for quantitative magnetic resonance imaging data","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Medical Imaging Techniques and Applications","field":"Medicine","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; Polytechnique Montréal; Montreal Heart Institute","funders":"Biotechnology and Biological Sciences Research Council; Canada First Research Excellence Fund; Fonds De La Recherche Scientifique - FNRS; Research Councils UK; International Society for Magnetic Resonance in Medicine; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Wellcome Trust; Directorate for Biological Sciences; Canadian Open Neuroscience Platform; Wellcome","keywords":"Extension (predicate logic); Magnetic resonance imaging; Functional magnetic resonance imaging; Nuclear magnetic resonance; Computer science; Information retrieval; Physics; Neuroscience; Psychology; Medicine; Radiology","score_opus":0.12701113299853875,"score_gpt":0.4141679176153345,"score_spread":0.2871567846167957,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293103396","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038423052,0.00044238137,0.903828,0.0030187229,0.00056799967,0.0012880296,0.028965296,0.047842026,0.0102051785],"genre_scores_gemma":[0.046582844,0.0007530292,0.8144992,0.0023239537,0.00045254067,0.0035013033,0.10401763,0.019670416,0.0081991],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.97996855,0.005955185,0.0043930323,0.0020631219,0.0064861304,0.0011339824],"domain_scores_gemma":[0.94417703,0.0108453985,0.003533229,0.025636114,0.012842525,0.0029657641],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.03477716,0.0015522423,0.0015769243,0.004705618,0.0017008471,0.008810547,0.007610179,0.002224899,0.014643028],"category_scores_gemma":[0.08235337,0.0024898436,0.0031729315,0.005591863,0.0027338266,0.01199247,0.02121101,0.0051577464,0.016608289],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023495946,0.00040515605,0.007794356,0.0018857461,0.00039954766,0.0007430877,0.00259707,0.011910414,0.013629773,0.36597276,0.3655811,0.22673142],"study_design_scores_gemma":[0.00041032585,0.00024057024,0.0029606977,0.0005246243,0.00008716966,0.00078106014,0.00036610884,0.03997793,0.013617345,0.19512847,0.7456459,0.00025975727],"about_ca_topic_score_codex":0.0070396354,"about_ca_topic_score_gemma":0.0057285,"teacher_disagreement_score":0.9923898,"about_ca_system_score_codex":0.002050164,"about_ca_system_score_gemma":0.009488639,"threshold_uncertainty_score":0.1839214},"labels":[],"label_agreement":null},{"id":"W4293321971","doi":"10.1038/s41597-022-01633-7","title":"Real-world behavioral dataset from two fully remote smartphone-based randomized clinical trials for depression","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Digital Mental Health Interventions","field":"Psychology","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; University of Toronto; Vector Institute; Centre for Addiction and Mental Health","funders":"National Institute of Mental Health; U.S. Department of Health and Human Services","keywords":"Mental health; Digital health; Anxiety; Mood; Randomized controlled trial; Health care; Phone; Depression (economics); Telemedicine; Computer science; Applied psychology; Psychology; Medicine; Psychiatry","score_opus":0.4865865810851913,"score_gpt":0.5961037204382227,"score_spread":0.10951713935303137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293321971","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08191889,0.0019421625,0.008714135,0.0035536159,0.00040114505,0.0031630832,0.89451677,0.0012513046,0.0045388737],"genre_scores_gemma":[0.168343,0.0009160956,0.02222926,0.0029844837,0.00025209214,0.013888903,0.78822696,0.00016666848,0.0029925506],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9931526,0.0037994294,0.0010678056,0.00084738253,0.0009226175,0.00021019184],"domain_scores_gemma":[0.96520495,0.022408161,0.0036595236,0.005742217,0.00202058,0.000964517],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009252509,0.00081501255,0.0013541507,0.0011875966,0.00065034966,0.0012056287,0.0018592868,0.0019749468,0.0099955015],"category_scores_gemma":[0.052371465,0.00048556758,0.0019674113,0.0020327324,0.00053487456,0.0005959027,0.0013683719,0.0016773065,0.0033273837],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.013706151,0.0032405942,0.13574606,0.011810343,0.0074697,0.0010074435,0.0007846345,0.015659703,0.0033690804,0.0069631445,0.69511825,0.10512488],"study_design_scores_gemma":[0.032081593,0.0071445685,0.50082785,0.003691161,0.0060776575,0.0019473418,0.0008500113,0.028951459,0.0055453028,0.021498524,0.39077365,0.0006108912],"about_ca_topic_score_codex":0.005498299,"about_ca_topic_score_gemma":0.014328428,"teacher_disagreement_score":0.0099955015,"about_ca_system_score_codex":0.0010585387,"about_ca_system_score_gemma":0.0024203302,"threshold_uncertainty_score":0.048932493},"labels":[],"label_agreement":null},{"id":"W4293463483","doi":"10.1038/s41597-022-01625-7","title":"Le Petit Prince multilingual naturalistic fMRI corpus","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neurobiology of Language and Bilingualism","field":"Neuroscience","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"Office of Naval Research; National Institute on Aging; Multidisciplinary University Research Initiative; Jiangsu University; National Institutes of Health; National Science Foundation; Jiangsu Normal University; New York University Abu Dhabi; Agence Nationale de la Recherche","keywords":"Functional magnetic resonance imaging; Generalizability theory; Computer science; Psychology; Active listening; Functional neuroimaging; Neuroimaging; Cognitive neuroscience; Cognitive psychology; Cognition; Communication","score_opus":0.07188631091391356,"score_gpt":0.31624786866279,"score_spread":0.2443615577488764,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293463483","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.68801606,0.0038189113,0.02742459,0.0019530376,0.0003695118,0.0026399372,0.23487149,0.0034379365,0.037468616],"genre_scores_gemma":[0.6013915,0.0010410955,0.054815948,0.0011234956,0.00032443082,0.008941431,0.30781737,0.0014373736,0.023107348],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994087,0.00019678255,0.00006728961,0.00017738242,0.000098893106,0.000050921975],"domain_scores_gemma":[0.99799407,0.0008924864,0.0001290808,0.0004498772,0.0003903393,0.00014412699],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00093920354,0.00096424855,0.0005125827,0.0013622235,0.0013042332,0.0008534177,0.00095726567,0.0011968663,0.013853379],"category_scores_gemma":[0.003391991,0.00040194363,0.00036407827,0.00093367754,0.0009891518,0.00053199753,0.0014812964,0.00088408816,0.0037578659],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006545268,0.0018075176,0.023573924,0.0070349807,0.00075460074,0.033470795,0.017108552,0.0040842807,0.22302978,0.0098902825,0.38263583,0.29006425],"study_design_scores_gemma":[0.0011626177,0.0006061925,0.22994167,0.0004727583,0.0002794591,0.039964337,0.0036371408,0.00517227,0.029819949,0.0042302534,0.68435514,0.0003582366],"about_ca_topic_score_codex":0.015718302,"about_ca_topic_score_gemma":0.025378015,"teacher_disagreement_score":0.015718302,"about_ca_system_score_codex":0.00061974593,"about_ca_system_score_gemma":0.0009647944,"threshold_uncertainty_score":0.04634416},"labels":[],"label_agreement":null},{"id":"W4293463901","doi":"10.1038/s41597-022-01645-3","title":"Probabilistic atlas for the language network based on precision fMRI data from &gt;800 individuals","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neurobiology of Language and Bilingualism","field":"Neuroscience","cited_by":149,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"National Institute on Deafness and Other Communication Disorders; Eunice Kennedy Shriver National Institute of Child Health and Human Development; Simons Foundation; U.S. Department of Health and Human Services","keywords":"Interpretability; Computer science; Probabilistic logic; Voxel; Artificial intelligence; Atlas (anatomy); Natural language processing; Machine learning","score_opus":0.12054851064406184,"score_gpt":0.3402758178060862,"score_spread":0.21972730716202432,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293463901","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09278527,0.00033031713,0.8833876,0.00025311077,0.000069091264,0.00041360658,0.010537723,0.00287423,0.009349083],"genre_scores_gemma":[0.4586694,0.00036051797,0.52290165,0.00018257716,0.00004586082,0.0019261733,0.009233929,0.001642094,0.005037787],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992772,0.00019632478,0.00006602575,0.0002325685,0.00016794792,0.0000599479],"domain_scores_gemma":[0.998126,0.0008473977,0.0002504485,0.00040156016,0.00032418387,0.00005036757],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018193144,0.0003533867,0.0005007636,0.0017473067,0.000548451,0.001248754,0.000655922,0.0007726556,0.006697354],"category_scores_gemma":[0.0042711818,0.0004983047,0.0006416643,0.0018539935,0.00075765164,0.00070923293,0.00087000174,0.00082624267,0.0013381566],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001719518,0.0002659981,0.04175858,0.0014623438,0.0005935246,0.0018001013,0.003786508,0.20183566,0.23592232,0.082268976,0.034103263,0.39448327],"study_design_scores_gemma":[0.00025137243,0.00062448595,0.20557173,0.00021325666,0.00042483123,0.007604717,0.00070171006,0.4869524,0.07549443,0.096811526,0.12492432,0.00042526418],"about_ca_topic_score_codex":0.006208375,"about_ca_topic_score_gemma":0.010110906,"teacher_disagreement_score":0.006697354,"about_ca_system_score_codex":0.00096726324,"about_ca_system_score_gemma":0.0014591675,"threshold_uncertainty_score":0.02240491},"labels":[],"label_agreement":null},{"id":"W4295367167","doi":"10.1038/s41597-022-01665-z","title":"A georeferenced rRNA amplicon database of aquatic microbiomes from South America","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Microbial Community Ecology and Physiology","field":"Environmental Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Agencia Santafesina de Ciencia, Tecnología e Innovación; Consejo Nacional de Investigaciones Científicas y Técnicas; Agencia Nacional de Promoción Científica y Tecnológica; Fundação de Amparo à Pesquisa do Estado de São Paulo","keywords":"Georeference; Ecology; Geography; Microbiome; Biology; Biodiversity; Database; Physical geography; Computer science; Bioinformatics","score_opus":0.0461665178413495,"score_gpt":0.25641245677011926,"score_spread":0.21024593892876975,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4295367167","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21274583,0.003495152,0.00902217,0.00022595403,0.00007927838,0.0003193472,0.7666059,0.0016665517,0.00583986],"genre_scores_gemma":[0.05101354,0.0015345953,0.031313978,0.000103468075,0.00001979529,0.0004991189,0.9141675,0.0003190409,0.0010289499],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987901,0.00009299186,0.00026179216,0.0003658273,0.00031614047,0.00017315209],"domain_scores_gemma":[0.9979108,0.00025021762,0.0005132849,0.0003389652,0.0006862477,0.00030034533],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00134923,0.0007597455,0.0009015657,0.008889553,0.0013136299,0.0011701225,0.0009000465,0.0005878914,0.0031209798],"category_scores_gemma":[0.0029796232,0.0005038362,0.00064488273,0.011967198,0.00038114676,0.0010678946,0.0022881604,0.000752957,0.0025393278],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022997777,0.0004435077,0.15575533,0.010886794,0.00071671756,0.0019010596,0.0076733534,0.0022736795,0.46172172,0.0024735248,0.090039216,0.2638154],"study_design_scores_gemma":[0.00010227599,0.00020574493,0.4748389,0.0010088207,0.00044920636,0.0010499593,0.001633653,0.0016109842,0.031191058,0.0012144829,0.48655453,0.00014024819],"about_ca_topic_score_codex":0.012823725,"about_ca_topic_score_gemma":0.029668264,"teacher_disagreement_score":0.012823725,"about_ca_system_score_codex":0.0007919989,"about_ca_system_score_gemma":0.0056467364,"threshold_uncertainty_score":0.025498152},"labels":[],"label_agreement":null},{"id":"W4295426917","doi":"10.1038/s41597-022-01655-1","title":"Sharkipedia: a curated open access database of shark and ray life history traits and abundance time-series","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ichthyology and Marine Biology","field":"Environmental Science","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; Dalhousie University","funders":"Save Our Seas Foundation; California Department of Fish and Wildlife; National Science Foundation","keywords":"Macroevolution; Biology; Macroecology; IUCN Red List; Database; Threatened species; Population; Ecology; Biodiversity; Phylogenetics; Computer science","score_opus":0.06562540695180588,"score_gpt":0.2972119327077994,"score_spread":0.2315865257559935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4295426917","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040597618,0.00046438587,0.0032166932,0.00011214232,0.00008195662,0.00009824324,0.9862796,0.004083863,0.0016032723],"genre_scores_gemma":[0.008120219,0.00039437253,0.007494068,0.00012331361,0.00005785306,0.00027744795,0.9821571,0.0006028479,0.00077271],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983907,0.00013819648,0.00065812,0.00040973845,0.00026967906,0.00013354846],"domain_scores_gemma":[0.9927704,0.0014206943,0.0021144534,0.0014272847,0.0014007315,0.0008665138],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.002473789,0.0013573135,0.0014607796,0.012294983,0.0005649845,0.0020037328,0.002313195,0.0011311823,0.02187798],"category_scores_gemma":[0.0110089965,0.0007303411,0.0009372516,0.011737491,0.0003816876,0.002009377,0.0037484928,0.0009906802,0.014405837],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015747154,0.00017021968,0.048410513,0.0074443184,0.0009208107,0.0010873527,0.0009841854,0.0015911638,0.011409257,0.0057587563,0.81462353,0.106025115],"study_design_scores_gemma":[0.00032330703,0.00013537664,0.14456233,0.0012380135,0.00048202017,0.000711668,0.00038116085,0.002120723,0.0036738997,0.0037200223,0.8424412,0.00021040539],"about_ca_topic_score_codex":0.00862446,"about_ca_topic_score_gemma":0.012756034,"teacher_disagreement_score":0.9976868,"about_ca_system_score_codex":0.0009053702,"about_ca_system_score_gemma":0.0046246033,"threshold_uncertainty_score":0.07318914},"labels":[],"label_agreement":null},{"id":"W4296031138","doi":"10.1038/s41597-022-01682-y","title":"An Open MRI Dataset For Multiscale Neuroscience","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":111,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Queen's University; McGill University; Montreal Neurological Institute and Hospital","funders":"Centre Azrieli de recherche sur l'autisme, Institut et Hôpital Neurologiques de Montréal; Canadian Institutes of Health Research; National Institute of Biomedical Imaging and Bioengineering; Fonds de Recherche du Québec - Santé; Canadian Open Neuroscience Platform; Health Canada; Institute for Basic Science; National Institute of Mental Health; Hospital for Sick Children; National Research Foundation of Korea; Inha University; National Institutes of Health; Sick Kids Foundation; Canada First Research Excellence Fund; Canada Research Chairs; National Research Foundation; Natural Sciences and Engineering Research Council of Canada; Institute for Information and Communications Technology Promotion; Government of Canada; Savoy Foundation; China Scholarship Council; Ministry of Science and ICT, South Korea; Fondation Brain Canada; McGill University","keywords":"Open science; Neuroscience; Data science; Neuroinformatics; Computer science; Psychology; Physics","score_opus":0.24469456429015796,"score_gpt":0.3971177424460014,"score_spread":0.15242317815584344,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4296031138","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026599404,0.0005041959,0.0040849005,0.00037798434,0.00011925887,0.00012241672,0.98713636,0.0032703476,0.0017245868],"genre_scores_gemma":[0.0043437444,0.00017818133,0.0062573845,0.0001422459,0.000038608647,0.0003994951,0.98743653,0.00039097018,0.0008128529],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989059,0.00021417052,0.00014450403,0.00036458776,0.00023066741,0.00014010753],"domain_scores_gemma":[0.9983941,0.00034505333,0.00017427465,0.000586278,0.00033573265,0.00016462883],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0015815747,0.001689052,0.0014292268,0.0025010377,0.0009512219,0.0018592427,0.0037629425,0.0025431921,0.028650941],"category_scores_gemma":[0.006875441,0.0006685249,0.0016194953,0.004073191,0.0006556707,0.001055285,0.003312056,0.0018260924,0.03843526],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041198192,0.000130791,0.0036993201,0.0013002421,0.00027344402,0.0003236506,0.00014915675,0.0018828516,0.0027848603,0.002482648,0.9629276,0.023633502],"study_design_scores_gemma":[0.00058572815,0.000088572924,0.01635147,0.0004646791,0.00018279982,0.0010467984,0.00021303011,0.0046912776,0.002976866,0.010222195,0.9630378,0.00013866577],"about_ca_topic_score_codex":0.013903275,"about_ca_topic_score_gemma":0.03169455,"teacher_disagreement_score":0.99623704,"about_ca_system_score_codex":0.0009365576,"about_ca_system_score_gemma":0.002027641,"threshold_uncertainty_score":0.09584695},"labels":[],"label_agreement":null},{"id":"W4296407701","doi":"10.1038/s41597-022-01672-0","title":"Publisher Correction: Fundamental social motives measured across forty-two cultures in two waves","year":2022,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Cultural Differences and Values","field":"Psychology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; University of Waterloo","funders":"","keywords":"Geography; Sociology; Anthropology","score_opus":0.2000200098705693,"score_gpt":0.4606010152912551,"score_spread":0.26058100542068574,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4296407701","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017855205,0.0009415449,0.0023536389,0.057812136,0.9174723,0.000041256255,0.0053022048,0.000769209,0.013522195],"genre_scores_gemma":[0.07499388,0.0071379803,0.017467143,0.044701625,0.11412774,0.00041045656,0.015480346,0.0054306285,0.7202502],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9958001,0.00069037225,0.0008486939,0.00051270996,0.0019349427,0.00021319697],"domain_scores_gemma":[0.94732445,0.013172982,0.0020021277,0.0036991064,0.03276056,0.0010408956],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037656836,0.0017040442,0.0013772013,0.0037525836,0.004269894,0.0035038348,0.0022970845,0.0025914442,0.07436061],"category_scores_gemma":[0.080471046,0.0007668673,0.00078286265,0.0048396513,0.0018080325,0.0024119148,0.0020782016,0.0068728724,0.028843353],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002481054,0.000006759668,0.00026311597,0.00007496387,0.000006840223,0.00018742443,0.00021357583,0.00002389928,0.000048274193,0.0015901832,0.98894787,0.008612176],"study_design_scores_gemma":[0.000027061373,0.000021223828,0.0029543114,0.0004460105,0.00004483235,0.0008098264,0.0007673131,0.00026603628,0.0006788028,0.0028333666,0.99109113,0.00006013363],"about_ca_topic_score_codex":0.026003195,"about_ca_topic_score_gemma":0.029080853,"teacher_disagreement_score":0.07436061,"about_ca_system_score_codex":0.0029792802,"about_ca_system_score_gemma":0.0045260894,"threshold_uncertainty_score":0.248761},"labels":[],"label_agreement":null},{"id":"W4298394231","doi":"10.1038/s41597-022-01716-5","title":"Niche deconvolution of the glioblastoma proteome reveals a distinct infiltrative phenotype within the proneural transcriptomic subgroup","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Glioma Diagnosis and Treatment","field":"Medicine","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto; University Health Network","funders":"Canadian Institutes of Health Research; Terry Fox Foundation; Government of Canada; Princess Margaret Cancer Foundation","keywords":"Biology; Phenotype; Transcriptome; Glioblastoma; MYB; Cancer research; Glioma; Mesenchymal stem cell; Gene expression profiling; Gene; Transcription factor; Gene expression; Cell biology; Genetics","score_opus":0.04000432322437747,"score_gpt":0.280134834045855,"score_spread":0.24013051082147752,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4298394231","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99388415,0.00027601857,0.004784597,0.000025282638,0.000003373073,0.000006847708,0.0006825434,0.000076469005,0.00026080108],"genre_scores_gemma":[0.9930356,0.00022518108,0.0043518688,0.0000297703,0.0000051052357,0.0000184805,0.0020102812,0.000026487776,0.00029715177],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9998957,0.000013581812,0.0000063649636,0.0000346015,0.000026032043,0.000023813827],"domain_scores_gemma":[0.99990654,0.000019833155,0.000027037602,0.0000111444415,0.000017547136,0.00001784685],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002723462,0.00025889053,0.00022149809,0.0007324528,0.00017473116,0.00031513392,0.0000990946,0.00013600313,0.000503426],"category_scores_gemma":[0.00038231906,0.00008913454,0.00026211934,0.00040305502,0.00015014231,0.00017339345,0.00033523346,0.00018197822,0.00022914376],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043414332,0.00004253398,0.03875477,0.00007330027,0.00006408287,0.00014461021,0.00015223594,0.000663211,0.94191885,0.00019891244,0.00026208485,0.017291276],"study_design_scores_gemma":[0.000022057118,0.00035896516,0.73254776,0.000024484583,0.00013280316,0.001833697,0.00042819235,0.024383722,0.23534055,0.0012944151,0.0035956164,0.000037841433],"about_ca_topic_score_codex":0.0008575562,"about_ca_topic_score_gemma":0.0015910199,"teacher_disagreement_score":0.0008575562,"about_ca_system_score_codex":0.00011897083,"about_ca_system_score_gemma":0.0001916532,"threshold_uncertainty_score":0.0017051101},"labels":[],"label_agreement":null},{"id":"W4301399201","doi":"10.1038/s41597-022-01704-9","title":"A comprehensive database of amphibian heat tolerance","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":56,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"National Research, Development and Innovation Office; Agencia Nacional de Investigación y Desarrollo; Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; National Natural Science Foundation of China; University of Queensland; Australian Research Council; University of New South Wales; Department of Education and Training; Deutsche Forschungsgemeinschaft","keywords":"Ectotherm; Amphibian; Evolvability; Trait; Vulnerability (computing); Resource (disambiguation); Ecology; Biology; Environmental science; Environmental resource management; Computer science; Evolutionary biology","score_opus":0.08464849075418321,"score_gpt":0.2873495062304601,"score_spread":0.20270101547627692,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301399201","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006073796,0.0033311949,0.0015981358,0.00014418038,0.00003684273,0.00004749325,0.9822352,0.0006018973,0.005931153],"genre_scores_gemma":[0.012408008,0.0027061703,0.0044256905,0.00011980607,0.000035525896,0.00023968241,0.9786386,0.00017836688,0.0012481215],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99851996,0.00019817299,0.0004752758,0.00038552415,0.00030380482,0.00011728424],"domain_scores_gemma":[0.994065,0.0018929823,0.0013559276,0.00070015166,0.0015739554,0.00041212942],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011923374,0.0008236431,0.0010399529,0.015435469,0.00049411884,0.001469303,0.0013341232,0.0007257468,0.032549176],"category_scores_gemma":[0.008523069,0.00040978793,0.00092930964,0.020444404,0.0002905722,0.0016576587,0.0013311603,0.00096577057,0.017663416],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006162561,0.00014377694,0.0665102,0.02457635,0.00069488166,0.0005067558,0.0013357762,0.004191894,0.005355967,0.007926203,0.6485797,0.23956226],"study_design_scores_gemma":[0.000044343516,0.000045888602,0.06762774,0.0018198952,0.00020847,0.0003683687,0.00033536484,0.00058981753,0.0009611324,0.0020473965,0.9258887,0.0000629978],"about_ca_topic_score_codex":0.0064620967,"about_ca_topic_score_gemma":0.009951267,"teacher_disagreement_score":0.032549176,"about_ca_system_score_codex":0.0007323728,"about_ca_system_score_gemma":0.0020534787,"threshold_uncertainty_score":0.10888785},"labels":[],"label_agreement":null},{"id":"W4303984248","doi":"10.1038/s41597-022-01693-9","title":"Long-term annual soil nitrogen surplus across Europe (1850–2019)","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil and Water Nutrient Dynamics","field":"Environmental Science","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Bundesministerium für Bildung und Forschung","keywords":"Agriculture; Water Framework Directive; Environmental science; Term (time); Eutrophication; Soil water; Scale (ratio); Equator; Water quality; Natural resource economics; Environmental resource management; Hydrology (agriculture); Geography; Nutrient; Economics; Ecology; Soil science; Latitude; Biology","score_opus":0.02268357152617,"score_gpt":0.2624966884403882,"score_spread":0.2398131169142182,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4303984248","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8648077,0.003982811,0.0047753197,0.0005440114,0.00009917563,0.00001638421,0.1203638,0.00033313,0.005077599],"genre_scores_gemma":[0.89016926,0.0018257585,0.004563189,0.00012433346,0.000048443875,0.00003458867,0.102093585,0.00007483661,0.0010658904],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.999734,0.000032691143,0.000037017642,0.000117027834,0.000044677443,0.00003456022],"domain_scores_gemma":[0.99962986,0.00005314545,0.000101795835,0.00005303619,0.00013103294,0.000031047635],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00081649335,0.00029461423,0.00030504342,0.0020279826,0.00016707473,0.0007491098,0.0003555815,0.0004225671,0.0010351988],"category_scores_gemma":[0.0011292321,0.0001338622,0.00059301907,0.0035984737,0.0003036246,0.0007198763,0.00069210364,0.00019481333,0.00030490878],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031900406,0.00006819306,0.8253189,0.0009536094,0.0014995865,0.0007920336,0.0006264922,0.06319407,0.0034317877,0.005022445,0.03558565,0.06318827],"study_design_scores_gemma":[0.000037377686,0.000033950302,0.88705206,0.0002517436,0.00021833247,0.00034940275,0.00051112636,0.01702306,0.001325331,0.0018567533,0.0912889,0.000052011797],"about_ca_topic_score_codex":0.03492419,"about_ca_topic_score_gemma":0.028913517,"teacher_disagreement_score":0.03492419,"about_ca_system_score_codex":0.0006292193,"about_ca_system_score_gemma":0.000462522,"threshold_uncertainty_score":0.069441795},"labels":[],"label_agreement":null},{"id":"W4304783064","doi":"10.1038/s41597-022-01695-7","title":"An analysis-ready and quality controlled resource for pediatric brain white-matter research","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université de Montréal; Institut Universitaire en Santé Mentale de Québec; Western University; Polytechnique Montréal; Concordia University; University of Toronto","funders":"Child Mind Institute; University of Pennsylvania; National Institute of Biomedical Imaging and Bioengineering; National Institute of Mental Health; U.S. Department of Health and Human Services","keywords":"White matter; Quality (philosophy); Computer science; Medicine; Physics","score_opus":0.3514407486686759,"score_gpt":0.5218421750821093,"score_spread":0.17040142641343337,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4304783064","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05154036,0.0024652416,0.37607276,0.005547233,0.0012535503,0.0065098368,0.4683323,0.068557896,0.01972077],"genre_scores_gemma":[0.09046646,0.0008645186,0.48309982,0.0011537786,0.0006028876,0.009598222,0.39452657,0.0149878245,0.0046999073],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9816243,0.0065564425,0.0026267932,0.0035908835,0.0047900253,0.0008114847],"domain_scores_gemma":[0.86573744,0.03759652,0.01082791,0.040171087,0.03704304,0.008624049],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03401476,0.0013038141,0.0016690205,0.006849507,0.0016857866,0.0030016834,0.004588642,0.0012283683,0.030278595],"category_scores_gemma":[0.09993859,0.0011310385,0.0014880027,0.0062017976,0.0016375912,0.0035132992,0.00872908,0.002352397,0.019685507],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004593239,0.0010785243,0.047526892,0.0033039688,0.00085574656,0.001583197,0.0022216982,0.0066134003,0.03577932,0.016229007,0.5712679,0.30894703],"study_design_scores_gemma":[0.0019431453,0.00080576405,0.07305292,0.0019194899,0.0006714873,0.001268965,0.000980023,0.024882717,0.063025735,0.025917284,0.8050598,0.0004726678],"about_ca_topic_score_codex":0.0094089275,"about_ca_topic_score_gemma":0.013223936,"teacher_disagreement_score":0.03401476,"about_ca_system_score_codex":0.001992706,"about_ca_system_score_gemma":0.013449852,"threshold_uncertainty_score":0.17988938},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"bench_or_experimental","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4306173614","doi":"10.1038/s41597-022-01737-0","title":"OSeMOSYS Global, an open-source, open data global electricity system model generator","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Integrated Energy Systems Optimization","field":"Engineering","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Sciences and Engineering Research Council of Canada; University College Cork; Loughborough University; Imperial College London; Science Foundation Ireland; Government of Canada; Government of the United Kingdom; Foreign, Commonwealth and Development Office; Colorado School of Mines","keywords":"Open source; Open data; Electricity; Open system (computing); Computer science; Generator (circuit theory); Electricity system; World Wide Web; Electricity generation; Operating system; Engineering; Electrical engineering; Software; Physics; Power (physics)","score_opus":0.06749732815584851,"score_gpt":0.2947616454382479,"score_spread":0.22726431728239938,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4306173614","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014110476,0.0005922229,0.39447942,0.002004934,0.0010113426,0.00081765425,0.28586718,0.23946361,0.061653245],"genre_scores_gemma":[0.103843875,0.0014600973,0.2942982,0.0010574101,0.0002359482,0.0020269481,0.4895598,0.08224541,0.025272379],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99937505,0.00016107087,0.000062111685,0.0001176043,0.00023030196,0.000053957803],"domain_scores_gemma":[0.99784184,0.00083309127,0.0001497882,0.0005430922,0.00045205426,0.0001802366],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0020324567,0.0013202865,0.00085595006,0.0016307506,0.00053813227,0.002203866,0.0028519605,0.00095103163,0.03108723],"category_scores_gemma":[0.007695378,0.00080002414,0.001539559,0.0024311973,0.00041235366,0.0033113349,0.0019320567,0.0017084944,0.013331381],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038157456,0.00022467677,0.00658776,0.0014580898,0.00032449787,0.00043537535,0.00050118635,0.2063548,0.0041265152,0.05044458,0.6540086,0.07515229],"study_design_scores_gemma":[0.00041948215,0.00006629353,0.0017678789,0.000226807,0.0000709639,0.00013790338,0.00016541005,0.33087674,0.004520981,0.048666064,0.61295015,0.00013123377],"about_ca_topic_score_codex":0.009133666,"about_ca_topic_score_gemma":0.013705087,"teacher_disagreement_score":0.99714804,"about_ca_system_score_codex":0.0009770246,"about_ca_system_score_gemma":0.002712575,"threshold_uncertainty_score":0.10399711},"labels":[],"label_agreement":null},{"id":"W4306292496","doi":"10.1038/s41597-022-01723-6","title":"GABB: A global dataset of alpine breeding birds and their ecological traits","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; Environment and Climate Change Canada","keywords":"Ecology; Geography; Biology","score_opus":0.07948226285215344,"score_gpt":0.2861313746013772,"score_spread":0.2066491117492238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4306292496","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014439958,0.000895279,0.0004642089,0.000074253374,0.000019823374,0.00002902958,0.9824432,0.00035226348,0.0012819445],"genre_scores_gemma":[0.014346943,0.00026631326,0.0019884708,0.000049146525,0.000010681095,0.000107452615,0.98276067,0.000055916596,0.0004144333],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996025,0.000069668175,0.000062300365,0.00013839973,0.00007428484,0.000052840165],"domain_scores_gemma":[0.99894637,0.0001527348,0.00031229918,0.00016401529,0.00026589274,0.0001586362],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00050032325,0.00085776666,0.0007689353,0.0033481107,0.00041234464,0.00092561543,0.00085424184,0.0006202916,0.0068797953],"category_scores_gemma":[0.0018954293,0.00030054757,0.0006177159,0.004628411,0.00017243209,0.0004990356,0.0012397079,0.000724515,0.0053308276],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005392718,0.00010498541,0.22784051,0.0046683284,0.0009994347,0.0005151085,0.0011474558,0.003357517,0.005989881,0.0021615962,0.68683046,0.06584551],"study_design_scores_gemma":[0.000174914,0.000056372144,0.4118453,0.00063632644,0.00015019874,0.0003520206,0.00040779533,0.0014516497,0.00060533563,0.00076564145,0.58350104,0.000053346066],"about_ca_topic_score_codex":0.038507387,"about_ca_topic_score_gemma":0.06134575,"teacher_disagreement_score":0.038507387,"about_ca_system_score_codex":0.00056934566,"about_ca_system_score_gemma":0.0008952031,"threshold_uncertainty_score":0.07656652},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.20529560955855886,"score_gpt":0.40425551406959226,"score_spread":0.1989599045110334,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4306292873","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017084599,0.00087867974,0.66775817,0.028295929,0.0037183508,0.005880868,0.04946842,0.08694684,0.13996819],"genre_scores_gemma":[0.07575819,0.0007995475,0.7387382,0.004517435,0.0008311382,0.006513831,0.097085655,0.03928371,0.036472227],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.91023684,0.030546898,0.009389677,0.0071207318,0.039708678,0.002997251],"domain_scores_gemma":[0.62438846,0.07838606,0.008236312,0.20752642,0.071591906,0.009870835],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.17980908,0.0010830201,0.0013783679,0.011260488,0.005242207,0.023583353,0.012812562,0.0040998165,0.020609196],"category_scores_gemma":[0.26025116,0.002374066,0.0023039838,0.017051253,0.006363176,0.025224643,0.019305468,0.007780536,0.012405726],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048261014,0.0005249726,0.009650652,0.0013503903,0.00019957435,0.0008005151,0.016768912,0.016036522,0.0037524437,0.27711746,0.42307657,0.25023943],"study_design_scores_gemma":[0.00009489205,0.000057978577,0.0024634292,0.0007339051,0.00003985573,0.00012297192,0.001252879,0.004674295,0.0032663269,0.048442483,0.9386331,0.00021782678],"about_ca_topic_score_codex":0.02191702,"about_ca_topic_score_gemma":0.014382166,"teacher_disagreement_score":0.98718745,"about_ca_system_score_codex":0.009247488,"about_ca_system_score_gemma":0.037451982,"threshold_uncertainty_score":0.9509327},"labels":[],"label_agreement":null},{"id":"W4306882708","doi":"10.1038/s41597-022-01714-7","title":"The Immune Signatures data resource, a compendium of systems vaccinology datasets","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Allergy and Infectious Diseases; Canadian Institutes of Health Research; National Institutes of Health; Division of Intramural Research, National Institute of Allergy and Infectious Diseases; U.S. Department of Health and Human Services","keywords":"Compendium; Reverse vaccinology; Data sharing; Systems biology; Immunogenicity; Computer science; Computational biology; Pooling; Data science; Immune system; Bioinformatics; Biology; Medicine; Immunology; Genome; Artificial intelligence; Geography; Gene","score_opus":0.035874715170587324,"score_gpt":0.2659537450915435,"score_spread":0.2300790299209562,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4306882708","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00047190182,0.00017246189,0.0022642228,0.00013148104,0.00003919222,0.00010787316,0.9950964,0.0008802781,0.0008362617],"genre_scores_gemma":[0.0015878208,0.000157861,0.0052485764,0.00015664654,0.000012748294,0.00057440135,0.99167866,0.00024074575,0.00034246972],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9964276,0.00088404317,0.0008517212,0.0007215092,0.00083695375,0.0002782418],"domain_scores_gemma":[0.98933804,0.0050033983,0.0014914669,0.001960461,0.0014179744,0.00078864367],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006732603,0.0013377232,0.0015531583,0.004584924,0.0008344897,0.0024815353,0.003314155,0.001606189,0.03172624],"category_scores_gemma":[0.023769703,0.0010235343,0.0019396943,0.0061559114,0.00051743665,0.0014170659,0.0035309754,0.0026342284,0.01908964],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00081094453,0.0001285202,0.014669494,0.0064289775,0.0007381185,0.0003017932,0.00026966687,0.0029628652,0.004700032,0.00909705,0.9243415,0.03555101],"study_design_scores_gemma":[0.00070922804,0.00011283148,0.015771652,0.00095838506,0.00029012654,0.000266808,0.0001437505,0.0014804478,0.0030170763,0.0096709635,0.96748215,0.000096618736],"about_ca_topic_score_codex":0.00742569,"about_ca_topic_score_gemma":0.01172135,"teacher_disagreement_score":0.03172624,"about_ca_system_score_codex":0.0013461899,"about_ca_system_score_gemma":0.005706562,"threshold_uncertainty_score":0.10613483},"labels":[],"label_agreement":null},{"id":"W4307565510","doi":"10.1038/s41597-022-01743-2","title":"Elaboration of a new framework for fine-grained epidemiological annotation","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Data-Driven Disease Surveillance","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université de Sherbrooke","funders":"Horizon 2020 Framework Programme; Fonds De La Recherche Scientifique - FNRS; FOD Volksgezondheid, Veiligheid van de Voedselketen en Leefmilieu; Centre de Coopération Internationale en Recherche Agronomique pour le Développement; Agence Nationale de la Recherche; European Commission","keywords":"Annotation; Computer science; Variety (cybernetics); Context (archaeology); Data science; Process (computing); Information retrieval; World Wide Web; Artificial intelligence; Geography","score_opus":0.1471525771918431,"score_gpt":0.40116059898123296,"score_spread":0.25400802178938986,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4307565510","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000888128,0.00020495622,0.9892812,0.0008616097,0.00014484643,0.00035961607,0.001768546,0.0046539595,0.0018371588],"genre_scores_gemma":[0.010726128,0.0001893162,0.9795415,0.0003516041,0.00015763612,0.00038878073,0.0061542066,0.00081009656,0.0016808618],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9888975,0.0033455547,0.0019337413,0.002855084,0.0024846736,0.0004834173],"domain_scores_gemma":[0.97695166,0.010363368,0.0014254655,0.0044962512,0.0058003226,0.00096291804],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.016172586,0.0020608068,0.0017841749,0.010555179,0.0034805422,0.008554767,0.0057348884,0.003448615,0.007406596],"category_scores_gemma":[0.024707993,0.0020133848,0.004505492,0.0071002766,0.0037815198,0.009745414,0.010376668,0.005839479,0.004457493],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003089092,0.0005242084,0.0045945146,0.0029697905,0.0004061774,0.0018982501,0.009145775,0.027463622,0.02405851,0.4444767,0.080726616,0.40342695],"study_design_scores_gemma":[0.00008599353,0.000121172634,0.0019150147,0.0010682321,0.00029424558,0.0010489898,0.0015747299,0.24266312,0.017099747,0.33348978,0.40035093,0.00028806244],"about_ca_topic_score_codex":0.024244592,"about_ca_topic_score_gemma":0.031430695,"teacher_disagreement_score":0.024244592,"about_ca_system_score_codex":0.0031750663,"about_ca_system_score_gemma":0.0075910725,"threshold_uncertainty_score":0.085529864},"labels":[],"label_agreement":null},{"id":"W4307891591","doi":"10.1038/s41597-022-01770-z","title":"Publisher Correction: An analysis-ready and quality controlled resource for pediatric brain white-matter research","year":2022,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Radiomics and Machine Learning in Medical Imaging","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université de Montréal; Institut Universitaire en Santé Mentale de Québec; Western University; Polytechnique Montréal; Concordia University; University of Toronto","funders":"","keywords":"White matter; Resource (disambiguation); Quality (philosophy); White paper; Computer science; Data science; Medicine; Political science; Physics","score_opus":0.10684226474260987,"score_gpt":0.43110366615052953,"score_spread":0.32426140140791965,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4307891591","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0002650674,0.0009920327,0.0046990355,0.05933104,0.9028467,0.00019507343,0.016133336,0.004769365,0.010768285],"genre_scores_gemma":[0.010981671,0.006229992,0.034786817,0.090218514,0.23035444,0.0014404679,0.03871403,0.019698348,0.56757575],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9837578,0.0022322377,0.0032457006,0.0012661107,0.008787566,0.0007105439],"domain_scores_gemma":[0.7601626,0.044327393,0.008815794,0.02129026,0.16106887,0.0043351017],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010997306,0.002050845,0.002553617,0.009720788,0.0033835017,0.008448954,0.0051554297,0.006070397,0.18787031],"category_scores_gemma":[0.18288167,0.00169716,0.0016760335,0.007534996,0.002751313,0.0033226176,0.0032198492,0.01150822,0.13560149],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017314838,0.0000041109365,0.000043660875,0.000070234855,0.0000070955666,0.000057586924,0.0000100595225,0.000018438739,0.000028432925,0.0003610323,0.99517053,0.004211538],"study_design_scores_gemma":[0.000043439115,0.000011938868,0.00055058656,0.00030851676,0.000031863066,0.00019896272,0.00005526903,0.00018073071,0.00037521948,0.001503018,0.996705,0.000035402278],"about_ca_topic_score_codex":0.017975736,"about_ca_topic_score_gemma":0.028083801,"teacher_disagreement_score":0.18787031,"about_ca_system_score_codex":0.0040206905,"about_ca_system_score_gemma":0.013396586,"threshold_uncertainty_score":0.6284888},"labels":[],"label_agreement":null},{"id":"W4308252523","doi":"10.1038/s41597-022-01806-4","title":"Pre- and post-surgery brain tumor multimodal magnetic resonance imaging data optimized for large scale computational modelling","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Libin Cardiovascular Institute of Alberta; University of Calgary","funders":"Horizon 2020 Framework Programme; Universiteit Gent; Belgian Federal Science Policy Office; Deutsche Forschungsgemeinschaft; European Commission","keywords":"Magnetic resonance imaging; Voxel; Functional magnetic resonance imaging; Brain tumor; Diffusion MRI; Blood-oxygen-level dependent; Neuroimaging; Cerebral blood flow; Computer science; Scale (ratio); Medicine; Radiology; Artificial intelligence; Psychology; Neuroscience; Pathology; Internal medicine; Physics","score_opus":0.06762944165670765,"score_gpt":0.29375143796004594,"score_spread":0.2261219963033383,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4308252523","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8749657,0.00062736124,0.0598877,0.00070698553,0.00012983906,0.00029808676,0.059840962,0.0018063721,0.0017369697],"genre_scores_gemma":[0.86481535,0.00035106446,0.035736434,0.000117674885,0.00004941988,0.00051622227,0.09674043,0.00022268966,0.0014506961],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99978906,0.00004227236,0.000016600352,0.00007812658,0.00003909439,0.00003484476],"domain_scores_gemma":[0.9994916,0.00019863104,0.00005079557,0.00013076815,0.00009443142,0.000033825432],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00047837134,0.00073094346,0.0006633271,0.00065231026,0.00027928627,0.000600993,0.0009620058,0.0010865306,0.0021486466],"category_scores_gemma":[0.0020289796,0.0003268278,0.0008705478,0.000702493,0.00048469935,0.00034052623,0.0004920147,0.0005987256,0.0008192241],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018246957,0.0011996189,0.037019413,0.0007821774,0.0006666564,0.0025542178,0.00021675474,0.80160576,0.04690816,0.0027046413,0.023665432,0.08085248],"study_design_scores_gemma":[0.00024875478,0.0007158643,0.09635823,0.000059354505,0.00021938552,0.0018943488,0.00022500292,0.8566685,0.022340141,0.00695918,0.014128188,0.00018312698],"about_ca_topic_score_codex":0.007823606,"about_ca_topic_score_gemma":0.008166686,"teacher_disagreement_score":0.007823606,"about_ca_system_score_codex":0.0004071273,"about_ca_system_score_gemma":0.00076507934,"threshold_uncertainty_score":0.015556157},"labels":[],"label_agreement":null},{"id":"W4308509762","doi":"10.1038/s41597-022-01558-1","title":"A curated collection of human vaccination response signatures","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"vaccines and immunoinformatics approaches","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of Allergy and Infectious Diseases; Canadian Institutes of Health Research; National Institutes of Health; Government of Canada; U.S. Department of Health and Human Services","keywords":"Ontology; Metadata; Computer science; Context (archaeology); Profiling (computer programming); UniProt; Dashboard; Computational biology; Data science; World Wide Web; Biology; Gene","score_opus":0.03444902112042952,"score_gpt":0.28963945436500976,"score_spread":0.25519043324458024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4308509762","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04031819,0.008710365,0.021678139,0.0010851737,0.00044493072,0.0005513997,0.90789557,0.006121781,0.013194446],"genre_scores_gemma":[0.047126446,0.0028599284,0.036967214,0.00078404567,0.0001500722,0.0007346056,0.9065419,0.00070645555,0.004129338],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9973909,0.0004084488,0.0004984498,0.0007530435,0.00079987023,0.00014933904],"domain_scores_gemma":[0.9900608,0.0034031032,0.001378289,0.0019381536,0.002574887,0.00064468355],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025809202,0.0008900851,0.0014675027,0.008083813,0.00093441666,0.001488537,0.00106533,0.0012584033,0.0089368485],"category_scores_gemma":[0.011614877,0.00039444453,0.0010480759,0.0068949615,0.00039757832,0.0007253243,0.0014718741,0.0009568772,0.007679202],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025578488,0.0005368035,0.054508537,0.020639526,0.0012772722,0.0038819057,0.001174603,0.0035701257,0.14261682,0.004549566,0.4553354,0.30935147],"study_design_scores_gemma":[0.00023470446,0.00041925936,0.1171373,0.0016722956,0.0010803409,0.003961715,0.0003646577,0.003220979,0.05078471,0.003782819,0.81718105,0.00016016449],"about_ca_topic_score_codex":0.005248831,"about_ca_topic_score_gemma":0.011213388,"teacher_disagreement_score":0.0089368485,"about_ca_system_score_codex":0.000887273,"about_ca_system_score_gemma":0.004987319,"threshold_uncertainty_score":0.029896736},"labels":[],"label_agreement":null},{"id":"W4308886066","doi":"10.1038/s41597-022-01796-3","title":"Near-real-time daily estimates of fossil fuel CO2 emissions from major high-emission cities in China","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric and Environmental Gas Dynamics","field":"Environmental Science","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; Hudbay Minerals (Canada); University of Toronto","funders":"National Natural Science Foundation of China","keywords":"China; Environmental science; Fossil fuel; Natural resource economics; Geography; Ecology; Economics; Biology; Archaeology","score_opus":0.010204367170759936,"score_gpt":0.22089903088848123,"score_spread":0.2106946637177213,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4308886066","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91281635,0.00015081302,0.0015987311,0.00010950869,0.000018057091,0.000033692453,0.08311084,0.00033578405,0.0018261286],"genre_scores_gemma":[0.8685166,0.00016272342,0.0023479136,0.000030044164,0.000014595804,0.0000638796,0.12800266,0.000032129992,0.00082946493],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997502,0.000023182221,0.0000234868,0.00008206319,0.00007556678,0.00004556879],"domain_scores_gemma":[0.9994979,0.000050956347,0.00008421777,0.000103950144,0.00019816136,0.00006478079],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004018239,0.000690655,0.00035744946,0.0016276358,0.0002968917,0.0005247871,0.00051163766,0.00034938045,0.0010567362],"category_scores_gemma":[0.0006056432,0.00022400355,0.0004230205,0.0026494837,0.00021092487,0.00049528,0.00051360484,0.00024225317,0.0003277458],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003060647,0.00024705022,0.7525861,0.00040881097,0.000436612,0.00045968892,0.0002914591,0.17243244,0.008307706,0.0012307377,0.020870192,0.04242307],"study_design_scores_gemma":[0.00004632418,0.00004010316,0.86105853,0.000031005282,0.00007800594,0.000054964967,0.00020786693,0.12148108,0.004471929,0.00037354478,0.012089596,0.00006708494],"about_ca_topic_score_codex":0.11664595,"about_ca_topic_score_gemma":0.1402444,"teacher_disagreement_score":0.11664595,"about_ca_system_score_codex":0.0010639522,"about_ca_system_score_gemma":0.0011829557,"threshold_uncertainty_score":0.23193395},"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,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.4877564196382878,"score_gpt":0.4680528498693035,"score_spread":0.019703569768984297,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309046817","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015999338,0.00024204457,0.9428819,0.0062055252,0.00025672285,0.000633333,0.0009530717,0.0013998376,0.03142824],"genre_scores_gemma":[0.20377667,0.0003316218,0.7814988,0.0012377911,0.00025518905,0.0019275277,0.002075334,0.0012556288,0.0076413783],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9172866,0.040287424,0.008117698,0.013017415,0.018326778,0.0029640668],"domain_scores_gemma":[0.7327711,0.08642834,0.015692228,0.13434185,0.02577109,0.0049954676],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.090589345,0.0010398276,0.0015619731,0.007838348,0.005658379,0.024395704,0.0071273176,0.0064263516,0.006507533],"category_scores_gemma":[0.2189975,0.0019350465,0.0035208077,0.007871805,0.014321614,0.047935434,0.015345435,0.0060695545,0.002690482],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000032065676,0.000049870087,0.0024634053,0.000058644993,0.00003653452,0.00006602002,0.0014628477,0.007711485,0.00031855903,0.97105473,0.0023948296,0.01435101],"study_design_scores_gemma":[0.00003737471,0.000030006117,0.00037876866,0.000124133,0.000028438035,0.00007226066,0.00050605735,0.020497805,0.00087011047,0.93744326,0.03996847,0.00004320518],"about_ca_topic_score_codex":0.011455981,"about_ca_topic_score_gemma":0.010737928,"teacher_disagreement_score":0.99287266,"about_ca_system_score_codex":0.008712525,"about_ca_system_score_gemma":0.015782598,"threshold_uncertainty_score":0.47908795},"labels":[],"label_agreement":null},{"id":"W4309333099","doi":"10.1038/s41597-022-01817-1","title":"A biomechanics dataset of healthy human walking at various speeds, step lengths and step widths","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Balance, Gait, and Falls Prevention","field":"Health Professions","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Biomechanics; Inverse dynamics; Ground reaction force; Motion capture; Gait; Computer science; Preferred walking speed; Treadmill; Simulation; Motion analysis; Gait analysis; Torque; Kinematics; Mathematics; Motion (physics); Physical medicine and rehabilitation; Artificial intelligence; Physics; Medicine; Anatomy; Physical therapy","score_opus":0.0758115345077446,"score_gpt":0.39332829701896466,"score_spread":0.3175167625112201,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309333099","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.27573466,0.0014230458,0.015336252,0.0004856678,0.00029993136,0.0006457964,0.6991264,0.0019471054,0.0050011524],"genre_scores_gemma":[0.1778526,0.00043443977,0.0105400635,0.000195379,0.00011341646,0.0012336174,0.8064137,0.00010406746,0.0031126295],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9996649,0.000058683476,0.000051562372,0.000096766904,0.000088797286,0.00003927038],"domain_scores_gemma":[0.999186,0.00016597564,0.00007655706,0.00014787485,0.0003267165,0.000096901684],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039393088,0.0008475092,0.0007452793,0.0011687374,0.00029558857,0.0003661416,0.0008424413,0.0011417293,0.0053945286],"category_scores_gemma":[0.0016601653,0.00018895323,0.0005644339,0.0009182754,0.00027140602,0.00025320175,0.00084872916,0.00038387155,0.003703931],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0042222575,0.0028673708,0.13410358,0.00444232,0.0011367899,0.0020167914,0.00058806327,0.015766378,0.034619633,0.0011144863,0.5053241,0.29379818],"study_design_scores_gemma":[0.00070323166,0.0017221145,0.8362963,0.0007738983,0.00030305499,0.0023673035,0.00066160224,0.02387445,0.0074451775,0.0024946355,0.12310949,0.00024875117],"about_ca_topic_score_codex":0.0038630408,"about_ca_topic_score_gemma":0.011988059,"teacher_disagreement_score":0.0053945286,"about_ca_system_score_codex":0.00014778675,"about_ca_system_score_gemma":0.00040172297,"threshold_uncertainty_score":0.018046439},"labels":[],"label_agreement":null},{"id":"W4309513471","doi":"10.1038/s41597-022-01816-2","title":"Author Correction: An analysis-ready and quality controlled resource for pediatric brain white-matter research","year":2022,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Neonatal and fetal brain pathology","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université de Montréal; Institut Universitaire en Santé Mentale de Québec; Western University; Polytechnique Montréal; Concordia University; University of Toronto","funders":"","keywords":"White matter; White paper; Quality (philosophy); Resource (disambiguation); White (mutation); Medicine; Computer science; Geography; Biology; Physics; Archaeology","score_opus":0.15712393059970867,"score_gpt":0.4317657935942526,"score_spread":0.27464186299454396,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309513471","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00047091258,0.0010332336,0.0076068835,0.07874796,0.87499154,0.00031732093,0.024105106,0.005406621,0.007320441],"genre_scores_gemma":[0.023355583,0.0067647817,0.078157105,0.16292338,0.2349829,0.0026107982,0.05180586,0.028092628,0.411307],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97792256,0.004439126,0.0048993677,0.0020727203,0.009610337,0.001055846],"domain_scores_gemma":[0.7355389,0.07035312,0.011064081,0.02260447,0.15477844,0.005660931],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015378903,0.0020360267,0.0021663206,0.007886073,0.0030378886,0.006826641,0.0042130747,0.005143228,0.14500819],"category_scores_gemma":[0.25450864,0.0013917476,0.0016180382,0.0054847263,0.0026477869,0.0024688342,0.0037044568,0.009740385,0.0842087],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004225563,0.000005720047,0.00010602882,0.00014168146,0.000015065458,0.000090358626,0.000023564218,0.000026802201,0.000053974025,0.000533372,0.99374324,0.0052180183],"study_design_scores_gemma":[0.00008912626,0.000015829633,0.000828937,0.00054569496,0.0000626999,0.00035575568,0.000087307606,0.0002566664,0.00067825615,0.00248589,0.9945439,0.00005001166],"about_ca_topic_score_codex":0.013932954,"about_ca_topic_score_gemma":0.02006913,"teacher_disagreement_score":0.14500819,"about_ca_system_score_codex":0.0037878319,"about_ca_system_score_gemma":0.0150617305,"threshold_uncertainty_score":0.48510075},"labels":[],"label_agreement":null},{"id":"W4309625182","doi":"10.1038/s41597-022-01812-6","title":"An expert curated global legume checklist improves the accuracy of occurrence, biodiversity and taxonomic data","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant and Fungal Species Descriptions","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Université de Montréal; Royal Botanical Gardens, Kew","keywords":"Checklist; Taxon; Taxonomic rank; Biodiversity; Biology; Global biodiversity; Ecology","score_opus":0.06531958014189927,"score_gpt":0.29584527117678433,"score_spread":0.23052569103488507,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309625182","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13650815,0.008145636,0.277292,0.017595043,0.010059262,0.009152693,0.40255168,0.019621206,0.11907431],"genre_scores_gemma":[0.091783166,0.0024756517,0.5891775,0.0033365248,0.0009655341,0.0042590355,0.286642,0.004157688,0.0172029],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9762824,0.00817543,0.004790342,0.0019339286,0.007859303,0.0009585097],"domain_scores_gemma":[0.84266245,0.02074204,0.016680388,0.022310035,0.09196902,0.005636064],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03744607,0.0011815481,0.0012943372,0.034770336,0.003001625,0.0046240017,0.0031761655,0.001282259,0.020621942],"category_scores_gemma":[0.093131036,0.0006708474,0.00097116106,0.01603439,0.0010767941,0.004090332,0.005743454,0.0020436642,0.012124503],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033012865,0.0001984583,0.043790404,0.004586087,0.00016587197,0.0009948527,0.0055150576,0.0008676983,0.023788206,0.0067437,0.4790177,0.43400177],"study_design_scores_gemma":[0.000050298648,0.000092661976,0.05994301,0.0023860661,0.00010732333,0.00062470866,0.0018386799,0.0010963402,0.003687263,0.002366114,0.9276642,0.0001432826],"about_ca_topic_score_codex":0.016589528,"about_ca_topic_score_gemma":0.033046596,"teacher_disagreement_score":0.03744607,"about_ca_system_score_codex":0.0033610607,"about_ca_system_score_gemma":0.015389001,"threshold_uncertainty_score":0.19803607},"labels":[],"label_agreement":null},{"id":"W4310060332","doi":"10.1038/s41597-022-01833-1","title":"TractoInferno - A large-scale, open-source, multi-site database for machine learning dMRI tractography","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Hôpital du Sacré-Cœur de Montréal; Université de Sherbrooke","funders":"Fonds de recherche du Québec – Nature et technologies; Compute Canada; Wellcome Trust","keywords":"Computer science; Tractography; Scale (ratio); Database; Artificial intelligence; Diffusion MRI; Cartography; Magnetic resonance imaging; Medicine; Geography","score_opus":0.17628452514973905,"score_gpt":0.40893639807620985,"score_spread":0.2326518729264708,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4310060332","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03144671,0.0033954068,0.17795327,0.0005302913,0.0004997774,0.0007537286,0.50942016,0.26459596,0.011404686],"genre_scores_gemma":[0.066949844,0.001058709,0.11020593,0.00032255377,0.00009971068,0.0011074381,0.7912237,0.023853337,0.0051788115],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99882776,0.00015277637,0.00014139374,0.00041051384,0.00036936344,0.00009823758],"domain_scores_gemma":[0.9967018,0.0008929997,0.00027903833,0.0012440505,0.00053101406,0.00035118105],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0021622488,0.0021808667,0.0019706658,0.0037634843,0.0007819169,0.0027087412,0.0042830845,0.0018499081,0.02711596],"category_scores_gemma":[0.008543902,0.0010882088,0.0016188486,0.0035646136,0.0005713143,0.0030231632,0.0029997502,0.0015239975,0.022153974],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018869534,0.00031313385,0.008408916,0.0028176026,0.00078450766,0.0010188412,0.00042795442,0.01888975,0.0103692375,0.007852771,0.7808417,0.16638862],"study_design_scores_gemma":[0.0012185171,0.00051863876,0.021411354,0.0007465009,0.00036347492,0.0040137656,0.0003172928,0.13655634,0.034973707,0.027416466,0.77188176,0.0005821644],"about_ca_topic_score_codex":0.0065324316,"about_ca_topic_score_gemma":0.01252851,"teacher_disagreement_score":0.9957169,"about_ca_system_score_codex":0.0009335543,"about_ca_system_score_gemma":0.001974867,"threshold_uncertainty_score":0.09071195},"labels":[],"label_agreement":null},{"id":"W4310480951","doi":"10.1038/s41597-022-01842-0","title":"HeliantHOME, a public and centralized database of phenotypic sunflower data","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Sunflower and Safflower Cultivation","field":"Agricultural and Biological Sciences","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria; University of British Columbia","funders":"Leibniz-Rechenzentrum; Leibniz-Gemeinschaft; Bayerische Akademie der Wissenschaften; Genome British Columbia; Technische Universität München; Genome Canada; National Science Foundation","keywords":"Domestication; Biology; Genomics; Adaptation (eye); Helianthus annuus; Sunflower; Helianthus; Population; Ranking (information retrieval); Biotechnology; Genome; Genetics; Computer science; Information retrieval; Gene; Agronomy","score_opus":0.1199364605218563,"score_gpt":0.27427903573872975,"score_spread":0.15434257521687345,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4310480951","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015795492,0.00078905595,0.009970061,0.00017897654,0.00009141454,0.00013991514,0.94820154,0.020548413,0.0042850566],"genre_scores_gemma":[0.014693995,0.00029002083,0.010957956,0.00006923815,0.000030824147,0.00018903267,0.9718032,0.0010704927,0.0008953663],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989881,0.000092281734,0.000174106,0.00036247427,0.00029401566,0.000088953726],"domain_scores_gemma":[0.99767894,0.00054434955,0.00046662503,0.00057232974,0.0003490716,0.0003888056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019360764,0.0013461495,0.00110833,0.0060080816,0.00092630053,0.0017602552,0.0020361624,0.000984473,0.012666541],"category_scores_gemma":[0.004337561,0.00044339066,0.0006439424,0.0060530766,0.00029795562,0.002230109,0.0019958967,0.0011499702,0.0054765805],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031577433,0.0005933739,0.039139334,0.004976804,0.0007122252,0.0017507653,0.0011456893,0.004049271,0.07487039,0.0070380513,0.6562169,0.20634939],"study_design_scores_gemma":[0.00045825358,0.00022185854,0.09850703,0.0005457248,0.00035496432,0.0010715136,0.0003334527,0.00642317,0.024400199,0.007225051,0.86019516,0.00026349392],"about_ca_topic_score_codex":0.006846465,"about_ca_topic_score_gemma":0.006954269,"teacher_disagreement_score":0.012666541,"about_ca_system_score_codex":0.0008021329,"about_ca_system_score_gemma":0.002295946,"threshold_uncertainty_score":0.042373836},"labels":[],"label_agreement":null},{"id":"W4310699919","doi":"10.1038/s41597-022-01837-x","title":"Mechanical Unloading of Engineered Human Meniscus Models Under Simulated Microgravity: A Transcriptomic Study","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Spaceflight effects on biology","field":"Medicine","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Alberta Innovates; University of British Columbia; Canada Foundation for Innovation; Government of Canada; Canadian Institutes of Health Research; Alberta Cancer Foundation; Alberta Innovates - Health Solutions; Women and Children's Health Research Institute","keywords":"Meniscus; Transcriptome; Computer science; Computational biology; Biology; Physics; Optics; Gene expression","score_opus":0.10437387391574139,"score_gpt":0.3480758117944599,"score_spread":0.24370193787871852,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4310699919","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9941929,0.00076001906,0.0035904835,0.0000316935,0.000024815145,0.00004753223,0.00087263767,0.000021552949,0.0004583884],"genre_scores_gemma":[0.9896361,0.0010142737,0.005035657,0.00009860297,0.000014002637,0.00017445379,0.002260682,0.000019671079,0.001746538],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998522,0.00001718036,0.000013776624,0.000041261756,0.000051295727,0.000024324827],"domain_scores_gemma":[0.9998845,0.00002371319,0.00003230035,0.00001712889,0.00002654579,0.00001582789],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002682797,0.00019690265,0.0001982186,0.00026528467,0.00021409424,0.00026600878,0.00011502975,0.00021435568,0.00063018844],"category_scores_gemma":[0.00015583415,0.00008787394,0.00033724896,0.00021679897,0.00019256285,0.00013646585,0.00018283392,0.00025960835,0.00017269733],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010862771,0.000029293662,0.0005867813,0.000045955534,0.0000083006935,0.000055689095,0.00005306866,0.000091759764,0.9979361,0.00003383753,0.000023405402,0.0010271806],"study_design_scores_gemma":[0.000036951515,0.0018289414,0.07515577,0.000034434586,0.00015062155,0.0008256864,0.0006005793,0.0035508012,0.9128476,0.00011995606,0.00481685,0.000031787135],"about_ca_topic_score_codex":0.00071807834,"about_ca_topic_score_gemma":0.001143833,"teacher_disagreement_score":0.00071807834,"about_ca_system_score_codex":0.0001213382,"about_ca_system_score_gemma":0.00015748822,"threshold_uncertainty_score":0.0021082163},"labels":[],"label_agreement":null},{"id":"W4311505052","doi":"10.1038/s41597-022-01888-0","title":"A data set of global river networks and corresponding water resources zones divisions v2","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"U.S. Geological Survey; Nature Conservancy; McGill University; Australian National University; World Wildlife Fund","keywords":"Data set; Set (abstract data type); Water resources; Geology; Water resource management; Environmental science; Computer science; Ecology; Biology","score_opus":0.0410215549892982,"score_gpt":0.27028991474207664,"score_spread":0.22926835975277843,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311505052","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034178887,0.00009861358,0.0032533826,0.000112415386,0.000039706134,0.00016648938,0.9580142,0.00091692206,0.003219536],"genre_scores_gemma":[0.03377692,0.000058475584,0.0052010603,0.00004176898,0.000009779411,0.00041391377,0.95977235,0.00006808238,0.00065766845],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99914896,0.00013060655,0.00014702945,0.0002382126,0.00024384444,0.00009140452],"domain_scores_gemma":[0.9979595,0.00043254002,0.0002648661,0.00050983345,0.0006467226,0.00018662978],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007372473,0.0009935659,0.00056244794,0.003849539,0.00068219175,0.00086569146,0.0015167805,0.0010893906,0.005922717],"category_scores_gemma":[0.0032083278,0.0003802942,0.0008484498,0.0070693693,0.0004277717,0.00097536616,0.0012064811,0.0009851528,0.0035749853],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007070788,0.0011177352,0.15566745,0.0029325958,0.00050293,0.001395828,0.0009748451,0.060453318,0.010470028,0.018544117,0.6340542,0.11317993],"study_design_scores_gemma":[0.00040701564,0.00021160649,0.22184062,0.0002790027,0.00015881835,0.00070737116,0.001140843,0.035286658,0.009357424,0.0069193947,0.72348106,0.00021019716],"about_ca_topic_score_codex":0.025501354,"about_ca_topic_score_gemma":0.026453217,"teacher_disagreement_score":0.025501354,"about_ca_system_score_codex":0.0010213718,"about_ca_system_score_gemma":0.0018366947,"threshold_uncertainty_score":0.05070579},"labels":[],"label_agreement":null},{"id":"W4311572563","doi":"10.1038/s41597-022-01774-9","title":"The global spectrum of plant form and function: enhanced species-level trait dataset","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ecology and Vegetation Dynamics Studies","field":"Environmental Science","cited_by":118,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Algoma University; University of Waterloo; University of Saskatchewan; Université de Sherbrooke","funders":"Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Russian Science Foundation; Natural Environment Research Council; Consejo Nacional de Investigaciones Científicas y Técnicas; Newton Fund; Fondo para la Investigación Científica y Tecnológica; Inter-American Institute for Global Change Research; Universidad Nacional de Córdoba","keywords":"Trait; Biology; Categorical variable; Vascular plant; Specific leaf area; Plant species; Taxonomic rank; Botany; Ecology; Statistics; Mathematics; Species richness; Computer science","score_opus":0.036174268526012715,"score_gpt":0.2472785866775703,"score_spread":0.21110431815155759,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311572563","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015031535,0.00038045272,0.0013090841,0.00007527126,0.000024648305,0.000024740364,0.98162043,0.0005675718,0.00096627395],"genre_scores_gemma":[0.009453119,0.000088400935,0.002282451,0.00004165241,0.000009710328,0.00008527461,0.98773324,0.000057238594,0.00024893635],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99903464,0.00013696656,0.00011706291,0.00037723253,0.00022158523,0.00011241534],"domain_scores_gemma":[0.99797624,0.0006015631,0.0003201128,0.00053202675,0.00036575465,0.00020429623],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00105616,0.0010318108,0.00097215024,0.0037653297,0.00044904507,0.0012079438,0.0015161574,0.001320377,0.0060426253],"category_scores_gemma":[0.0030920692,0.0003083475,0.0011145571,0.0044293283,0.00041145016,0.00082927744,0.0019700488,0.0010017611,0.007162382],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001223096,0.00054807257,0.15522544,0.007378362,0.0011545041,0.0009606872,0.0007396042,0.013112579,0.024000265,0.00539512,0.71181846,0.07844388],"study_design_scores_gemma":[0.00033863867,0.00019001143,0.30574495,0.00045411027,0.00029175443,0.0009736228,0.00038399253,0.005945882,0.0038129417,0.0033312046,0.678344,0.00018887126],"about_ca_topic_score_codex":0.0044257506,"about_ca_topic_score_gemma":0.00836944,"teacher_disagreement_score":0.0060426253,"about_ca_system_score_codex":0.0005209255,"about_ca_system_score_gemma":0.0007210033,"threshold_uncertainty_score":0.020214617},"labels":[],"label_agreement":null},{"id":"W4311582290","doi":"10.1038/s41597-022-01860-y","title":"Particle filtration efficiency measured using sodium chloride and polystyrene latex sphere test methods","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Infection Control and Ventilation","field":"Medicine","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada","funders":"Public Health Agency; Public Health Agency of Canada","keywords":"Polystyrene; Dispersity; Filtration (mathematics); Materials science; Sodium; Equivalence (formal languages); Chloride; Particle size; Particle (ecology); Mathematics; Chemistry; Polymer chemistry; Statistics; Composite material; Metallurgy; Physical chemistry","score_opus":0.09924052525318396,"score_gpt":0.36494054651851177,"score_spread":0.2657000212653278,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311582290","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95436895,0.0014154642,0.031638656,0.00012600527,0.000072863135,0.0003140319,0.0016741221,0.00033713557,0.010052835],"genre_scores_gemma":[0.9680895,0.00088342844,0.024850652,0.00013139346,0.000020179148,0.00027834254,0.0018116309,0.00004303456,0.0038918366],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99424464,0.0006466298,0.00043292978,0.00046767876,0.0040126313,0.00019551498],"domain_scores_gemma":[0.99689686,0.0010382084,0.00045564474,0.00024818187,0.0012467229,0.0001143604],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002197182,0.00048794155,0.00042902684,0.0016477612,0.00042588846,0.00073461415,0.00076777075,0.0006673319,0.0014978268],"category_scores_gemma":[0.003938184,0.00013999123,0.00040060154,0.001114949,0.00045963397,0.0006364996,0.00063373696,0.00038728814,0.00073625444],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012062053,0.00057023234,0.081095085,0.0004990664,0.00020466748,0.00016435694,0.0009260904,0.0028953622,0.8520704,0.0013470951,0.0017857179,0.057235796],"study_design_scores_gemma":[0.000024160048,0.0015633169,0.066682436,0.000030151745,0.000061512364,0.00033328994,0.00017808737,0.00567332,0.9223717,0.00021730798,0.0028041922,0.000060527702],"about_ca_topic_score_codex":0.0029259722,"about_ca_topic_score_gemma":0.0025929161,"teacher_disagreement_score":0.0029259722,"about_ca_system_score_codex":0.00059411855,"about_ca_system_score_gemma":0.00053402584,"threshold_uncertainty_score":0.0116199255},"labels":[],"label_agreement":null},{"id":"W4311715174","doi":"10.1038/s41597-022-01887-1","title":"Comparative transcriptome profiles of four sexually size dimorphic fish","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fish Ecology and Management Studies","field":"Environmental Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"National Key Research and Development Program of China; National Natural Science Foundation of China","keywords":"Sexual dimorphism; Transcriptome; Biology; Fish <Actinopterygii>; Zoology; Evolutionary biology; Fishery; Genetics; Gene expression; Gene","score_opus":0.06658906232373124,"score_gpt":0.27213229318616233,"score_spread":0.2055432308624311,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311715174","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99643576,0.00020011871,0.0010447297,0.000026600683,0.000006796788,0.000010867723,0.0016580033,0.00001769829,0.0005995589],"genre_scores_gemma":[0.98560065,0.00027382647,0.0037446402,0.00014591312,0.0000056378235,0.00009091067,0.006712554,0.00003228208,0.0033935024],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9999143,0.0000051369634,0.0000065830454,0.000040296367,0.000019070703,0.000014687493],"domain_scores_gemma":[0.99991345,0.000017371312,0.000019574392,0.000006211845,0.000027990744,0.000015346543],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000097923315,0.00016837285,0.00021063747,0.0004167306,0.00028305285,0.0001851071,0.00009042195,0.00022242527,0.0009704918],"category_scores_gemma":[0.00018425622,0.00012998632,0.00032104852,0.00034111305,0.00020763469,0.00015535389,0.00032893373,0.00020278055,0.0002497713],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002185651,0.000011248229,0.02901382,0.00009460859,0.000031141466,0.00011426284,0.0002591209,0.00015744635,0.9657291,0.000044476667,0.00009382303,0.004232345],"study_design_scores_gemma":[0.00001566514,0.00032319143,0.92395586,0.00001964096,0.00011955314,0.00051328255,0.0009971187,0.0014152748,0.0690036,0.00018746046,0.0034201576,0.000029229546],"about_ca_topic_score_codex":0.0015683785,"about_ca_topic_score_gemma":0.0038383242,"teacher_disagreement_score":0.0015683785,"about_ca_system_score_codex":0.00016263821,"about_ca_system_score_gemma":0.00020571378,"threshold_uncertainty_score":0.0032466054},"labels":[],"label_agreement":null},{"id":"W4313439128","doi":"10.1038/s41597-022-01899-x","title":"MIMIC-IV, a freely accessible electronic health record dataset","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":2791,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children","funders":"National Institutes of Health; National Institute of Biomedical Imaging and Bioengineering; U.S. Department of Health and Human Services","keywords":"Electronic health record; Computer science; Health records; Information retrieval; World Wide Web; Health care; Political science","score_opus":0.10846043845443115,"score_gpt":0.39437885807324796,"score_spread":0.2859184196188168,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313439128","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005239309,0.0003324289,0.0017348717,0.0003753951,0.00007964215,0.00028316575,0.98858184,0.00083643507,0.002536972],"genre_scores_gemma":[0.011639457,0.00026931727,0.0047514304,0.00025464766,0.00007288622,0.0006529427,0.98116106,0.00008435563,0.0011139485],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99770856,0.00046371692,0.0006604253,0.00048548824,0.00052983366,0.00015190677],"domain_scores_gemma":[0.99213374,0.0022681626,0.0016002408,0.0018822611,0.0014440491,0.0006715113],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.001916602,0.0006202996,0.00085032627,0.0041789208,0.00042844095,0.0016798234,0.0017249312,0.001367663,0.013040653],"category_scores_gemma":[0.020860117,0.0003703873,0.00072013837,0.0048268507,0.00032120742,0.0008913463,0.0021301412,0.000981804,0.010797725],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011369723,0.00021140654,0.047017958,0.003066475,0.00038126297,0.0005551895,0.0003880393,0.0018163448,0.0025798911,0.0039721774,0.88103175,0.05784248],"study_design_scores_gemma":[0.0007018671,0.0002753893,0.10228066,0.0008369495,0.00027097657,0.0009272221,0.00047562836,0.0039665303,0.0028518604,0.0045927926,0.88268375,0.00013643931],"about_ca_topic_score_codex":0.008860976,"about_ca_topic_score_gemma":0.013200989,"teacher_disagreement_score":0.99827504,"about_ca_system_score_codex":0.0012387183,"about_ca_system_score_gemma":0.0043483586,"threshold_uncertainty_score":0.043625355},"labels":[],"label_agreement":null},{"id":"W4313528201","doi":"10.1038/s41597-022-01880-8","title":"Familydemic Cross Country and Gender Dataset on work and family outcomes during COVID-19 pandemic","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Work-Family Balance Challenges","field":"Social Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Brock University; University of British Columbia","funders":"Forskningsrådet om Hälsa, Arbetsliv och Välfärd; Social Sciences and Humanities Research Council of Canada; Universität Hamburg; Università degli Studi di Firenze; Narodowe Centrum Nauki; Vetenskapsrådet; Government of Canada; Royal Swedish Academy of Sciences; Davidson College; Fordham University","keywords":"Pandemic; Work (physics); Welfare; Coronavirus disease 2019 (COVID-19); Demographic economics; Division of labour; Cross country; Geography; Political science; Economic growth; Economics; Medicine","score_opus":0.20641003253429405,"score_gpt":0.418678501366601,"score_spread":0.21226846883230696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313528201","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004370505,0.00008576629,0.00010411069,0.00009426647,0.000022355003,0.000020998217,0.994383,0.000054921526,0.0008641449],"genre_scores_gemma":[0.0068787406,0.00007530545,0.00043831233,0.00006007923,0.000013167219,0.00014171239,0.99181443,0.000013525004,0.0005647817],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993794,0.00012768853,0.000076293574,0.00015243393,0.00011891577,0.00014533853],"domain_scores_gemma":[0.9983993,0.00035019143,0.00034661512,0.0002473155,0.00042785564,0.00022872814],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008695492,0.000778564,0.00068971404,0.0022583469,0.0005489641,0.0010324813,0.0011570442,0.0012215135,0.00896971],"category_scores_gemma":[0.0039269435,0.00027364783,0.000717699,0.003742029,0.00020240697,0.00049114943,0.0014258383,0.0008143321,0.0050533004],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003481445,0.00012964102,0.08835742,0.00081845524,0.00019014225,0.0002836147,0.00033028444,0.0015806408,0.0003410242,0.0018673997,0.892797,0.012956329],"study_design_scores_gemma":[0.00027890797,0.00009222916,0.33690974,0.00082117075,0.000111227215,0.00043275382,0.0013717278,0.0016172571,0.00060110283,0.0015293459,0.6561425,0.0000920758],"about_ca_topic_score_codex":0.0507417,"about_ca_topic_score_gemma":0.06655655,"teacher_disagreement_score":0.0507417,"about_ca_system_score_codex":0.00096178456,"about_ca_system_score_gemma":0.0014474929,"threshold_uncertainty_score":0.10089272},"labels":[],"label_agreement":null},{"id":"W4313544955","doi":"10.1038/s41597-022-01896-0","title":"Author Correction: COVIDiSTRESS diverse dataset on psychological and behavioural outcomes one year into the COVID-19 pandemic","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"COVID-19 and Mental Health","field":"Psychology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Pandemic; Coronavirus disease 2019 (COVID-19); 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Psychology; Medicine; Virology; Outbreak; Infectious disease (medical specialty)","score_opus":0.4524044860773293,"score_gpt":0.5205371422595774,"score_spread":0.06813265618224806,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313544955","genre_codex":"dataset","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015868375,0.0012491189,0.002912447,0.0754561,0.26100728,0.00014217883,0.6390067,0.0024106093,0.016228784],"genre_scores_gemma":[0.041970935,0.004117858,0.010942287,0.05200851,0.032197956,0.0011894573,0.6576913,0.008087602,0.19179411],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9895655,0.0019545478,0.0023205052,0.0015158314,0.003988591,0.00065498997],"domain_scores_gemma":[0.9344625,0.020855248,0.005718552,0.008488503,0.028625298,0.0018499619],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010257239,0.0015842521,0.0015705588,0.005225323,0.0021483256,0.004530528,0.00218732,0.0027179504,0.11623523],"category_scores_gemma":[0.16177163,0.0008945969,0.0013941666,0.008719133,0.0012965031,0.0022530283,0.0039849603,0.006574892,0.076441675],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025038238,0.0000019771103,0.0005965001,0.000070459886,0.000011974736,0.000036704878,0.000027367163,0.000039317605,0.0000070496612,0.00042717825,0.9963568,0.0023996353],"study_design_scores_gemma":[0.000111395384,0.000011755083,0.004374884,0.00087509275,0.000056143235,0.0002566742,0.0002055337,0.00025609226,0.00020555564,0.002296862,0.9912798,0.0000703473],"about_ca_topic_score_codex":0.051449142,"about_ca_topic_score_gemma":0.06095,"teacher_disagreement_score":0.11623523,"about_ca_system_score_codex":0.003359103,"about_ca_system_score_gemma":0.009549675,"threshold_uncertainty_score":0.38884556},"labels":[],"label_agreement":null},{"id":"W4315567907","doi":"10.1038/s41597-022-01915-0","title":"A 30-year dataset of CO2 in flowing freshwaters in the United States","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ocean Acidification Effects and Responses","field":"Earth and Planetary Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Winnipeg","funders":"Louisiana Board of Regents","keywords":"Alkalinity; Environmental science; Water quality; Carbon dioxide; Range (aeronautics); Water chemistry; Limnology; Ecosystem; Dissolved organic carbon; Oceanography; Freshwater ecosystem; Ecology; Environmental chemistry; Hydrology (agriculture); Chemistry; Biology; Geology; Environmental engineering","score_opus":0.06029442004987093,"score_gpt":0.2847824655910135,"score_spread":0.22448804554114257,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4315567907","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.106864974,0.0002894452,0.0004018882,0.00033441867,0.000043874068,0.000026867041,0.8903607,0.00036068322,0.0013172631],"genre_scores_gemma":[0.11263492,0.00012629952,0.001474391,0.00012183427,0.000021604044,0.000088343564,0.88503724,0.000025647412,0.00046975125],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99953115,0.000064256616,0.00007050145,0.0001639902,0.000106977524,0.00006308678],"domain_scores_gemma":[0.99846363,0.00022003357,0.00032065573,0.00021777458,0.00059780676,0.00018017733],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00059898954,0.00040490972,0.0004326094,0.0020316124,0.0004254928,0.0006989177,0.00069895375,0.00054946553,0.0018208242],"category_scores_gemma":[0.0020057831,0.00021685915,0.00041059306,0.005764891,0.00026698902,0.0005883373,0.000978008,0.0004469802,0.0009433707],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002709112,0.00021251687,0.5814307,0.0007241208,0.00038942462,0.0002811986,0.00029804753,0.011009614,0.0017159148,0.0012191691,0.3775573,0.024891159],"study_design_scores_gemma":[0.0001080909,0.00004731987,0.83724767,0.0001483954,0.0000667088,0.00011855656,0.00064765534,0.014217179,0.0013371434,0.0007759111,0.1452162,0.00006910807],"about_ca_topic_score_codex":0.17130743,"about_ca_topic_score_gemma":0.18237114,"teacher_disagreement_score":0.17130743,"about_ca_system_score_codex":0.0010045622,"about_ca_system_score_gemma":0.0012962882,"threshold_uncertainty_score":0.34062064},"labels":[],"label_agreement":null},{"id":"W4315782561","doi":"10.1038/s41597-023-01935-4","title":"Interbasin water transfers in the United States and Canada","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Water resources management and optimization","field":"Engineering","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"U.S. Geological Survey; U.S. Department of the Interior; National Science Foundation","keywords":"Geospatial analysis; Water transfer; Water use; Water resource management; Environmental science; Environmental resource management; Hydrology (agriculture); Geography; Remote sensing; Ecology; Engineering","score_opus":0.02280844734738341,"score_gpt":0.19856969990625967,"score_spread":0.17576125255887626,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4315782561","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.66944,0.0011122453,0.0018791986,0.0008718706,0.000028603024,0.00020701298,0.29288247,0.00043368846,0.033145044],"genre_scores_gemma":[0.79687154,0.0014238653,0.004977856,0.00026521928,0.000008922914,0.00015280077,0.18712744,0.000072954645,0.009099381],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99948347,0.000023451195,0.00002724304,0.00009218031,0.00021871421,0.00015477525],"domain_scores_gemma":[0.9983329,0.00013468492,0.000170316,0.00009004007,0.001056725,0.00021535603],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00026974623,0.00035406149,0.00029761152,0.0037657698,0.0015182152,0.0015411077,0.00075855077,0.0002817387,0.002327055],"category_scores_gemma":[0.0018049867,0.0001732287,0.00037464272,0.012764732,0.00059079065,0.0005485621,0.0010692669,0.0004055117,0.00033366092],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026974335,0.00017389185,0.72765976,0.00039115112,0.0003590839,0.0006543212,0.0026065242,0.041550353,0.000980305,0.013739229,0.12460285,0.08701275],"study_design_scores_gemma":[0.00003078712,0.000015360089,0.84721017,0.00022718875,0.00007419897,0.00011520813,0.005704228,0.029002173,0.0011520841,0.0017405759,0.11464263,0.00008545097],"about_ca_topic_score_codex":0.9966808,"about_ca_topic_score_gemma":0.9980939,"teacher_disagreement_score":0.023654733,"about_ca_system_score_codex":0.023654733,"about_ca_system_score_gemma":0.030237608,"threshold_uncertainty_score":0.17162788},"labels":[],"label_agreement":null},{"id":"W4316097136","doi":"10.1038/s41597-022-01889-z","title":"Satellite-derived multivariate world-wide lake physical variable timeseries for climate studies","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and coastal ecosystems","field":"Earth and Planetary Sciences","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Centre for Ecology and Hydrology; Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences; Università degli Studi di Perugia; National Centre for Earth Observation; Eesti Maaülikool; Fisheries and Oceans Canada; European Space Agency; Uppsala Universitet; Universidad del Valle; Technion-Israel Institute of Technology; European Commission; National Institute for Environmental Studies; Latvijas Universitate; University of Wisconsin-Madison; Università degli Studi di Trento; Suomen Ympäristökeskus; Tartu Ülikool; Agri-Food and Biosciences Institute; Centre National d’Etudes Spatiales; Environmental Protection Agency; Environment and Climate Change Canada; University of Stirling; Sight Research UK; Irkutsk State University; Compute Canada; University of Queensland; Natural Environment Research Council; U.S. Environmental Protection Agency; Nipissing University; KU Leuven; Creighton University","keywords":"Environmental science; Satellite imagery; Satellite; Latitude; Limnology; Climate change; Climatology; Longitude; Remote sensing; Physical geography; Geography; Oceanography; Geology","score_opus":0.08653126989409696,"score_gpt":0.30584255371121843,"score_spread":0.21931128381712148,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4316097136","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02216784,0.0001437614,0.00511533,0.00012468963,0.000041528365,0.000106625135,0.96892494,0.0011524495,0.0022228213],"genre_scores_gemma":[0.02101233,0.00006947379,0.008331769,0.00002405956,0.0000151750855,0.00027743125,0.96915215,0.000113787864,0.0010037125],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9995751,0.00007252524,0.000052450425,0.00011730476,0.00012665501,0.000055969882],"domain_scores_gemma":[0.9987375,0.00010370854,0.00026906744,0.00037321635,0.00035941173,0.00015711384],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007188812,0.0006218798,0.0004733566,0.0018608759,0.00039518884,0.0008193572,0.00073538045,0.00036840714,0.008885428],"category_scores_gemma":[0.0024408416,0.0002841376,0.00062795065,0.0054818084,0.00015854472,0.00079612015,0.0015229726,0.00075703935,0.00600731],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005928854,0.00036002044,0.19804369,0.0010321898,0.000634439,0.00021744659,0.00050226995,0.017832806,0.008042023,0.006117725,0.65495867,0.11166591],"study_design_scores_gemma":[0.00022058917,0.000106949774,0.5170541,0.00017574185,0.000116310526,0.00015307288,0.00032498932,0.023384888,0.0037289553,0.0032499768,0.45138603,0.00009840636],"about_ca_topic_score_codex":0.028579097,"about_ca_topic_score_gemma":0.04732388,"teacher_disagreement_score":0.028579097,"about_ca_system_score_codex":0.0006387718,"about_ca_system_score_gemma":0.0018192191,"threshold_uncertainty_score":0.05682546},"labels":[],"label_agreement":null},{"id":"W4316499895","doi":"10.1038/s41597-023-01945-2","title":"Author Correction: MIMIC-IV, a freely accessible electronic health record dataset","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":166,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children","funders":"","keywords":"Computer science; Electronic health record; Information retrieval; Health records; Data science; World Wide Web; Political science; Health care","score_opus":0.10061696061468166,"score_gpt":0.3922536860077259,"score_spread":0.29163672539304425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4316499895","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016314742,0.0013525402,0.0093741845,0.090367794,0.5758983,0.0002882282,0.29713812,0.005004091,0.01894522],"genre_scores_gemma":[0.039795775,0.004963281,0.038349245,0.07269885,0.066061236,0.0019966993,0.39418736,0.020306608,0.361641],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98449206,0.0032722766,0.0036924258,0.0020195388,0.005886678,0.00063701964],"domain_scores_gemma":[0.87406635,0.03987772,0.007014852,0.01608085,0.060739666,0.002220589],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013397809,0.0015839603,0.0016598627,0.0068528946,0.0029654978,0.0051682335,0.0030196141,0.0025968219,0.1918794],"category_scores_gemma":[0.21197626,0.001121763,0.0015676446,0.009312159,0.0017794042,0.0025641175,0.0043325117,0.0058213365,0.10531163],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028574636,0.000003277628,0.00017576518,0.00012432135,0.0000126802315,0.00006649373,0.000029940365,0.00004437416,0.00002410923,0.00059812755,0.9954554,0.0034369598],"study_design_scores_gemma":[0.00009029893,0.000012914454,0.0013178434,0.00062599615,0.00005800555,0.0003241621,0.00013105292,0.0002689062,0.000373043,0.002618008,0.99412525,0.000054524702],"about_ca_topic_score_codex":0.025382467,"about_ca_topic_score_gemma":0.029822098,"teacher_disagreement_score":0.1918794,"about_ca_system_score_codex":0.0034221024,"about_ca_system_score_gemma":0.010995832,"threshold_uncertainty_score":0.64190054},"labels":[],"label_agreement":null},{"id":"W4317500295","doi":"10.1038/s41597-022-01919-w","title":"The global historical climate database HCLIM","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Tree-ring climate responses","field":"Earth and Planetary Sciences","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"European Commission; Sight Research UK; University of Bern; Natural Environment Research Council; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Precipitation; Environmental science; Climatology; Database; Climate change; Data collection; Series (stratigraphy); Meteorology; Geography; Computer science; Statistics; Mathematics; Geology","score_opus":0.08066699518869384,"score_gpt":0.29728364273494867,"score_spread":0.21661664754625481,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317500295","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00085454015,0.00022069535,0.0010080924,0.00006270837,0.00004348481,0.000024816958,0.99189407,0.000996469,0.004895089],"genre_scores_gemma":[0.0028129972,0.00022826326,0.0016079483,0.000041320982,0.000022473349,0.000093679766,0.99390453,0.0001910067,0.0010978268],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993418,0.00009448014,0.0001319327,0.00018903536,0.00017829059,0.00006445094],"domain_scores_gemma":[0.9977864,0.00034484954,0.00037515405,0.0007055813,0.0006101547,0.00017774303],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008988266,0.0008003062,0.0007695397,0.0066030957,0.00038174106,0.0015106493,0.0013771338,0.00053990376,0.052185677],"category_scores_gemma":[0.0036967655,0.00040903545,0.0004242294,0.015096113,0.00017025348,0.0015623434,0.0013627341,0.00080646854,0.039333098],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014503954,0.000027459342,0.007070834,0.0013185652,0.00009768485,0.00012652142,0.00016080217,0.0031076502,0.00077451306,0.006152068,0.92884475,0.052174147],"study_design_scores_gemma":[0.000031238953,0.000008785608,0.009341362,0.000125229,0.000022103894,0.000081304795,0.000058410045,0.0008163679,0.00033077915,0.0013843292,0.9877781,0.00002193914],"about_ca_topic_score_codex":0.01094848,"about_ca_topic_score_gemma":0.010298177,"teacher_disagreement_score":0.052185677,"about_ca_system_score_codex":0.0008121728,"about_ca_system_score_gemma":0.0018485361,"threshold_uncertainty_score":0.17457849},"labels":[],"label_agreement":null},{"id":"W4318617212","doi":"10.1038/s41597-023-01949-y","title":"Quality control and removal of technical variation of NMR metabolic biomarker data in ~120,000 UK Biobank participants","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":177,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"NIHR Cambridge Biomedical Research Centre; Economic and Social Research Council; Engineering and Physical Sciences Research Council; Chief Scientist Office, Scottish Government Health and Social Care Directorate; Medical Research Council; Public Health Agency; University of Cambridge; Department of Health and Social Care; Canadian Institutes of Health Research; Health and Social Care Research and Development Division; National Institute for Health and Care Research; Science and Technology Facilities Council; British Heart Foundation; Scottish Government; Dell EMC; Wellcome Trust","keywords":"Biobank; Biomarker; Biomarker discovery; Computer science; Computational biology; Medicine; Bioinformatics; Chemistry; Biology; Proteomics; Biochemistry","score_opus":0.10249251048695286,"score_gpt":0.36813642298749344,"score_spread":0.2656439125005406,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4318617212","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47734147,0.0036012586,0.41380233,0.0029969863,0.001109562,0.0047941217,0.079971865,0.008378145,0.008004301],"genre_scores_gemma":[0.5503451,0.0011434772,0.33883283,0.0019663097,0.00039315128,0.009949598,0.08353007,0.004491985,0.00934742],"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9492206,0.024726525,0.0082099,0.0081729805,0.008188397,0.00148159],"domain_scores_gemma":[0.889829,0.03888051,0.012728486,0.033201948,0.024186742,0.0011733588],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.06454731,0.0010489657,0.0017397171,0.0031757425,0.001743557,0.0029455577,0.002074475,0.001522144,0.008179365],"category_scores_gemma":[0.17083351,0.0010388211,0.001492133,0.0060367375,0.0015912433,0.0009617543,0.0035375499,0.001082671,0.0038575267],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006559118,0.0004919937,0.43659392,0.0049062553,0.0024810764,0.0014928846,0.011188858,0.008169628,0.059653,0.005620116,0.104271516,0.3585717],"study_design_scores_gemma":[0.00064750103,0.00081060483,0.6343642,0.0009935396,0.0013326325,0.0022140532,0.0011525975,0.019306565,0.05198591,0.0041756043,0.28265586,0.00036083514],"about_ca_topic_score_codex":0.014770342,"about_ca_topic_score_gemma":0.01583468,"teacher_disagreement_score":0.06454731,"about_ca_system_score_codex":0.0011872516,"about_ca_system_score_gemma":0.0032928823,"threshold_uncertainty_score":0.3413629},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"bench_or_experimental","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4318766583","doi":"10.1038/s41597-023-01977-8","title":"Reaction profiles for quantum chemistry-computed [3 + 2] cycloaddition reactions","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Click Chemistry and Applications","field":"Chemistry","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Vetenskapsrådet","keywords":"Cycloaddition; Workflow; Click chemistry; Reactivity (psychology); Computer science; Chemistry; Computational chemistry; Quantum chemistry; Benchmarking; Reaction mechanism; Combinatorial chemistry; Computational science; Biochemical engineering; Organic chemistry; Database; Catalysis","score_opus":0.08451263699742557,"score_gpt":0.32619986212294105,"score_spread":0.2416872251255155,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4318766583","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8060605,0.0008465122,0.1541044,0.0004707584,0.00018742107,0.0002636905,0.00818391,0.0041911067,0.025691733],"genre_scores_gemma":[0.94516087,0.00038842863,0.047836125,0.00008686675,0.000020426112,0.0003157613,0.0041190167,0.0006515653,0.0014209511],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996013,0.00006856236,0.00002951644,0.0000642177,0.00015501566,0.000081308855],"domain_scores_gemma":[0.9987282,0.000735508,0.000078393896,0.00012355672,0.00027494546,0.00005944587],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009802436,0.00091280503,0.00094660115,0.0013778467,0.0009083101,0.0009804948,0.001221588,0.0012832817,0.007088376],"category_scores_gemma":[0.0027623444,0.00040003387,0.0009177248,0.0019308592,0.00051138387,0.0011124744,0.00044608655,0.0013043408,0.0008665956],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002633501,0.00017955647,0.0017698949,0.00031794517,0.00004807823,0.00018428425,0.00012499679,0.96257544,0.008296185,0.013188665,0.0017652828,0.011286232],"study_design_scores_gemma":[0.000027048456,0.00003977876,0.00033052324,0.000008224074,0.000008686925,0.000017087366,0.000025950278,0.9917846,0.00522544,0.0019222758,0.0005956745,0.000014656609],"about_ca_topic_score_codex":0.007977837,"about_ca_topic_score_gemma":0.0057058614,"teacher_disagreement_score":0.007977837,"about_ca_system_score_codex":0.0016681055,"about_ca_system_score_gemma":0.0016407047,"threshold_uncertainty_score":0.023712933},"labels":[],"label_agreement":null},{"id":"W4320038879","doi":"10.1038/s41597-022-01811-7","title":"The Psychological Science Accelerator’s COVID-19 rapid-response dataset","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 and Mental Health","field":"Psychology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Wilfrid Laurier University","funders":"National Cancer Institute; National Institute of Mental Health; Fundação para a Ciência e a Tecnologia; Japan Society for the Promotion of Science; Morsani College of Medicine; National Institutes of Health; Universidad de los Andes; Agentúra na Podporu Výskumu a Vývoja; Agence Nationale de la Recherche; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; National Research University Higher School of Economics; UK Research and Innovation; Agencia Nacional de Investigación y Desarrollo; Uniwersytet Wrocławski; Univerzita Karlova v Praze; Kingston University; Economic and Social Research Council; European Commission; Narodowym Centrum Nauki; Amazon Web Services","keywords":"Framing (construction); Coronavirus disease 2019 (COVID-19); Pandemic; Autonomy; Psychology; Framing effect; Raw data; Cognition; Psychological science; Social psychology; Medicine; Geography; Statistics; Mathematics; Psychiatry","score_opus":0.3940632159279168,"score_gpt":0.5637136789219475,"score_spread":0.16965046299403064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4320038879","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012288636,0.000057990048,0.0002753876,0.00031042664,0.000053966756,0.00013768466,0.9966653,0.00010988627,0.0011604293],"genre_scores_gemma":[0.0055153673,0.000092904666,0.0016101351,0.00040109904,0.00005703937,0.002544071,0.98799896,0.00007989613,0.0017005571],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9961055,0.0014947167,0.0006216012,0.0006661469,0.0007073141,0.00040465855],"domain_scores_gemma":[0.9804473,0.009002341,0.0023161191,0.0032381704,0.0038886385,0.001107432],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005180792,0.00071849214,0.0009577295,0.0031349116,0.0011833373,0.0019244993,0.0023281057,0.0020795995,0.045413442],"category_scores_gemma":[0.039364822,0.0005719793,0.00085509574,0.0058810934,0.000374779,0.00084898993,0.0029470553,0.00198869,0.024787156],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023551726,0.00007006303,0.010527988,0.0007111823,0.00006528413,0.000057866233,0.00017129902,0.00038471803,0.00008641157,0.0012376081,0.98023486,0.00621717],"study_design_scores_gemma":[0.0006944212,0.00009944568,0.05341509,0.0009891987,0.00008762377,0.00013030651,0.0006869894,0.00096325774,0.0003067659,0.0029540902,0.93957305,0.000099682395],"about_ca_topic_score_codex":0.029812643,"about_ca_topic_score_gemma":0.045276113,"teacher_disagreement_score":0.045413442,"about_ca_system_score_codex":0.0019388103,"about_ca_system_score_gemma":0.0038920396,"threshold_uncertainty_score":0.15192306},"labels":[],"label_agreement":null},{"id":"W4320857731","doi":"10.1038/s41597-023-01942-5","title":"A longitudinal microstructural MRI dataset in healthy C57Bl/6 mice at 9.4 Tesla","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Robarts Clinical Trials; Western University","funders":"Congressionally Directed Medical Research Programs; Canada First Research Excellence Fund; Natural Sciences and Engineering Research Council of Canada; Government of Canada; U.S. Department of Defense","keywords":"Diffusion MRI; Fractional anisotropy; Magnetic resonance imaging; Neuroimaging; Human Connectome Project; Computer science; Magnetization transfer; Nuclear magnetic resonance; Artificial intelligence; Pattern recognition (psychology); Neuroscience; Radiology; Physics; Medicine; Biology; Functional connectivity","score_opus":0.20877625383579276,"score_gpt":0.4405094157300159,"score_spread":0.23173316189422313,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4320857731","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.40675414,0.003213667,0.057699308,0.0007772492,0.00027003,0.00052864314,0.5202417,0.0050239894,0.0054913154],"genre_scores_gemma":[0.26038736,0.0019860908,0.07058014,0.000550897,0.000097710436,0.0010001503,0.6596695,0.0011164513,0.004611693],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9997056,0.000038678685,0.000035294404,0.00010361284,0.00007664217,0.000040202212],"domain_scores_gemma":[0.999071,0.00012138583,0.00016838149,0.00023768481,0.00028976917,0.00011178215],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00077171857,0.0010684427,0.00073950464,0.0014860372,0.0004933129,0.0005614945,0.00070137624,0.0010371153,0.0028931538],"category_scores_gemma":[0.0009640487,0.0003688378,0.00067980174,0.0010899195,0.00042664888,0.00036427862,0.00071182,0.00080256065,0.0018421052],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003767632,0.0014975311,0.044250023,0.0032510662,0.0016369836,0.0026413465,0.0005163588,0.020228444,0.6388961,0.0022880498,0.1665693,0.11445725],"study_design_scores_gemma":[0.00072786584,0.0029352822,0.45214123,0.0008226308,0.0013967516,0.013481945,0.00062701455,0.04166577,0.18498184,0.010937639,0.2896877,0.00059432915],"about_ca_topic_score_codex":0.0040998263,"about_ca_topic_score_gemma":0.011571258,"teacher_disagreement_score":0.0040998263,"about_ca_system_score_codex":0.00039697537,"about_ca_system_score_gemma":0.00075457495,"threshold_uncertainty_score":0.009678543},"labels":[],"label_agreement":null},{"id":"W4321076624","doi":"10.1038/s41597-023-01973-y","title":"GLORIA - A globally representative hyperspectral in situ dataset for optical sensing of water quality","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and coastal ecosystems","field":"Earth and Planetary Sciences","cited_by":143,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"Bundesministerium für Wirtschaft und Energie; Bundesministerium für Bildung und Forschung; U.S. Geological Survey; European Commission; Eesti Teadusagentuur; Horizon 2020 Framework Programme; Nuclear Safety and Security Commission; National Foundation for Science and Technology Development; National Aeronautics and Space Administration","keywords":"Hyperspectral imaging; Remote sensing; Water quality; Environmental science; In situ; Computer science; Geography; Ecology; Meteorology; Biology","score_opus":0.09523647153939807,"score_gpt":0.33412230150700406,"score_spread":0.238885829967606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321076624","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.046206314,0.000524729,0.008533645,0.0003840721,0.00013911643,0.00020995797,0.92961514,0.007208536,0.0071785804],"genre_scores_gemma":[0.035903614,0.00016851771,0.012408074,0.00015962395,0.000030233796,0.00024675883,0.94943506,0.00035908606,0.0012890978],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999283,0.00008724166,0.000059158025,0.00022534696,0.00023750048,0.000107631095],"domain_scores_gemma":[0.99929345,0.00006133612,0.00006960468,0.0002512666,0.00026317753,0.00006123646],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006864305,0.0013210098,0.0009861736,0.0018349288,0.00053633086,0.0009192077,0.0016234925,0.0012704779,0.0042496524],"category_scores_gemma":[0.0012207387,0.00036931262,0.0012561602,0.002901018,0.000433352,0.0009558557,0.0016194221,0.0011610499,0.006410852],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005946824,0.0006692727,0.04498643,0.0021288039,0.0008166083,0.00057320605,0.0006116615,0.021991238,0.05444392,0.0022361383,0.7566462,0.11430191],"study_design_scores_gemma":[0.0006972566,0.00020511498,0.26778728,0.0004700891,0.00023113441,0.0005608468,0.0010622585,0.05335634,0.023946766,0.00459686,0.6466874,0.00039874561],"about_ca_topic_score_codex":0.025802545,"about_ca_topic_score_gemma":0.053597216,"teacher_disagreement_score":0.025802545,"about_ca_system_score_codex":0.0007408197,"about_ca_system_score_gemma":0.0012839834,"threshold_uncertainty_score":0.051304758},"labels":[],"label_agreement":null},{"id":"W4322764463","doi":"10.1038/s41597-023-02015-3","title":"The Three Terms Task - an open benchmark to compare human and artificial semantic representations","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institut Universitaire de Gériatrie de Montréal","funders":"","keywords":"Computer science; Semantic memory; Natural language processing; Artificial intelligence; Semantic similarity; Task (project management); Benchmark (surveying); Word (group theory); Similarity (geometry); Associative property; Noun; Representation (politics); Cognition; Linguistics; Psychology; Mathematics","score_opus":0.14608141974920205,"score_gpt":0.4256313126982627,"score_spread":0.27954989294906063,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4322764463","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.50011706,0.005151339,0.070388794,0.0031528617,0.0018169576,0.0032098165,0.33369517,0.006973332,0.0754946],"genre_scores_gemma":[0.38206822,0.000840344,0.087405026,0.0013145242,0.00042530734,0.005145132,0.5089397,0.0013636689,0.012498082],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99551857,0.0015837745,0.0007601236,0.00090956927,0.0010482455,0.00017965032],"domain_scores_gemma":[0.98647004,0.006318623,0.0012365766,0.0033092555,0.00186306,0.00080243155],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034268654,0.0015816364,0.0009776242,0.0031136316,0.0016348641,0.002119375,0.0022247508,0.0026981079,0.008412775],"category_scores_gemma":[0.024098983,0.00028676412,0.0015892129,0.0028478885,0.001257949,0.004622595,0.0035434782,0.0020872273,0.007981929],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004965624,0.0027097582,0.072603606,0.007860576,0.0010509333,0.0014070534,0.005018494,0.011631368,0.027510576,0.025012452,0.46404728,0.37618217],"study_design_scores_gemma":[0.0014762176,0.002583086,0.16224457,0.0011318498,0.0004218051,0.0050915377,0.006430395,0.07181837,0.02435546,0.1175997,0.6062579,0.00058912835],"about_ca_topic_score_codex":0.0033046359,"about_ca_topic_score_gemma":0.0057385005,"teacher_disagreement_score":0.008412775,"about_ca_system_score_codex":0.0010142202,"about_ca_system_score_gemma":0.0012960562,"threshold_uncertainty_score":0.028143525},"labels":[],"label_agreement":null},{"id":"W4324344293","doi":"10.1038/s41597-023-02059-5","title":"Newly reconstructed Arctic surface air temperatures for 1979–2021 with deep learning method","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Arctic and Antarctic ice dynamics","field":"Earth and Planetary Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Centers for Environmental Information; National Oceanic and Atmospheric Administration; Norges Forskningsråd; Alberta Agricultural Research Institute; U.S. Department of Energy","keywords":"Arctic; The arctic; Environmental science; Surface air temperature; Oceanography; Geology; Climate change","score_opus":0.0260754778403675,"score_gpt":0.2633308319054671,"score_spread":0.23725535406509957,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4324344293","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.701559,0.0021369543,0.05654809,0.00090547826,0.0010880003,0.000068670335,0.22703521,0.0028094917,0.007849062],"genre_scores_gemma":[0.6660923,0.0006256524,0.028734492,0.00012048368,0.00016821615,0.000097867,0.29982004,0.0002205972,0.0041204374],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99978703,0.000024362753,0.000015059124,0.00009346353,0.00003583618,0.000044271696],"domain_scores_gemma":[0.999653,0.000036376063,0.000033135475,0.00005185779,0.00019670323,0.00002894549],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045878039,0.0008525346,0.00043314032,0.0015689116,0.00031776525,0.00058803795,0.00066895917,0.0006182248,0.0021664288],"category_scores_gemma":[0.0010678128,0.00022436846,0.0009868856,0.0015477019,0.00018795757,0.0006801864,0.00048688432,0.0007454317,0.0013784472],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00056905893,0.00027651465,0.1725935,0.00024138043,0.00061233324,0.00036194376,0.00008654992,0.61795783,0.0051186127,0.0017773741,0.06321563,0.13718928],"study_design_scores_gemma":[0.00006275921,0.000045771427,0.06249701,0.00006424627,0.000116522955,0.000065649634,0.00009864204,0.91291136,0.004349795,0.0014505313,0.018265935,0.000071804076],"about_ca_topic_score_codex":0.07065853,"about_ca_topic_score_gemma":0.0773133,"teacher_disagreement_score":0.07065853,"about_ca_system_score_codex":0.00083641545,"about_ca_system_score_gemma":0.0012626431,"threshold_uncertainty_score":0.14049453},"labels":[],"label_agreement":null},{"id":"W4327811479","doi":"10.1038/s41597-023-02048-8","title":"A dataset of rodent cerebrovasculature from in vivo multiphoton fluorescence microscopy imaging","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Sunnybrook Health Science Centre","funders":"CIHR Skin Research Training Centre; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; U.S. Department of Health and Human Services; National Institutes of Health; Government of Canada; Japan Agency for Medical Research and Development","keywords":"Fluorescence; Microscopy; Multiphoton fluorescence microscope; Fluorescence microscope; In vivo; Fluorescence-lifetime imaging microscopy; Rodent; Chemistry; Optics; Biology; Physics; Ecology","score_opus":0.015508928196742267,"score_gpt":0.3055314629526942,"score_spread":0.2900225347559519,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4327811479","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23155741,0.013273775,0.058467817,0.0010107972,0.00055267004,0.000754912,0.66451025,0.02079699,0.009075296],"genre_scores_gemma":[0.094132856,0.0027205974,0.05556697,0.00035817074,0.00007058173,0.0006797104,0.8425387,0.001307859,0.00262454],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993772,0.00006123792,0.0000625161,0.00022214603,0.00019714505,0.00007979005],"domain_scores_gemma":[0.99905866,0.0001695931,0.00012481288,0.00027351512,0.00027202923,0.000101352234],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00073452987,0.0016410048,0.0012809577,0.002848624,0.0006847901,0.0011876504,0.0019954199,0.0016732033,0.003672701],"category_scores_gemma":[0.001642808,0.0006331031,0.0013389343,0.0024555684,0.0006194498,0.0006210834,0.0012095268,0.0009998231,0.0032778396],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019598568,0.0009900382,0.029788656,0.008841551,0.0013885556,0.0035487171,0.00048120256,0.0323599,0.2604734,0.0035261582,0.4146678,0.24197412],"study_design_scores_gemma":[0.0005159696,0.0011256577,0.1760339,0.0011863254,0.0011273487,0.017010005,0.00064134836,0.07427153,0.18508078,0.01168989,0.53070414,0.00061312906],"about_ca_topic_score_codex":0.008594854,"about_ca_topic_score_gemma":0.024968999,"teacher_disagreement_score":0.008594854,"about_ca_system_score_codex":0.0008842398,"about_ca_system_score_gemma":0.0014185369,"threshold_uncertainty_score":0.017089605},"labels":[],"label_agreement":null},{"id":"W4328048033","doi":"10.1038/s41597-023-02030-4","title":"A synthetic population for agent-based modelling in Canada","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"demographic modeling and climate adaptation","field":"Decision Sciences","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Economic and Social Research Council; Research Councils UK","keywords":"Population; Computer science; Data science; Demography; Sociology","score_opus":0.4265268679097579,"score_gpt":0.4053863315042952,"score_spread":0.021140536405462707,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4328048033","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6237908,0.0014888849,0.17953895,0.0061460487,0.0005976677,0.0006693181,0.11358248,0.0025891815,0.071596645],"genre_scores_gemma":[0.89660245,0.0005717772,0.06301102,0.00044843458,0.000043815457,0.00038644316,0.027575666,0.00018385386,0.011176501],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997273,0.00008062828,0.000011953595,0.000059191327,0.00005187891,0.00006909939],"domain_scores_gemma":[0.9987581,0.00053685164,0.000055468932,0.00008932008,0.00044555325,0.00011479046],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00067039044,0.00047579128,0.00037381556,0.00058840966,0.0009955155,0.0011208913,0.0014913633,0.0008986069,0.00481815],"category_scores_gemma":[0.0036658554,0.00027475497,0.00071363925,0.0012911003,0.0007263896,0.0005119306,0.0008512853,0.00096353865,0.00035165093],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052517542,0.000025970216,0.005452582,0.00004275232,0.000030909676,0.00008615632,0.00009786094,0.9645299,0.00010430246,0.013345834,0.011643268,0.0045879977],"study_design_scores_gemma":[0.00003800729,0.000007381419,0.0018568181,0.000016845426,0.000009259045,0.000015167834,0.00012883144,0.9827488,0.00010365293,0.00507484,0.009982846,0.000017525781],"about_ca_topic_score_codex":0.8917935,"about_ca_topic_score_gemma":0.84188217,"teacher_disagreement_score":0.10820651,"about_ca_system_score_codex":0.008967706,"about_ca_system_score_gemma":0.008972072,"threshold_uncertainty_score":0.21768749},"labels":[],"label_agreement":null},{"id":"W4362664032","doi":"10.1038/s41597-023-01946-1","title":"Data and Tools Integration in the Canadian Open Neuroscience Platform","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Douglas Mental Health University Institute; Robarts Clinical Trials; Concordia University; McGill University and Génome Québec Innovation Centre; Indoc Research; Ontario Brain Institute; Krembil Foundation; McGill University Health Centre; McGill University; University of Toronto; Western University; McGill Genome Centre; Montreal Neurological Institute and Hospital","funders":"Institut de Cardiologie de Montréal; University of Toronto; McGill University; Simon Fraser University; National Institutes of Health; Canada First Research Excellence Fund; Canada Research Chairs; Health Canada; Government of Ontario; Concordia University; Réseau en Bio-Imagerie du Quebec; Compute Canada; National Institute of Mental Health; Ontario Brain Institute; Fondation Brain Canada; National Institute of Biomedical Imaging and Bioengineering; Université Laval","keywords":"Computer science; Metadata; Data sharing; World Wide Web; Reuse; Data science; Software; Open science; Point (geometry); Data management; Open data; Database; Operating system","score_opus":0.7302211163254171,"score_gpt":0.4968576880176869,"score_spread":0.23336342830773016,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4362664032","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.090860166,0.004023466,0.22607507,0.026583979,0.0011828748,0.004303747,0.21514475,0.08267914,0.34914687],"genre_scores_gemma":[0.21752661,0.0034442155,0.47915003,0.0039952607,0.000250126,0.0028145486,0.21845967,0.0065753185,0.067784235],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9877429,0.0010053782,0.0008431631,0.0015823382,0.006846698,0.0019794628],"domain_scores_gemma":[0.97556204,0.0018635925,0.00084522297,0.0048530805,0.01293605,0.0039399425],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.010322252,0.0011648411,0.0011396555,0.013588074,0.0077092997,0.012393359,0.0046059065,0.001489528,0.01149114],"category_scores_gemma":[0.019931724,0.0009896632,0.0013880255,0.021906404,0.0026051435,0.007950343,0.015247297,0.002394781,0.0056786733],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010453601,0.00041097135,0.027079621,0.0009069804,0.00026548567,0.0012250899,0.0069939094,0.0063303676,0.0055322656,0.13201793,0.48376125,0.33443084],"study_design_scores_gemma":[0.000108350316,0.000053591277,0.019449078,0.000396377,0.00009446176,0.00024663206,0.003083248,0.010189113,0.003970319,0.03288611,0.9292281,0.000294566],"about_ca_topic_score_codex":0.92555535,"about_ca_topic_score_gemma":0.92450815,"teacher_disagreement_score":0.9953941,"about_ca_system_score_codex":0.04098216,"about_ca_system_score_gemma":0.13580549,"threshold_uncertainty_score":0.29734772},"labels":[],"label_agreement":null},{"id":"W4362673363","doi":"10.1038/s41597-023-02069-3","title":"Author Correction: GLORIA - A globally representative hyperspectral in situ dataset for optical sensing of water quality","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Water Quality Monitoring and Analysis","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"","keywords":"Hyperspectral imaging; In situ; Remote sensing; Environmental science; Water quality; Quality (philosophy); Geography; Biology; Physics; Ecology; Meteorology","score_opus":0.12555651376025498,"score_gpt":0.38843788958646935,"score_spread":0.2628813758262144,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4362673363","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008386521,0.00071824685,0.0073992056,0.02902121,0.90404606,0.00010653564,0.03376033,0.0041481303,0.019961676],"genre_scores_gemma":[0.025340343,0.0043167863,0.03573499,0.029412264,0.066970445,0.0005422661,0.093731895,0.021058967,0.722892],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9957158,0.0004278729,0.0006675089,0.00062128954,0.0023625072,0.00020503408],"domain_scores_gemma":[0.9576885,0.0046551,0.0012788457,0.003900313,0.031692788,0.00078444125],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034273395,0.0014928366,0.0012261483,0.0039262944,0.0029717644,0.0040421393,0.0021928572,0.002480517,0.11670678],"category_scores_gemma":[0.0524888,0.0008036523,0.001069107,0.0044211983,0.0014293949,0.0025987518,0.0029466387,0.0051155756,0.08697766],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014005944,0.0000026576367,0.00007246904,0.000060269187,0.0000042263623,0.00006373196,0.00002156369,0.00005429177,0.00006801542,0.00050089246,0.9939156,0.00522219],"study_design_scores_gemma":[0.000015757912,0.0000059022395,0.00042365017,0.0001426492,0.000013562836,0.00015462452,0.00007180627,0.00019174014,0.00045394347,0.00094929885,0.9975555,0.000021407166],"about_ca_topic_score_codex":0.025328303,"about_ca_topic_score_gemma":0.03498912,"teacher_disagreement_score":0.11670678,"about_ca_system_score_codex":0.0022307604,"about_ca_system_score_gemma":0.005937922,"threshold_uncertainty_score":0.39042312},"labels":[],"label_agreement":null},{"id":"W4362717754","doi":"10.1038/s41597-023-02096-0","title":"A standardized catalogue of spectral indices to advance the use of remote sensing in Earth system research","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":182,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"European Space Agency; Deutsche Forschungsgemeinschaft; Niedersächsische Ministerium für Wissenschaft und Kultur","keywords":"Remote sensing; Information retrieval; Earth (classical element); Computer science; Geography; Astronomy; Physics","score_opus":0.12454228153508677,"score_gpt":0.3377334745517816,"score_spread":0.2131911930166948,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4362717754","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011074865,0.002921483,0.49052554,0.0009829164,0.001386371,0.0014777911,0.3411281,0.060497843,0.09000504],"genre_scores_gemma":[0.018527292,0.0031734419,0.4486486,0.0007627443,0.0004735366,0.0018725771,0.4885919,0.018293282,0.019656643],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99723595,0.0004009894,0.0006014709,0.0003136405,0.0012874218,0.00016051151],"domain_scores_gemma":[0.990858,0.0015024449,0.0008109485,0.0023207676,0.0039664507,0.00054146204],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035232643,0.0015818698,0.0010211602,0.011241169,0.00086554757,0.0029068955,0.0020108377,0.0009564392,0.028536756],"category_scores_gemma":[0.011968684,0.0007008494,0.0013176034,0.01251644,0.0006501494,0.0037442518,0.0031768219,0.0022482546,0.030931175],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018632175,0.00019015794,0.0046950565,0.0023669624,0.00009233549,0.0001768278,0.00037377316,0.0037131992,0.01072874,0.029305154,0.64155525,0.30661634],"study_design_scores_gemma":[0.000033032124,0.000039263447,0.0070146285,0.0004630527,0.000038606424,0.00031522135,0.00009243191,0.003197528,0.0038460465,0.013842505,0.9710109,0.00010662321],"about_ca_topic_score_codex":0.0049655167,"about_ca_topic_score_gemma":0.0058643236,"teacher_disagreement_score":0.028536756,"about_ca_system_score_codex":0.00096716057,"about_ca_system_score_gemma":0.0043443968,"threshold_uncertainty_score":0.095465004},"labels":[],"label_agreement":null},{"id":"W4365515241","doi":"10.1038/s41597-023-02106-1","title":"Open science and data sharing in cognitive neuroscience with MouseBytes and MouseBytes+","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Robarts Clinical Trials; Western University","funders":"Canadian Open Neuroscience Platform; Canada First Research Excellence Fund","keywords":"Open science; Cognitive neuroscience; Neuroinformatics; Cognition; Neuroscience; Open data; Cognitive science; Data sharing; Psychology; Data science; Computer science; World Wide Web; Medicine; Physics","score_opus":0.19230217751921452,"score_gpt":0.4030885047671502,"score_spread":0.21078632724793567,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4365515241","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010367807,0.0014277534,0.55126894,0.003147505,0.0008364962,0.0010770535,0.06410025,0.35413587,0.01363824],"genre_scores_gemma":[0.09099859,0.0029393295,0.5203103,0.0031296432,0.00051381084,0.00508241,0.2778685,0.082971394,0.016186118],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99383485,0.0012322621,0.0010549347,0.0017433852,0.0017037404,0.00043082525],"domain_scores_gemma":[0.97744286,0.0045555294,0.001749983,0.011579038,0.0020624362,0.0026101393],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.016910909,0.0022752956,0.0020376977,0.005241521,0.0014463121,0.007446328,0.0068042953,0.0020424188,0.020336624],"category_scores_gemma":[0.027371056,0.0019261378,0.003032707,0.0055854977,0.0025993711,0.012200481,0.017566726,0.0039807195,0.018097814],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0064798063,0.0007346098,0.012650665,0.0046956013,0.0024003158,0.0019857858,0.003151099,0.009516703,0.04551072,0.14311332,0.55788326,0.2118782],"study_design_scores_gemma":[0.0010381939,0.00048055645,0.008110806,0.0008440759,0.000505922,0.0012698898,0.00049898075,0.031739753,0.041092295,0.15951729,0.7543618,0.00054047833],"about_ca_topic_score_codex":0.002424559,"about_ca_topic_score_gemma":0.003019818,"teacher_disagreement_score":0.9931957,"about_ca_system_score_codex":0.001233389,"about_ca_system_score_gemma":0.0047212476,"threshold_uncertainty_score":0.089434505},"labels":[],"label_agreement":null},{"id":"W4366086895","doi":"10.1038/s41597-023-02094-2","title":"CarbonMonitor-Power near-real-time monitoring of global power generation on hourly to daily scales","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric and Environmental Gas Dynamics","field":"Environmental Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hudbay Minerals (Canada); University of Toronto","funders":"China Postdoctoral Science Foundation; Tsinghua University; National Natural Science Foundation of China","keywords":"Renewable energy; Environmental science; Wind power; Solar power; Meteorology; Electricity generation; Range (aeronautics); Fossil fuel; Climatology; Geography; Power (physics); Ecology; Engineering","score_opus":0.021934896263920173,"score_gpt":0.2620825650098672,"score_spread":0.24014766874594703,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366086895","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.076373115,0.0006764695,0.005975445,0.00084924715,0.0002757747,0.0001257791,0.90507746,0.003177437,0.00746931],"genre_scores_gemma":[0.12645346,0.00041031654,0.0072016194,0.00016899961,0.00010875075,0.00016413249,0.86389375,0.0002648917,0.0013341389],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99953294,0.00008033386,0.000046030516,0.0001438036,0.00014174219,0.000055161763],"domain_scores_gemma":[0.99923396,0.00013009392,0.00014618403,0.00017449014,0.00023090123,0.000084495405],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004992549,0.00097530254,0.0005224038,0.0018323646,0.00024351386,0.0007221978,0.0010032066,0.00083985395,0.0031231702],"category_scores_gemma":[0.0022361288,0.00027525364,0.0006151637,0.003225014,0.0002872698,0.0011220488,0.0007807967,0.0008922308,0.0017959038],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046299116,0.00035471877,0.13112116,0.0012919959,0.00064708036,0.0006260107,0.0003573698,0.10143989,0.0038612578,0.00535378,0.699477,0.055006694],"study_design_scores_gemma":[0.00053822243,0.00020167527,0.25543767,0.0003840271,0.00020275074,0.0005632154,0.00060835405,0.18097089,0.008435784,0.008137487,0.5443055,0.0002144346],"about_ca_topic_score_codex":0.018000891,"about_ca_topic_score_gemma":0.029132867,"teacher_disagreement_score":0.018000891,"about_ca_system_score_codex":0.00047767878,"about_ca_system_score_gemma":0.00057237933,"threshold_uncertainty_score":0.03579223},"labels":[],"label_agreement":null},{"id":"W4366280062","doi":"10.1038/s41597-023-02136-9","title":"Author Correction: MIMIC-IV, a freely accessible electronic health record dataset","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":146,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children","funders":"","keywords":"Electronic health record; Computer science; Information retrieval; Health records; World Wide Web; Data science; Health care; Political science","score_opus":0.10061696061468166,"score_gpt":0.3922536860077259,"score_spread":0.29163672539304425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366280062","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001876028,0.001684488,0.010392751,0.10161153,0.67326665,0.0002879503,0.18662778,0.0049669226,0.01928585],"genre_scores_gemma":[0.05687463,0.0067331064,0.04885465,0.08307681,0.07551968,0.0021129716,0.33162826,0.021132559,0.3740674],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9830346,0.0036547487,0.004126071,0.0020146784,0.006552644,0.00061728345],"domain_scores_gemma":[0.86181855,0.039859764,0.0070874556,0.01953425,0.06931688,0.0023830712],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0145501755,0.0013540289,0.0014696493,0.0066806255,0.0032988396,0.005419677,0.0029449344,0.0026751386,0.14779717],"category_scores_gemma":[0.22310162,0.0009783217,0.0013720832,0.008713432,0.0020207048,0.00266577,0.0050564134,0.00502034,0.079385035],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029875297,0.00000315151,0.00021309039,0.00014845027,0.000014499946,0.00009809568,0.00004843079,0.000047601567,0.000034134024,0.0009638288,0.9934237,0.004975156],"study_design_scores_gemma":[0.00006624313,0.000010342915,0.001075847,0.0005945654,0.00005014634,0.0003567891,0.00015216364,0.00024662845,0.0003565661,0.0027528221,0.99429166,0.000046220153],"about_ca_topic_score_codex":0.020912072,"about_ca_topic_score_gemma":0.027979046,"teacher_disagreement_score":0.14779717,"about_ca_system_score_codex":0.00351738,"about_ca_system_score_gemma":0.011809548,"threshold_uncertainty_score":0.49443078},"labels":[],"label_agreement":null},{"id":"W4366824320","doi":"10.1038/s41597-023-02082-6","title":"Three-Dimensional Motion Capture Data of a Movement Screen from 183 Athletes","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Sports Performance and Training","field":"Medicine","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"China Scholarship Council; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Motion capture; Athletes; Movement (music); Motion (physics); Computer science; Physical medicine and rehabilitation; Artificial intelligence; Medicine; Physics; Physical therapy; Acoustics","score_opus":0.12092658849877143,"score_gpt":0.3219926699935155,"score_spread":0.2010660814947441,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366824320","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2511526,0.0014224638,0.016675279,0.00037667245,0.00044325567,0.0014353961,0.7109553,0.0036149332,0.013924142],"genre_scores_gemma":[0.20238926,0.00064660195,0.013010796,0.00014365897,0.00009792076,0.001370386,0.774827,0.00018940159,0.007324991],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9994241,0.00006405903,0.00007774745,0.00014132577,0.00019136752,0.000101446574],"domain_scores_gemma":[0.99931693,0.000116610885,0.00006571616,0.00013261466,0.0003040447,0.00006411505],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005386894,0.0011643465,0.0008149007,0.0019614385,0.0004034883,0.00060566136,0.0007541185,0.0010599055,0.008648175],"category_scores_gemma":[0.0013805723,0.0002785643,0.0008051152,0.0015799128,0.00034726466,0.000269702,0.0008336326,0.00041438465,0.0067471494],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031579915,0.0013968272,0.12324062,0.0044751354,0.0005589634,0.0016048581,0.0006390062,0.015335863,0.049697597,0.0010584648,0.3580107,0.44082394],"study_design_scores_gemma":[0.00035558996,0.0016741778,0.73788714,0.0011163038,0.0003349148,0.0027491467,0.001182404,0.032371756,0.022383502,0.0015005689,0.19813833,0.0003061766],"about_ca_topic_score_codex":0.007847899,"about_ca_topic_score_gemma":0.026769575,"teacher_disagreement_score":0.008648175,"about_ca_system_score_codex":0.00027022013,"about_ca_system_score_gemma":0.00067073427,"threshold_uncertainty_score":0.028931081},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4367312806","doi":"10.1038/s41597-023-02137-8","title":"Author Correction: An analysis-ready and quality controlled resource for pediatric brain white-matter research","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Neonatal and fetal brain pathology","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université de Montréal; Institut Universitaire en Santé Mentale de Québec; Western University; Polytechnique Montréal; Concordia University; University of Toronto","funders":"","keywords":"White matter; Resource (disambiguation); White paper; Quality (philosophy); White (mutation); Psychology; Computer science; Data science; Medicine; Geography; Biology; Physics; Archaeology","score_opus":0.18463636716419232,"score_gpt":0.44278307247781695,"score_spread":0.2581467053136246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4367312806","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00086778397,0.0014261539,0.021886762,0.103652135,0.8051876,0.00043480555,0.04439894,0.0073098503,0.0148360245],"genre_scores_gemma":[0.031442817,0.00698538,0.11910655,0.12463416,0.14564072,0.0024457711,0.07851399,0.033623524,0.45760703],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.973505,0.0052190716,0.0061729164,0.0027136444,0.0114813065,0.0009081391],"domain_scores_gemma":[0.75314945,0.056191124,0.009341727,0.020790333,0.15690196,0.0036253952],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.020281753,0.0019121505,0.0020536722,0.0068641123,0.0035008027,0.006753963,0.004251782,0.004243114,0.14401385],"category_scores_gemma":[0.26436356,0.001350151,0.0017509517,0.0060285972,0.0028095813,0.0026149037,0.00400013,0.008508953,0.07707488],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000046023484,0.0000052130226,0.0001764336,0.00016019974,0.000018809698,0.00014044571,0.00005126614,0.000043320953,0.00008025023,0.0012105585,0.9891766,0.008890952],"study_design_scores_gemma":[0.00006644855,0.000013404317,0.0008921165,0.0006359258,0.000068270965,0.00058127445,0.00009038509,0.00030168187,0.00067028555,0.0027420092,0.9938836,0.00005465861],"about_ca_topic_score_codex":0.017367493,"about_ca_topic_score_gemma":0.024754032,"teacher_disagreement_score":0.14401385,"about_ca_system_score_codex":0.0037583085,"about_ca_system_score_gemma":0.013601105,"threshold_uncertainty_score":0.48177433},"labels":[],"label_agreement":null},{"id":"W4367314599","doi":"10.1038/s41597-023-02104-3","title":"Network Analysis of Academic Medical Center Websites in the United States","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Office of Science; U.S. Department of Energy","keywords":"Centrality; Computer science; World Wide Web; Network analysis; Data science; Identification (biology); Health care; Key (lock); Web resource; Resource (disambiguation); Information retrieval; Web page; Computer security; Computer network; Political science; Engineering","score_opus":0.09259889426856219,"score_gpt":0.34146165408168183,"score_spread":0.24886275981311964,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4367314599","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97624016,0.0006547783,0.0023161788,0.0002913019,0.000012267626,0.0000513344,0.008781478,0.0001841976,0.011468354],"genre_scores_gemma":[0.9871501,0.00039030175,0.0034304736,0.000023920444,0.000007608004,0.00003670693,0.007422624,0.000016771332,0.0015215289],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9994947,0.0001924264,0.000040762574,0.000078938814,0.00014280633,0.000050355804],"domain_scores_gemma":[0.99638546,0.0019987307,0.00081405666,0.00012176387,0.00050104293,0.0001789866],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0005305024,0.0001327654,0.0001440986,0.006472495,0.00047165493,0.00069862406,0.00018551826,0.00017673755,0.0025587159],"category_scores_gemma":[0.0044819866,0.00008436948,0.00019846557,0.008672968,0.00018054663,0.0007063174,0.00038349483,0.00015531492,0.0003109302],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031340847,0.00021158541,0.819736,0.0006347093,0.0002486859,0.0007835701,0.0029414482,0.021118887,0.0034210316,0.009455852,0.017901206,0.12323352],"study_design_scores_gemma":[0.000012621939,0.00011165586,0.9115244,0.00009695651,0.00006899588,0.00057963794,0.0037056943,0.060873963,0.0015519789,0.0028570017,0.01859571,0.000021344635],"about_ca_topic_score_codex":0.02878248,"about_ca_topic_score_gemma":0.039295964,"teacher_disagreement_score":0.99352753,"about_ca_system_score_codex":0.00096120825,"about_ca_system_score_gemma":0.00060001656,"threshold_uncertainty_score":0.057229936},"labels":[],"label_agreement":null},{"id":"W4368360783","doi":"10.1038/s41597-023-02160-9","title":"A dataset of direct observations of sea ice drift and waves in ice","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Arctic and Antarctic ice dynamics","field":"Earth and Planetary Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada; Centre For Cold Ocean Resources Engineering","funders":"Norges Forskningsråd; U.S. Department of Defense","keywords":"Sea ice; Drift ice; Sea ice thickness; Geology; Arctic ice pack; Antarctic sea ice; Sea ice concentration; Cryosphere; Lead (geology); Breakup; Sea ice growth processes; Climatology; Oceanography; Physics; Geomorphology","score_opus":0.06736690355056935,"score_gpt":0.26852421829017853,"score_spread":0.20115731473960918,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4368360783","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2030494,0.0006679757,0.0037670743,0.00008136352,0.00008696592,0.00015457718,0.7871072,0.00062762713,0.004457773],"genre_scores_gemma":[0.10768234,0.00033160648,0.0062524853,0.000052783293,0.00005405352,0.00025758732,0.88405436,0.00007340776,0.0012413809],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9994368,0.000044170676,0.0000654039,0.00017302377,0.00018883782,0.0000916853],"domain_scores_gemma":[0.99865294,0.000177959,0.00023505758,0.00027219456,0.0005028478,0.00015900271],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00044464864,0.0007412283,0.0007447763,0.0021927883,0.00038236388,0.0006203209,0.000689107,0.00063952536,0.002689819],"category_scores_gemma":[0.0010976917,0.00023115001,0.0006497701,0.0037196402,0.0002899938,0.00054719415,0.0008892882,0.00051681366,0.0026559595],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00087605696,0.0013744585,0.52563775,0.003135776,0.00086089544,0.0016576557,0.0012179984,0.031859852,0.041850295,0.0018740419,0.25317138,0.13648377],"study_design_scores_gemma":[0.0001425438,0.00023494239,0.8035797,0.00025528594,0.00012885891,0.00046421582,0.000714745,0.008396461,0.008092653,0.00064607774,0.17725891,0.00008560964],"about_ca_topic_score_codex":0.012251311,"about_ca_topic_score_gemma":0.028804181,"teacher_disagreement_score":0.012251311,"about_ca_system_score_codex":0.00042235304,"about_ca_system_score_gemma":0.001074618,"threshold_uncertainty_score":0.024360001},"labels":[],"label_agreement":null},{"id":"W4375850863","doi":"10.1038/s41597-023-02171-6","title":"Haplotype-resolved genome assembly of Coriaria nepalensis a non-legume nitrogen-fixing shrub","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Legume Nitrogen Fixing Symbiosis","field":"Agricultural and Biological Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; University of British Columbia","funders":"National Natural Science Foundation of China","keywords":"Biology; Genome; Contig; Frankia; Botany; Sequence assembly; Nitrogen fixation; Whole genome sequencing; Gene; Genetics; Root nodule; Bacteria; Transcriptome","score_opus":0.04886845499612701,"score_gpt":0.2565052895636224,"score_spread":0.20763683456749538,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4375850863","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8445623,0.0019736106,0.071482815,0.00036685372,0.000107490356,0.00048117957,0.073453344,0.0014389714,0.006133358],"genre_scores_gemma":[0.5758078,0.0021011853,0.1772267,0.0001802665,0.000041927713,0.0005430847,0.23726927,0.0008264745,0.0060033933],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99988675,0.0000095594505,0.00000783614,0.00005671854,0.000023669163,0.000015353773],"domain_scores_gemma":[0.9997267,0.000065566295,0.000037804108,0.000031570493,0.000099454286,0.000038902093],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00017358121,0.0005008574,0.0005373045,0.0012934671,0.0005495148,0.0005271226,0.00042980668,0.00043744428,0.0015896601],"category_scores_gemma":[0.0006755444,0.00030109767,0.00080199377,0.0014564223,0.00015750894,0.00025244654,0.00038729073,0.0007238662,0.0009390022],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007015223,0.00012565097,0.015567893,0.001077626,0.00017649216,0.001746856,0.0011105217,0.0063844095,0.915268,0.00089290214,0.0024721543,0.054475803],"study_design_scores_gemma":[0.00047411234,0.0006698495,0.5109195,0.0006278315,0.0011277421,0.0033329413,0.0018219079,0.08364064,0.24963532,0.003137724,0.14429177,0.00032068146],"about_ca_topic_score_codex":0.012682193,"about_ca_topic_score_gemma":0.021132415,"teacher_disagreement_score":0.012682193,"about_ca_system_score_codex":0.00048101402,"about_ca_system_score_gemma":0.001071616,"threshold_uncertainty_score":0.025216758},"labels":[],"label_agreement":null},{"id":"W4376127614","doi":"10.1038/s41597-023-02080-8","title":"Social and moral psychology of COVID-19 across 69 countries","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 and Mental Health","field":"Psychology","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; Western University; Carleton University; University of Waterloo; Toronto Metropolitan University; University of Toronto; University of British Columbia","funders":"Biotechnology and Biological Sciences Research Council; Medical Research Council","keywords":"Pandemic; Public health; Psychology; Social psychology; General Social Survey; Survey data collection; Personality psychology; Coronavirus disease 2019 (COVID-19); Personality; Sociology; Social science; Medicine","score_opus":0.40192822931789274,"score_gpt":0.5808020181795318,"score_spread":0.17887378886163907,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4376127614","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98913157,0.00032787488,0.00020112707,0.00064263574,0.000015085336,0.0000732218,0.006600772,0.000006632824,0.0030011379],"genre_scores_gemma":[0.99176955,0.0004310281,0.00041298094,0.00031460673,0.000015844673,0.0002625073,0.0062188385,0.000009845483,0.0005648841],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9978942,0.0011598703,0.00016496136,0.00020860728,0.00021128841,0.0003610612],"domain_scores_gemma":[0.9938403,0.0022248207,0.0017829739,0.00046609654,0.0009078748,0.0007780208],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028928004,0.00020664086,0.00033037126,0.0017880634,0.0014174464,0.0013824715,0.00033123567,0.0004865551,0.0030013167],"category_scores_gemma":[0.009583076,0.00023055443,0.00034671358,0.0030517553,0.0010267938,0.00082838064,0.0023434388,0.0012470976,0.00057132717],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009410941,0.0001390485,0.9617731,0.00014675106,0.000057349083,0.0000953416,0.016067008,0.000212332,0.00014384897,0.00079160795,0.009102345,0.011377048],"study_design_scores_gemma":[0.000011858741,0.00003716774,0.9771408,0.00012075044,0.000011167169,0.000061854706,0.015567639,0.00016168428,0.000070013135,0.00025709436,0.0065430296,0.000016940592],"about_ca_topic_score_codex":0.029610164,"about_ca_topic_score_gemma":0.026880113,"teacher_disagreement_score":0.029610164,"about_ca_system_score_codex":0.0011082898,"about_ca_system_score_gemma":0.0009962579,"threshold_uncertainty_score":0.05887562},"labels":[],"label_agreement":null},{"id":"W4376133730","doi":"10.1038/s41597-023-02163-6","title":"A spatio-temporal dataset on food flows for four West African cities","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Urban Agriculture and Sustainability","field":"Agricultural and Biological Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Bundesministerium für Bildung und Forschung; Consortium of International Agricultural Research Centers; Deutsche Forschungsgemeinschaft","keywords":"Food security; Geography; Sustainability; Psychological intervention; Resilience (materials science); Socioeconomics; Environmental planning; Business; Agriculture; Ecology; Economics","score_opus":0.10629855697655285,"score_gpt":0.2652192563099462,"score_spread":0.15892069933339334,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4376133730","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.122494526,0.00031397407,0.0006775488,0.00030450177,0.000029510858,0.00012568787,0.87406045,0.00018366215,0.0018101004],"genre_scores_gemma":[0.10816235,0.00036457326,0.0042818943,0.000058273337,0.000016909382,0.0005473885,0.88531446,0.00003309673,0.0012210045],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99952614,0.0000917731,0.00008077026,0.00009960394,0.00010285336,0.0000988784],"domain_scores_gemma":[0.9984738,0.0004077941,0.0003576756,0.00021001152,0.00040041344,0.0001502288],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00071860204,0.00059744413,0.0004818306,0.004261724,0.00063796237,0.0008572592,0.0006533633,0.00081908517,0.0033008656],"category_scores_gemma":[0.0029935716,0.00029657345,0.0005097477,0.0093626585,0.0002559613,0.0006811519,0.00094816275,0.0006618258,0.001722268],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010350904,0.0006696444,0.5449519,0.0033750231,0.00050778553,0.002022885,0.004315959,0.023155749,0.0055807773,0.00485208,0.33696628,0.07256672],"study_design_scores_gemma":[0.00016698132,0.000101079844,0.70323735,0.00074547983,0.00009537757,0.0004044154,0.0074637234,0.0131157795,0.0026207443,0.0013607275,0.2705764,0.000112103065],"about_ca_topic_score_codex":0.095513724,"about_ca_topic_score_gemma":0.107896306,"teacher_disagreement_score":0.095513724,"about_ca_system_score_codex":0.0011849017,"about_ca_system_score_gemma":0.0016466908,"threshold_uncertainty_score":0.18991554},"labels":[],"label_agreement":null},{"id":"W4376503834","doi":"10.1038/s41597-023-02177-0","title":"An inventory of greenhouse gas emissions due to natural gas pipeline incidents in the United States and Canada from 1980s to 2021","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric and Environmental Gas Dynamics","field":"Environmental Science","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Calgary","funders":"Natural Science Foundation of Jiangsu Province; Government of Jiangsu Province; Tencent","keywords":"Greenhouse gas; Natural gas; Environmental science; Pipeline transport; Pipeline (software); Methane; Fugitive emissions; Compressor station; Emission inventory; Natural resource economics; Environmental protection; Environmental engineering; Waste management; Engineering; Geography; Meteorology; Ecology; Economics; Air quality index","score_opus":0.015316847315076224,"score_gpt":0.24122568541476386,"score_spread":0.22590883809968765,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4376503834","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.44756007,0.00277802,0.0025587038,0.0005597515,0.000038051538,0.00016443308,0.52532125,0.00041947106,0.020600185],"genre_scores_gemma":[0.585932,0.0065984325,0.0057619363,0.00016606288,0.000017933355,0.00011943245,0.3906688,0.00005826932,0.010677073],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99956995,0.0000115851035,0.000028127824,0.00006474186,0.00021802273,0.00010755168],"domain_scores_gemma":[0.9984251,0.000058643567,0.00014835401,0.000046224563,0.0011844074,0.0001372875],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003179677,0.0008257724,0.00035906216,0.007576488,0.0013616438,0.0013618441,0.00075226714,0.0002491052,0.0013408695],"category_scores_gemma":[0.0008498229,0.0003697165,0.0007280986,0.0143343005,0.0003169251,0.00062612916,0.00056894944,0.0004698593,0.00032614492],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000319518,0.0001862728,0.73194826,0.0007956698,0.0007552539,0.0008371771,0.0010804062,0.08859576,0.0023470582,0.004985625,0.076608926,0.09154009],"study_design_scores_gemma":[0.000018276212,0.00003095699,0.8909959,0.00027157526,0.00022919496,0.00014651421,0.0017896767,0.022703357,0.0019378434,0.00054808776,0.08122888,0.000099762045],"about_ca_topic_score_codex":0.995415,"about_ca_topic_score_gemma":0.9959973,"teacher_disagreement_score":0.029049262,"about_ca_system_score_codex":0.029049262,"about_ca_system_score_gemma":0.037878636,"threshold_uncertainty_score":0.21076816},"labels":[],"label_agreement":null},{"id":"W4376959108","doi":"10.1038/s41597-023-02214-y","title":"Responses of pyramidal cell somata and apical dendrites in mouse visual cortex over multiple days","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neural dynamics and brain function","field":"Neuroscience","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université de Montréal; Vector Institute; Canadian Institute for Advanced Research; Montreal Neurological Institute and Hospital; University of Toronto; Mila - Quebec Artificial Intelligence Institute; The Scarborough Hospital; York University","funders":"Natural Sciences and Engineering Research Council of Canada; Alliance de recherche numérique du Canada; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Government of Canada; Canadian Institute for Advanced Research; Alfred P. Sloan Foundation","keywords":"Visual cortex; Pyramidal cell; Neuroscience; Biology; Cortex (anatomy); Cerebral cortex; Anatomy; Hippocampus","score_opus":0.06469699247135695,"score_gpt":0.31159061401491306,"score_spread":0.2468936215435561,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4376959108","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98878014,0.00017838685,0.0018719769,0.000044897308,0.000006711159,0.000011123827,0.008432967,0.00020572243,0.00046802545],"genre_scores_gemma":[0.97681105,0.0002811567,0.003910459,0.000075084274,0.0000090433105,0.000062153864,0.017710885,0.00015384387,0.000986259],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9998142,0.0000089163195,0.00000880206,0.00006698365,0.00005763821,0.00004343464],"domain_scores_gemma":[0.99952817,0.000062081985,0.00010882562,0.000060036768,0.000119475284,0.00012139054],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00017891465,0.00021229668,0.00044673507,0.0008090155,0.00020355989,0.00028966914,0.00030984182,0.00041126314,0.0007253599],"category_scores_gemma":[0.0005070618,0.00021004365,0.00037922402,0.00046374043,0.00025909895,0.00024413085,0.0004811991,0.0005001938,0.00029006737],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040159893,0.00006833158,0.03582578,0.00019340617,0.0001356993,0.00017621677,0.000190854,0.0033182572,0.9447312,0.00015855962,0.0018383691,0.012961832],"study_design_scores_gemma":[0.000019770181,0.00015103685,0.9109675,0.000023604742,0.00006345703,0.00044797754,0.00013836255,0.009190714,0.0767978,0.000421441,0.0017349116,0.00004334192],"about_ca_topic_score_codex":0.0049268473,"about_ca_topic_score_gemma":0.010591735,"teacher_disagreement_score":0.0049268473,"about_ca_system_score_codex":0.00048036114,"about_ca_system_score_gemma":0.00027578272,"threshold_uncertainty_score":0.009796381},"labels":[],"label_agreement":null},{"id":"W4377093063","doi":"10.1038/s41597-023-02174-3","title":"M100 ExaData: a data collection campaign on the CINECA’s Marconi100 Tier-0 supercomputer","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Supercomputer; Computer science; Workload; Software; Data science; Operating system","score_opus":0.15939045995022397,"score_gpt":0.31590441355818877,"score_spread":0.1565139536079648,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377093063","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20059331,0.0015815403,0.0024673068,0.0039286627,0.0007941352,0.00043905855,0.7612064,0.011215774,0.017773828],"genre_scores_gemma":[0.080270216,0.00042905597,0.0045435424,0.00037052465,0.00026790556,0.00022706143,0.91108835,0.00058521784,0.002218205],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99648595,0.0005374681,0.00026660346,0.0007892175,0.001499631,0.0004212316],"domain_scores_gemma":[0.9944865,0.00059241435,0.00036477143,0.0014331363,0.002066018,0.0010571588],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003130001,0.0012342199,0.0010008652,0.0035632371,0.00158733,0.0019831236,0.0016378103,0.0011371841,0.003818905],"category_scores_gemma":[0.006539869,0.00043824792,0.0005173287,0.0064119305,0.00090520235,0.0020217502,0.0021775828,0.0017068703,0.0037688406],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008740369,0.0005176201,0.03717201,0.00063491066,0.00013174831,0.00048097825,0.000721627,0.004266414,0.006072037,0.0028587459,0.9179256,0.028344262],"study_design_scores_gemma":[0.0004917468,0.0002834975,0.21362907,0.00028712692,0.000087578206,0.0003911792,0.0020127185,0.022612425,0.013756584,0.0024544154,0.7438113,0.00018230092],"about_ca_topic_score_codex":0.041894037,"about_ca_topic_score_gemma":0.040964697,"teacher_disagreement_score":0.041894037,"about_ca_system_score_codex":0.0020124384,"about_ca_system_score_gemma":0.0030568137,"threshold_uncertainty_score":0.08330041},"labels":[],"label_agreement":null},{"id":"W4377096557","doi":"10.1038/s41597-023-02167-2","title":"FAIR in action - a flexible framework to guide FAIRification","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Research Data Management Practices","field":"Computer Science","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"Biotechnology and Biological Sciences Research Council; Innovative Medicines Initiative","keywords":"Action (physics); Computer science; Physics","score_opus":0.4095442173766862,"score_gpt":0.49234319063888665,"score_spread":0.08279897326220043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377096557","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013704533,0.00015245087,0.97336596,0.0035570522,0.00020707292,0.0009736025,0.00071852515,0.012220071,0.00743487],"genre_scores_gemma":[0.038973186,0.0001890717,0.9478697,0.0014625238,0.00013190776,0.0016836018,0.001961033,0.0033673176,0.0043616965],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9341899,0.03471262,0.007536028,0.009581501,0.010119719,0.0038602035],"domain_scores_gemma":[0.8796653,0.07045101,0.0046364293,0.028776841,0.011012554,0.0054578516],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.13828212,0.0026223613,0.0019362773,0.0084407,0.0062571047,0.021094663,0.009420951,0.0074786977,0.017253852],"category_scores_gemma":[0.17195505,0.0026044918,0.005072617,0.0033698669,0.010702271,0.020730084,0.026207866,0.011157737,0.006777682],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038480823,0.0003339886,0.0037032983,0.0004628485,0.00016759858,0.0006189523,0.005135241,0.030946083,0.0018773882,0.83211863,0.030885125,0.09336599],"study_design_scores_gemma":[0.00016202968,0.00006714121,0.0003936126,0.0006683881,0.000065921566,0.00018379024,0.00079542253,0.12302561,0.0045500286,0.68669754,0.18318678,0.00020380352],"about_ca_topic_score_codex":0.01811848,"about_ca_topic_score_gemma":0.017517122,"teacher_disagreement_score":0.99057907,"about_ca_system_score_codex":0.008437393,"about_ca_system_score_gemma":0.020644942,"threshold_uncertainty_score":0.7313145},"labels":[],"label_agreement":null},{"id":"W4378530983","doi":"10.1038/s41597-023-02222-y","title":"A compendium of bacterial and archaeal single-cell amplified genomes from oxygen deficient marine waters","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Microbial Community Ecology and Physiology","field":"Environmental Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Genome British Columbia; University of British Columbia, Okanagan Campus; University of British Columbia","funders":"CHIST-ERA; Leibniz-Gemeinschaft; Office of Science; Directorate for Biological Sciences; Agencia Nacional de Investigación y Desarrollo; Natural Sciences and Engineering Research Council of Canada; Universidad de Concepción; Genome British Columbia; University of Hawai'i; Agenția Națională pentru Cercetare și Dezvoltare; Joint Genome Institute; Canadian Institute for Advanced Research; U.S. Department of Energy; Georgia Institute of Technology; Southern California Gas Company; Simons Foundation; National Science Foundation","keywords":"Biogeochemical cycle; Anoxic waters; Oxygen minimum zone; Compendium; Ecology; Biology; Ecosystem; Organism; Environmental science; Geography; Upwelling","score_opus":0.051002630215706043,"score_gpt":0.23699682538906905,"score_spread":0.18599419517336302,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4378530983","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12976395,0.009314232,0.04566644,0.0004044465,0.00037795794,0.00044560546,0.80443287,0.0031331752,0.006461263],"genre_scores_gemma":[0.03805165,0.0033961511,0.09244708,0.00023932544,0.00008521572,0.0004934857,0.8629591,0.0006189606,0.001709058],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9987991,0.0001148588,0.00019418659,0.00036971387,0.00039859858,0.00012351599],"domain_scores_gemma":[0.9975453,0.0007108125,0.000574024,0.00039048173,0.00053485896,0.0002444945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011399954,0.0017211599,0.0015150959,0.0044180406,0.00083183846,0.0018323263,0.0013923834,0.00091777294,0.0031051284],"category_scores_gemma":[0.0037369714,0.0007746765,0.0015278741,0.0053940862,0.00039272147,0.00085753336,0.0013407966,0.0016278225,0.0037329863],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014083241,0.00016568424,0.012783365,0.006415937,0.00047167327,0.001565268,0.0010876154,0.0019100007,0.82166225,0.001009847,0.023258261,0.12826188],"study_design_scores_gemma":[0.00030564316,0.00071285234,0.23559879,0.0021436457,0.0016047028,0.005795756,0.00083443563,0.005918378,0.14574519,0.0021145788,0.5987957,0.0004303675],"about_ca_topic_score_codex":0.0016748171,"about_ca_topic_score_gemma":0.004636383,"teacher_disagreement_score":0.0044180406,"about_ca_system_score_codex":0.0003890911,"about_ca_system_score_gemma":0.0020174051,"threshold_uncertainty_score":0.010387719},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4379259280","doi":"10.1038/s41597-023-02264-2","title":"The FORCIS database: A global census of planktonic Foraminifera from ocean waters","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Natural Environment Research Council; Labex OT-Med; Max-Planck-Institut für Chemie; Fondation pour la Recherche sur la Biodiversite; Agence Nationale de la Recherche; Sight Research UK","keywords":"Foraminifera; Census; Oceanography; Plankton; Geography; Geology; Benthic zone; Population; Demography","score_opus":0.0734677861201946,"score_gpt":0.30287168208383597,"score_spread":0.22940389596364136,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379259280","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009653219,0.0006008616,0.00041803118,0.000062292136,0.000022593185,0.00007465218,0.9874127,0.00028351299,0.0014720957],"genre_scores_gemma":[0.0123430835,0.00055511337,0.0025353269,0.00006166168,0.000027091643,0.0003189718,0.98350334,0.000076067605,0.00057928136],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986526,0.00009532752,0.00051728456,0.00031896893,0.0002792431,0.00013648813],"domain_scores_gemma":[0.9942688,0.00045572818,0.0019451273,0.0006964796,0.002113898,0.0005200151],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012282454,0.0015982753,0.0014240193,0.010546766,0.0005276232,0.0013371685,0.0019837476,0.0008495617,0.011429449],"category_scores_gemma":[0.0052092145,0.00052402064,0.00082296215,0.014864214,0.000300147,0.0017989906,0.002627732,0.00079828553,0.0074344156],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00067055353,0.00009765082,0.30584186,0.007037703,0.0008930355,0.0005955075,0.0009857058,0.0011945703,0.0061840275,0.0026385132,0.5725529,0.10130792],"study_design_scores_gemma":[0.00021289119,0.000082567625,0.40022743,0.00093548314,0.00029244137,0.0003327017,0.00086883095,0.0006942184,0.0018779997,0.00086983404,0.59348863,0.00011703675],"about_ca_topic_score_codex":0.041709356,"about_ca_topic_score_gemma":0.0307703,"teacher_disagreement_score":0.041709356,"about_ca_system_score_codex":0.00085651584,"about_ca_system_score_gemma":0.00337189,"threshold_uncertainty_score":0.08293319},"labels":[],"label_agreement":null},{"id":"W4379520140","doi":"10.1038/s41597-023-02058-6","title":"A 100-member ensemble simulations of global historical (1951–2010) wave heights","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ocean Waves and Remote Sensing","field":"Earth and Planetary Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"Japan Agency for Marine-Earth Science and Technology; Ministry of Education, Culture, Sports, Science and Technology","keywords":"Climatology; Scale (ratio); Submarine pipeline; Climate change; Environmental science; Temporal scales; Meteorology; Geography; Geology; Oceanography; Cartography","score_opus":0.12310294191506743,"score_gpt":0.28208088045440594,"score_spread":0.1589779385393385,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379520140","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9748617,0.00013000947,0.0048854323,0.00025654482,0.000115628194,0.000039742572,0.015256941,0.00036527388,0.0040887496],"genre_scores_gemma":[0.9689408,0.0000653194,0.005673566,0.00008316663,0.00003573604,0.00008656271,0.024292387,0.0000747284,0.0007477435],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997688,0.000070865746,0.000015222125,0.00006892724,0.000031091327,0.000045030964],"domain_scores_gemma":[0.9990252,0.00038564566,0.00006351764,0.00020040928,0.0002104241,0.00011485993],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013766636,0.0006152335,0.00071909564,0.0005015454,0.0005476698,0.0006017083,0.0010729388,0.0010945799,0.0020946495],"category_scores_gemma":[0.0016078998,0.00042003355,0.0013231667,0.00080697704,0.00043153646,0.00074181444,0.0005860413,0.001069107,0.00040414286],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022737368,0.00019069738,0.022620609,0.00004080343,0.00022797729,0.00006794766,0.000036648653,0.9638634,0.00075176463,0.0007514535,0.0049650357,0.0062563703],"study_design_scores_gemma":[0.00012646035,0.00007943485,0.017955776,0.000011905419,0.00005835092,0.000015026139,0.000047146434,0.9788199,0.0005179939,0.0004690173,0.0018697768,0.000029144438],"about_ca_topic_score_codex":0.043737087,"about_ca_topic_score_gemma":0.035088852,"teacher_disagreement_score":0.043737087,"about_ca_system_score_codex":0.0007618371,"about_ca_system_score_gemma":0.00088385947,"threshold_uncertainty_score":0.086965024},"labels":[],"label_agreement":null},{"id":"W4379875502","doi":"10.1038/s41597-023-02284-y","title":"Carbon Monitor Europe near-real-time daily CO2 emissions for 27 EU countries and the United Kingdom","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Air Quality and Health Impacts","field":"Environmental Science","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Key Research and Development Program of China; National Natural Science Foundation of China","keywords":"Greenhouse gas; Aviation; European union; Business; Environmental science; Natural resource economics; International trade; Economics; Engineering","score_opus":0.10521872005893303,"score_gpt":0.3466945938191512,"score_spread":0.24147587376021817,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379875502","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3610608,0.00093902944,0.0021613387,0.00024809464,0.00007613167,0.00006249354,0.62714463,0.000443864,0.007863545],"genre_scores_gemma":[0.2884322,0.0006700382,0.0025333287,0.00009523394,0.000021526228,0.00012823552,0.7052918,0.0001047964,0.0027227798],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9994603,0.000089579575,0.00007250212,0.00015840118,0.00014609197,0.00007312917],"domain_scores_gemma":[0.9990495,0.00016148298,0.0002402061,0.00015659246,0.0003290968,0.000063186344],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036831372,0.00049688085,0.00045455722,0.0016547793,0.00018379983,0.00074404385,0.00028437856,0.0004707903,0.002348881],"category_scores_gemma":[0.001398338,0.00021662087,0.0004952741,0.0032698526,0.00015561363,0.00059546094,0.0006307488,0.0002835939,0.0012339889],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010988452,0.00022948204,0.59516,0.0016600051,0.0011773536,0.0009858866,0.00044631714,0.06527128,0.0051632603,0.003779183,0.2656451,0.059383217],"study_design_scores_gemma":[0.000098254015,0.00010941049,0.81156933,0.00018500697,0.00013358888,0.00038111865,0.00044387954,0.013988925,0.005566175,0.00033600934,0.16710557,0.000082713166],"about_ca_topic_score_codex":0.04994437,"about_ca_topic_score_gemma":0.04443912,"teacher_disagreement_score":0.04994437,"about_ca_system_score_codex":0.0004994691,"about_ca_system_score_gemma":0.0004442679,"threshold_uncertainty_score":0.0993073},"labels":[],"label_agreement":null},{"id":"W4381684967","doi":"10.1038/s41597-023-02273-1","title":"A Database of Snow on Sea Ice in the Central Arctic Collected during the MOSAiC expedition","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Arctic and Antarctic ice dynamics","field":"Earth and Planetary Sciences","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"Swiss Polar Institute; European Commission; WSL-Institut für Schnee- und Lawinenforschung SLF; National Science Foundation","keywords":"Snow; Arctic; Sea ice; Sea ice thickness; Arctic ice pack; Geology; Cryosphere; Atmospheric sciences; Environmental science; Climatology; Oceanography; Geomorphology","score_opus":0.03270510775488677,"score_gpt":0.24251858032489781,"score_spread":0.20981347257001104,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4381684967","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13894072,0.00079983106,0.00036049733,0.000048775448,0.00004205216,0.000079244775,0.85712326,0.00024116591,0.002364488],"genre_scores_gemma":[0.047562003,0.00030970096,0.0015932366,0.000027288392,0.000021382473,0.000117110605,0.94969225,0.00003862325,0.0006384392],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995834,0.000030860047,0.00006779171,0.00014298323,0.00010682296,0.00006811107],"domain_scores_gemma":[0.9989774,0.00010318467,0.00024934462,0.00014342532,0.00034381935,0.00018284887],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036967068,0.00058067613,0.0006025143,0.004029204,0.00062944146,0.0008066134,0.0006277816,0.00058664626,0.0022744082],"category_scores_gemma":[0.0010133615,0.00023862792,0.00043468748,0.0055869995,0.00027713642,0.00040218452,0.00072315166,0.0002841154,0.0020194848],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014472106,0.0004146448,0.59658974,0.003197547,0.0006162784,0.0025059483,0.0025258134,0.006616114,0.015656758,0.0010575742,0.28286296,0.086509384],"study_design_scores_gemma":[0.00009238357,0.00009272915,0.78220916,0.0004813862,0.00009660245,0.00041492382,0.0014039095,0.001745281,0.0020075087,0.00027260825,0.21113247,0.000051019935],"about_ca_topic_score_codex":0.059816945,"about_ca_topic_score_gemma":0.1252887,"teacher_disagreement_score":0.059816945,"about_ca_system_score_codex":0.0007084426,"about_ca_system_score_gemma":0.001960356,"threshold_uncertainty_score":0.11893755},"labels":[],"label_agreement":null},{"id":"W4381839070","doi":"10.1038/s41597-023-02303-y","title":"DOCU-CLIM: A global documentary climate dataset for climate reconstructions","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Tree-ring climate responses","field":"Earth and Planetary Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"National Supercomputing Centre Singapore; Centro Svizzero di Calcolo Scientifico; European Commission; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Documentary evidence; Climatology; Climate change; Proxy (statistics); Geography; Precipitation; Series (stratigraphy); Newspaper; Meteorology; Archaeology; Geology; Computer science; Political science; Oceanography","score_opus":0.06823592943993621,"score_gpt":0.3269455598863753,"score_spread":0.2587096304464391,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4381839070","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00085044996,0.00004617755,0.00026240924,0.00004666269,0.000019814648,0.00001791708,0.9976217,0.0007301964,0.00040469944],"genre_scores_gemma":[0.0018475057,0.000031568958,0.0010445253,0.000017096309,0.000008700336,0.000079096186,0.9966852,0.000096485186,0.00018988256],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993831,0.00011594514,0.000104018036,0.00016434633,0.00015212168,0.0000804721],"domain_scores_gemma":[0.99793303,0.00042749228,0.00027123728,0.00053482194,0.0006007097,0.0002326565],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001048229,0.001502832,0.00095074804,0.0025793263,0.00049877475,0.0011404081,0.0024791027,0.0013988314,0.022462424],"category_scores_gemma":[0.004003747,0.0005442317,0.0009523709,0.00518382,0.00026566585,0.0011017512,0.0010637074,0.0016688901,0.024625821],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013374152,0.00006400572,0.0061207633,0.00088784326,0.00014262789,0.000101927384,0.00008243899,0.0043663355,0.0007689262,0.0010040798,0.97758806,0.00873924],"study_design_scores_gemma":[0.000746423,0.0000506069,0.045704115,0.00026993474,0.000110436165,0.00017771462,0.00023533126,0.012973294,0.002016268,0.0021079562,0.93548626,0.00012160079],"about_ca_topic_score_codex":0.026723417,"about_ca_topic_score_gemma":0.036077958,"teacher_disagreement_score":0.026723417,"about_ca_system_score_codex":0.00085968996,"about_ca_system_score_gemma":0.0020194184,"threshold_uncertainty_score":0.07514429},"labels":[],"label_agreement":null},{"id":"W4382501545","doi":"10.1038/s41597-023-02320-x","title":"Author Correction: MusMorph, a database of standardized mouse morphology data for morphometric meta-analyses","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Bat Biology and Ecology Studies","field":"Agricultural and Biological Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Alberta Children's Hospital; University of Alberta; Alberta Bone and Joint Health Institute; University of Calgary","funders":"","keywords":"Morphology (biology); Meta-analysis; Database; Information retrieval; Computer science; Biology; Zoology; Medicine; Pathology","score_opus":0.4249550949419449,"score_gpt":0.3974539525286545,"score_spread":0.02750114241329038,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382501545","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015398597,0.0011797809,0.020108324,0.035157755,0.8130407,0.00014436088,0.096831836,0.009431031,0.022566343],"genre_scores_gemma":[0.02970183,0.0043142363,0.06767988,0.030373702,0.06828236,0.00080052705,0.16733943,0.0488327,0.58267534],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9949818,0.0007001995,0.0010192324,0.0007602175,0.00232505,0.00021357264],"domain_scores_gemma":[0.9524537,0.011056222,0.002072749,0.0050057597,0.028409442,0.001002261],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005620503,0.0018784872,0.0015640038,0.0056955204,0.0028146252,0.0038110635,0.0027705426,0.0024353305,0.23727705],"category_scores_gemma":[0.08032227,0.0012298978,0.0013477696,0.005124896,0.0013551629,0.0026181194,0.0031134302,0.005565075,0.11256215],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028826324,0.0000037751845,0.00014753091,0.00013473255,0.000013241614,0.00010383431,0.00003450882,0.00007486533,0.00011706303,0.000847429,0.99164397,0.0068502133],"study_design_scores_gemma":[0.000030503768,0.0000083392815,0.0007226814,0.0002796634,0.000036617967,0.00034179204,0.000055986682,0.00021223463,0.00050144346,0.0017445971,0.99603206,0.000034166576],"about_ca_topic_score_codex":0.013421195,"about_ca_topic_score_gemma":0.022227013,"teacher_disagreement_score":0.23727705,"about_ca_system_score_codex":0.0020143976,"about_ca_system_score_gemma":0.005294459,"threshold_uncertainty_score":0.7937708},"labels":[],"label_agreement":null},{"id":"W4383711048","doi":"10.1038/s41597-023-02614-0","title":"A comparison of neuroelectrophysiology databases","year":2023,"lang":"en","type":"review","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Baycrest Hospital; Indoc Research; Ontario Brain Institute","funders":"National Institutes of Health; National Institute of Neurological Disorders and Stroke; National Institute of Mental Health; University of Southern California; U.S. Department of Health and Human Services","keywords":"Neuroinformatics; Data sharing; Computer science; Data science; Data integration; Field (mathematics); World Wide Web; Metadata; Database","score_opus":0.6093313146232971,"score_gpt":0.49693759619651007,"score_spread":0.11239371842678708,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4383711048","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030802966,0.97694105,0.005622914,0.0015524824,0.00041332896,0.00017605524,0.001689526,0.000106982385,0.010417319],"genre_scores_gemma":[0.0225712,0.95256054,0.016303157,0.0011054651,0.00032818402,0.0004632062,0.005269413,0.00007152489,0.0013273746],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99686027,0.000855134,0.00078222324,0.00041947028,0.00096580345,0.00011716682],"domain_scores_gemma":[0.989368,0.0062069166,0.0008390408,0.0005584651,0.0026952797,0.00033222596],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008540276,0.0006497329,0.0014972306,0.012062209,0.0004848738,0.002908051,0.0020919482,0.00092354,0.0037508118],"category_scores_gemma":[0.016632494,0.00043806396,0.0011211145,0.013456255,0.0006774653,0.0040944056,0.0019511796,0.0010775446,0.0016034433],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030650408,0.000053435608,0.00194173,0.021173544,0.00049685326,0.00012500546,0.00030954738,0.0004762021,0.000678,0.016308516,0.011774712,0.94635594],"study_design_scores_gemma":[0.00009082079,0.00017995127,0.00666164,0.017616816,0.0009268846,0.0013040097,0.0006050162,0.00045821714,0.0017504788,0.009008793,0.96131945,0.000077984645],"about_ca_topic_score_codex":0.002057909,"about_ca_topic_score_gemma":0.0027201972,"teacher_disagreement_score":0.012062209,"about_ca_system_score_codex":0.0014538455,"about_ca_system_score_gemma":0.00547935,"threshold_uncertainty_score":0.045165837},"labels":[],"label_agreement":null},{"id":"W4383907966","doi":"10.1038/s41597-023-02356-z","title":"Haplotype-resolved chromosomal-level assembly of wasabi (Eutrema japonicum) genome","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Virus Research Studies","field":"Agricultural and Biological Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics; Japan Society for the Promotion of Science","keywords":"Haplotype; Genome; Biology; Genetics; Computational biology; Evolutionary biology; Gene; Genotype","score_opus":0.21633691570144425,"score_gpt":0.3191772619451976,"score_spread":0.10284034624375332,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4383907966","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8677643,0.0023315365,0.089238524,0.00025269118,0.00017225191,0.00027058073,0.033649933,0.0021037408,0.0042164875],"genre_scores_gemma":[0.6020652,0.0018751133,0.2224041,0.00014598231,0.00005026677,0.00041697908,0.16740151,0.0013487317,0.0042921915],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99977785,0.000023581311,0.000021577955,0.000104541235,0.000045637862,0.000026800624],"domain_scores_gemma":[0.99979156,0.00003024516,0.000037187998,0.00003367212,0.00007505422,0.00003228688],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003500491,0.000560703,0.000606782,0.0011236425,0.0006143115,0.00047366126,0.0004772095,0.00033833983,0.0013749144],"category_scores_gemma":[0.00066327903,0.00039707075,0.0007639993,0.0013703,0.00015080463,0.00040412808,0.00048494636,0.00072733127,0.0006467094],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006215241,0.00007167024,0.008196881,0.0008673634,0.00010188057,0.00053310377,0.00065185595,0.0028745097,0.9487996,0.0009306189,0.0018212232,0.03452974],"study_design_scores_gemma":[0.0003921534,0.0009338613,0.31778997,0.00038155864,0.0011301995,0.002226923,0.00121836,0.06415431,0.4795122,0.002539158,0.12940857,0.00031265107],"about_ca_topic_score_codex":0.006418036,"about_ca_topic_score_gemma":0.009644217,"teacher_disagreement_score":0.006418036,"about_ca_system_score_codex":0.0004328856,"about_ca_system_score_gemma":0.0007761838,"threshold_uncertainty_score":0.012761414},"labels":[],"label_agreement":null},{"id":"W4384025130","doi":"10.1038/s41597-023-02330-9","title":"Magnetic resonance imaging datasets with anatomical fiducials for quality control and registration","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced MRI Techniques and Applications","field":"Medicine","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Robarts Clinical Trials; University of British Columbia; McMaster University; Western University","funders":"Canada Foundation for Innovation; National Center for Research Resources; National Institute of Mental Health; NIH Blueprint for Neuroscience Research; National Institute on Aging; McDonnell Center for Systems Neuroscience; National Institutes of Health; Canada Research Chairs; Canada First Research Excellence Fund; Mitacs; Natural Sciences and Engineering Research Council of Canada; Fondation Brain Canada","keywords":"Fiducial marker; Neuroimaging; Computer science; Magnetic resonance imaging; Artificial intelligence; Image registration; Neuroanatomy; Computer vision; Medical physics; Medicine; Radiology; Image (mathematics); Anatomy","score_opus":0.06958832490861831,"score_gpt":0.4077585015990662,"score_spread":0.33817017669044785,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4384025130","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.038064677,0.0017611849,0.8348916,0.00094953465,0.00064919825,0.002822176,0.07121357,0.040910635,0.008737401],"genre_scores_gemma":[0.08535782,0.00082594063,0.7543704,0.0006443514,0.00019463473,0.008927443,0.1291082,0.016187109,0.004384001],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98989534,0.002072947,0.0026946238,0.0018404193,0.0032299578,0.00026675375],"domain_scores_gemma":[0.9610513,0.006442207,0.003371722,0.020324023,0.008407223,0.0004036487],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012436271,0.0015467817,0.0013940007,0.0035194315,0.001533006,0.0028303794,0.0031142104,0.0019282061,0.012186582],"category_scores_gemma":[0.053314965,0.0012842501,0.0014594801,0.0041707098,0.001437745,0.0023044958,0.0041377526,0.002233238,0.01441662],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023000005,0.0004892141,0.014623221,0.0046909247,0.0009067693,0.000879423,0.0031356586,0.013868582,0.113438286,0.014394085,0.29499212,0.53628176],"study_design_scores_gemma":[0.00074340665,0.0009389291,0.06105837,0.0011908927,0.0006776487,0.0053463806,0.00082308176,0.0421645,0.17325996,0.026390953,0.68660945,0.0007964826],"about_ca_topic_score_codex":0.003126012,"about_ca_topic_score_gemma":0.006528125,"teacher_disagreement_score":0.012436271,"about_ca_system_score_codex":0.0008963072,"about_ca_system_score_gemma":0.0031125306,"threshold_uncertainty_score":0.06577009},"labels":[],"label_agreement":null},{"id":"W4385343011","doi":"10.1038/s41597-023-02413-7","title":"Author Correction: A Database of Snow on Sea Ice in the Central Arctic Collected during the MOSAiC expedition","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Arctic and Antarctic ice dynamics","field":"Earth and Planetary Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Mosaic; Snow; Arctic; Sea ice; Oceanography; Physical geography; The arctic; Climatology; Geology; Geography; Meteorology; Archaeology","score_opus":0.035128624466839044,"score_gpt":0.2545644915385897,"score_spread":0.21943586707175064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385343011","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021168338,0.0011795029,0.005672323,0.02526972,0.8872669,0.00010339105,0.046170846,0.0035599496,0.028660543],"genre_scores_gemma":[0.039416615,0.0059698345,0.04056684,0.024034185,0.08471009,0.0004258759,0.11729384,0.019356562,0.66822624],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9969777,0.00027356748,0.00050342246,0.00042373018,0.0016377193,0.0001838671],"domain_scores_gemma":[0.97673756,0.0029349232,0.0009839381,0.001991001,0.016801024,0.00055153936],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022695237,0.0012982999,0.0010782393,0.0049368236,0.0027524685,0.0035053038,0.0017885333,0.0017483291,0.117431164],"category_scores_gemma":[0.03113942,0.00085508474,0.0010686533,0.005129683,0.0010251851,0.0017959534,0.0019802917,0.0043466617,0.05881247],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019208604,0.0000040312566,0.00016099848,0.00008552995,0.0000067432593,0.000096134754,0.000035182897,0.000042251177,0.00007740426,0.0004669891,0.992078,0.0069275773],"study_design_scores_gemma":[0.000015693218,0.0000064119245,0.001313576,0.00018462416,0.000022136092,0.0002952388,0.00010030875,0.00012076949,0.00041116544,0.0005313278,0.9969764,0.00002245588],"about_ca_topic_score_codex":0.040650655,"about_ca_topic_score_gemma":0.053031754,"teacher_disagreement_score":0.117431164,"about_ca_system_score_codex":0.0023763324,"about_ca_system_score_gemma":0.0056060744,"threshold_uncertainty_score":0.3928464},"labels":[],"label_agreement":null},{"id":"W4386026821","doi":"10.1038/s41597-023-02377-8","title":"A longitudinal resource for population neuroscience of school-age children and adolescents in China","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University Health Centre; Université de Montréal; Western University; Hospital for Sick Children; University of Toronto; Montreal Neurological Institute and Hospital; Douglas Mental Health University Institute; McGill University; Centre Hospitalier Universitaire Sainte-Justine; Queen's University; Holland Bloorview Kids Rehabilitation Hospital","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Social Science Fund of China; National Key Research and Development Program of China; Beijing Municipal Science and Technology Commission; Medical Research Council; Chinese Academy of Sciences; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Beijing Normal University; National Natural Science Foundation of China; Child Mind Institute","keywords":"Population; Psychology; Longitudinal study; Resource (disambiguation); Cognition; Computer science; Neuroscience; Demography; Medicine; Sociology","score_opus":0.09694316819790338,"score_gpt":0.32209723093467496,"score_spread":0.22515406273677158,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386026821","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033445418,0.0005314065,0.0048401244,0.00039026284,0.000086727305,0.00073109666,0.95388234,0.0007750604,0.005317505],"genre_scores_gemma":[0.050387047,0.0005233493,0.009255848,0.00030962564,0.00006378393,0.0041241883,0.9296348,0.00020126447,0.0055000456],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99827707,0.00027580818,0.00037324714,0.0004358974,0.00041689703,0.00022113738],"domain_scores_gemma":[0.992796,0.0005970854,0.00068444136,0.0017843699,0.0030660643,0.0010719557],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027963982,0.00078876707,0.0009391315,0.0055780415,0.0012125219,0.0010434138,0.0024129224,0.0005530767,0.018436683],"category_scores_gemma":[0.009056248,0.00061605143,0.0005846232,0.008759532,0.0002990989,0.00092444546,0.002504503,0.0006532108,0.005273451],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005024367,0.00020000136,0.37050095,0.0016766589,0.00041632258,0.00077766465,0.0018396941,0.00089539477,0.0025073534,0.0043420345,0.5043434,0.11199819],"study_design_scores_gemma":[0.00018595015,0.00007202967,0.61928,0.0004856752,0.00023733763,0.00030021838,0.00087970257,0.0010539028,0.001245957,0.0017100536,0.37444898,0.00010022565],"about_ca_topic_score_codex":0.18450621,"about_ca_topic_score_gemma":0.15231785,"teacher_disagreement_score":0.18450621,"about_ca_system_score_codex":0.002544321,"about_ca_system_score_gemma":0.014188638,"threshold_uncertainty_score":0.3668645},"labels":[],"label_agreement":null},{"id":"W4386070053","doi":"10.1038/s41597-023-02439-x","title":"Spatial and Temporal Availability of Cloud-free Optical Observations in the Tropics to Monitor Deforestation","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; Ste. Anne's Hospital; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Marshall Space Flight Center; National Aeronautics and Space Administration","keywords":"Tropics; Cloud computing; Deforestation (computer science); Environmental science; Remote sensing; Satellite imagery; Metadata; Computer science; Database; Geography; Meteorology; Ecology","score_opus":0.05593065947997162,"score_gpt":0.2739383081052171,"score_spread":0.21800764862524546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386070053","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89054114,0.00039102646,0.0030885062,0.00031356775,0.000032118383,0.00005954029,0.10030585,0.00028412562,0.0049841423],"genre_scores_gemma":[0.8992834,0.0002633157,0.009949294,0.0000960544,0.000031765314,0.00014245218,0.08954446,0.00007482372,0.00061445305],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997198,0.000045052984,0.000029114592,0.000085094514,0.00006356689,0.000057431847],"domain_scores_gemma":[0.99916434,0.0001562323,0.00024810573,0.00010984414,0.0002514684,0.00007000834],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007379719,0.0002777647,0.00018650776,0.0016022986,0.00022832307,0.00047310398,0.00031876264,0.0002970984,0.0014505376],"category_scores_gemma":[0.0012843657,0.00014635912,0.00032996235,0.0028072349,0.00023038367,0.00060083356,0.00040137788,0.0003384334,0.00030017045],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038851524,0.000276542,0.88196176,0.0004625028,0.00039341598,0.00046836314,0.00082184933,0.035761885,0.020031244,0.0014976811,0.014448859,0.043487407],"study_design_scores_gemma":[0.000035836507,0.000033094835,0.9576255,0.00007390786,0.00007482872,0.00015575072,0.0005077571,0.022266986,0.0033954135,0.00035550416,0.015447178,0.000028167316],"about_ca_topic_score_codex":0.055391274,"about_ca_topic_score_gemma":0.06910684,"teacher_disagreement_score":0.055391274,"about_ca_system_score_codex":0.00057585357,"about_ca_system_score_gemma":0.0006090652,"threshold_uncertainty_score":0.11013776},"labels":[],"label_agreement":null},{"id":"W4386127240","doi":"10.1038/s41597-023-02478-4","title":"Greenhouse gas emissions from U.S. crude oil pipeline accidents: 1968 to 2020","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric and Environmental Gas Dynamics","field":"Environmental Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Natural Science Foundation of Jiangsu Province; Government of Jiangsu Province","keywords":"Greenhouse gas; Environmental science; Pipeline (software); Crude oil; Petroleum engineering; Engineering; Geology; Oceanography","score_opus":0.02372505400748101,"score_gpt":0.2601706665585461,"score_spread":0.23644561255106508,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386127240","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9706956,0.00043768677,0.001960957,0.00038894077,0.000036523274,0.00001617025,0.021774799,0.00012537939,0.0045638294],"genre_scores_gemma":[0.99076277,0.00033384227,0.00072167657,0.000042501288,0.0000088960605,0.000008207567,0.0074021528,0.0000071505247,0.0007128747],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998796,0.0000129678565,0.000008874603,0.00002967138,0.000044485685,0.00002449667],"domain_scores_gemma":[0.9996265,0.000050766543,0.0001185997,0.000021838623,0.00015943802,0.00002279011],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00024409384,0.00039880894,0.000118567754,0.000904358,0.0002364715,0.00041960948,0.00017240184,0.00025500543,0.0009358306],"category_scores_gemma":[0.0006750057,0.00016344819,0.00046232628,0.0011832139,0.00016974175,0.0005278146,0.00034632048,0.0003499722,0.0002535704],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025946685,0.00010075327,0.80354345,0.00013274165,0.00020498718,0.0003585023,0.00012963456,0.1630617,0.0025573145,0.0014667325,0.0066363555,0.021548383],"study_design_scores_gemma":[0.000014520108,0.000118347314,0.8745915,0.00011854331,0.00015145921,0.00013932891,0.00053631596,0.102790944,0.005631019,0.0013875377,0.014465508,0.00005498342],"about_ca_topic_score_codex":0.11208644,"about_ca_topic_score_gemma":0.11213913,"teacher_disagreement_score":0.11208644,"about_ca_system_score_codex":0.0017382328,"about_ca_system_score_gemma":0.00081950525,"threshold_uncertainty_score":0.22286808},"labels":[],"label_agreement":null},{"id":"W4386195873","doi":"10.1038/s41597-023-02464-w","title":"Digitization of the Australian Parliamentary Debates, 1998–2022","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Parliament; Digitization; Democracy; Politics; Resource (disambiguation); Metadata; Public administration; Political science; Library science; Data science; Sociology; Computer science; World Wide Web; Law","score_opus":0.1295125417729573,"score_gpt":0.39690934396953004,"score_spread":0.26739680219657275,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386195873","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13914227,0.006252722,0.0034191469,0.005057581,0.0010256732,0.0008466785,0.6130989,0.0009899165,0.23016709],"genre_scores_gemma":[0.25849435,0.007036698,0.017876592,0.0017692206,0.0005312529,0.002922929,0.48767784,0.0006723448,0.22301874],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967397,0.00038166647,0.0003643459,0.00034291376,0.001901674,0.00026967537],"domain_scores_gemma":[0.98875797,0.0017669161,0.0018127331,0.0017610359,0.005128806,0.0007725452],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002913886,0.0004318311,0.00043800342,0.021345386,0.0023533371,0.0028293987,0.0011614586,0.0007275153,0.019560339],"category_scores_gemma":[0.015199766,0.00078667206,0.00036106457,0.037678182,0.0007918038,0.0016651136,0.0029780057,0.0013380252,0.005737589],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044884533,0.00012782324,0.022973897,0.0035554862,0.00008519444,0.0007448245,0.05058613,0.0012377822,0.0040078815,0.022360649,0.59808743,0.29578397],"study_design_scores_gemma":[0.000008484227,0.000012830595,0.17635727,0.00026795824,0.000017232998,0.00013041396,0.0020187623,0.00025814085,0.00072994235,0.0006183619,0.8195564,0.000024227978],"about_ca_topic_score_codex":0.36132184,"about_ca_topic_score_gemma":0.505362,"teacher_disagreement_score":0.36132184,"about_ca_system_score_codex":0.007411446,"about_ca_system_score_gemma":0.010751333,"threshold_uncertainty_score":0.71843743},"labels":[],"label_agreement":null},{"id":"W4386416992","doi":"10.1038/s41597-023-02467-7","title":"Plant traits and associated data from a warming experiment, a seabird colony, and along elevation in Svalbard","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Senter for Internasjonalisering av Utdanning; Universitetet i Bergen","keywords":"Ecosystem; Microclimate; Seabird; Ecology; Climate change; Arctic; Vegetation (pathology); Trait; Environmental science; Environmental change; Global warming; Physical geography; Geography; Biology","score_opus":0.1270747668648577,"score_gpt":0.3179991794282997,"score_spread":0.190924412563442,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386416992","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95466024,0.0001259338,0.00025294538,0.000022530052,0.0000086067175,0.000021260876,0.04419312,0.000052821786,0.0006624698],"genre_scores_gemma":[0.8107468,0.00014645974,0.00255903,0.00006086532,0.000011813512,0.00012351839,0.1850108,0.00003535454,0.0013054119],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997167,0.000042333482,0.000022723516,0.00010973268,0.00006668,0.00004171176],"domain_scores_gemma":[0.9990952,0.00017440302,0.00017650361,0.00023405977,0.00018215925,0.00013775722],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004329681,0.0002141169,0.00039530848,0.0012315518,0.0005301971,0.00046981906,0.0002902411,0.00028503517,0.0011319967],"category_scores_gemma":[0.0005462577,0.00013986422,0.00026589798,0.0012782996,0.00036245235,0.00015025762,0.00041154886,0.000284175,0.0003597759],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00063694996,0.00042650616,0.94721556,0.00016392469,0.00037168045,0.00028789442,0.00061163143,0.0021883217,0.018945282,0.00022793312,0.009572368,0.019351933],"study_design_scores_gemma":[0.000010468533,0.000025724667,0.99555415,0.0000068094846,0.00002533872,0.000044350767,0.0001235089,0.0003325483,0.00058478175,0.000019253,0.0032667015,0.000006434056],"about_ca_topic_score_codex":0.11916704,"about_ca_topic_score_gemma":0.29311693,"teacher_disagreement_score":0.11916704,"about_ca_system_score_codex":0.0005945451,"about_ca_system_score_gemma":0.0005363464,"threshold_uncertainty_score":0.23694682},"labels":[],"label_agreement":null},{"id":"W4386573853","doi":"10.1038/s41597-023-02488-2","title":"One high quality genome and two transcriptome datasets for new species of Mantamonas, a deep-branching eukaryote clade","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Protist diversity and phylogeny","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"European Commission; Government of Canada; Simons Foundation","keywords":"Clade; Eukaryote; Biology; Evolutionary biology; Transcriptome; Genome; Branching (polymer chemistry); Computational biology; Phylogenetics; Gene; Genetics; Gene expression","score_opus":0.08410650693930481,"score_gpt":0.3129157311745487,"score_spread":0.22880922423524386,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386573853","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5258786,0.003668088,0.04306684,0.0012384568,0.0005483619,0.0004296721,0.4136675,0.0028844194,0.008618104],"genre_scores_gemma":[0.17503053,0.001182518,0.064711675,0.00032841816,0.00007431114,0.00033936813,0.7551768,0.0004496442,0.0027067407],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993231,0.000049613514,0.00007717354,0.00028862947,0.00018458723,0.00007687721],"domain_scores_gemma":[0.99888676,0.00023257235,0.0002225246,0.00017334003,0.00030788596,0.0001768455],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00057426677,0.00086260115,0.00067079946,0.0016348906,0.0011252236,0.0009612802,0.00048659946,0.0009840718,0.003798932],"category_scores_gemma":[0.0016169407,0.0004036939,0.0010844263,0.0024680747,0.00034265162,0.00082809094,0.0011028097,0.0014025852,0.0021639008],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062946806,0.00018635418,0.018655919,0.0019053319,0.00024491546,0.0008976149,0.00066688476,0.0014633506,0.93107253,0.00116635,0.006206131,0.03690507],"study_design_scores_gemma":[0.00030404492,0.00095021515,0.3993722,0.0006247593,0.0010168746,0.004544755,0.0016152748,0.011466702,0.17355357,0.004017336,0.40225998,0.0002744086],"about_ca_topic_score_codex":0.0019011608,"about_ca_topic_score_gemma":0.0082210535,"teacher_disagreement_score":0.003798932,"about_ca_system_score_codex":0.0005159414,"about_ca_system_score_gemma":0.0015800514,"threshold_uncertainty_score":0.012708664},"labels":[],"label_agreement":null},{"id":"W4386740911","doi":"10.1038/s41597-023-02501-8","title":"Shared metadata for data-centric materials science","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"Horizon 2020 Framework Programme; Humboldt-Universität zu Berlin; Engineering and Physical Sciences Research Council; Deutsche Forschungsgemeinschaft; European Commission","keywords":"Metadata; Computer science; Interoperability; Workflow; Data sharing; Data science; Data curation; Big data; World Wide Web; Database; Data mining","score_opus":0.14759973735751378,"score_gpt":0.3791868367131539,"score_spread":0.23158709935564012,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386740911","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003483273,0.0031566648,0.9561078,0.018563258,0.0012753756,0.0003631939,0.001729651,0.00201514,0.013305717],"genre_scores_gemma":[0.08532074,0.004434502,0.89107305,0.00366239,0.001167737,0.00086499075,0.0069715963,0.0009612763,0.0055437754],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.97646344,0.010676704,0.0035827293,0.002295849,0.0061240154,0.0008572521],"domain_scores_gemma":[0.9162734,0.019103387,0.0028730945,0.0490251,0.009513633,0.0032112892],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.051076174,0.0008497479,0.0015334082,0.00729623,0.0042749126,0.016428277,0.0059803757,0.0038470163,0.005864365],"category_scores_gemma":[0.06975617,0.0011156767,0.002261653,0.009096772,0.009623608,0.04205313,0.01884763,0.0067962958,0.0025698377],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000031574353,0.000037156307,0.00051259616,0.00026253698,0.000032084794,0.00006767206,0.00045094037,0.0023739478,0.000600915,0.96106493,0.0081567615,0.026408862],"study_design_scores_gemma":[0.000018382803,0.000026613443,0.00018862392,0.0004988386,0.000030478377,0.000117408454,0.00043687396,0.00850444,0.0020017778,0.825652,0.16247562,0.0000489688],"about_ca_topic_score_codex":0.003681009,"about_ca_topic_score_gemma":0.0036449044,"teacher_disagreement_score":0.9940196,"about_ca_system_score_codex":0.006219385,"about_ca_system_score_gemma":0.014448681,"threshold_uncertainty_score":0.27011985},"labels":[],"label_agreement":null},{"id":"W4386798604","doi":"10.1038/s41597-023-02502-7","title":"Whole genomes from bacteria collected at diagnostic units around the world 2020","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre Hospitalier Universitaire de Sherbrooke","funders":"Horizon 2020 Framework Programme; Novo Nordisk Fonden; European Commission","keywords":"Genome; Bacteria; Biology; Computational biology; Evolutionary biology; Genetics; Gene","score_opus":0.04050191828275441,"score_gpt":0.25942383480733133,"score_spread":0.21892191652457693,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386798604","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.037245966,0.0023921304,0.0011303532,0.00031379377,0.000057600944,0.00007735583,0.9563438,0.0002980782,0.002140927],"genre_scores_gemma":[0.01595177,0.00086884864,0.0031104088,0.00012100705,0.00001479988,0.00007146644,0.9794901,0.0000288117,0.00034280276],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9989084,0.00012849717,0.00013750297,0.00028211274,0.0003540255,0.000189429],"domain_scores_gemma":[0.9978143,0.00036182522,0.0004973679,0.00027545908,0.0005771857,0.00047387579],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013861451,0.0009834433,0.0008747431,0.0059983446,0.0007858164,0.001275311,0.00057279604,0.0010114479,0.0041734297],"category_scores_gemma":[0.0038418577,0.00042812948,0.0009354647,0.009467348,0.00037124188,0.0007396297,0.0017319369,0.0008384254,0.003571822],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005619759,0.00072554365,0.25755668,0.012972023,0.0014939734,0.0016721651,0.0017011807,0.0059633427,0.1206582,0.0041643055,0.397037,0.19043583],"study_design_scores_gemma":[0.00022457016,0.0003580082,0.53551906,0.0011616581,0.00045701518,0.0011627637,0.0014320771,0.001294402,0.01368017,0.0018625334,0.4427057,0.00014208615],"about_ca_topic_score_codex":0.017358089,"about_ca_topic_score_gemma":0.025262095,"teacher_disagreement_score":0.017358089,"about_ca_system_score_codex":0.0009369445,"about_ca_system_score_gemma":0.0027773476,"threshold_uncertainty_score":0.03451413},"labels":[],"label_agreement":null},{"id":"W4386829292","doi":"10.1038/s41597-023-02535-y","title":"A daily high-resolution (1 km) human thermal index collection over the North China Plain from 2003 to 2020","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Urban Heat Island Mitigation","field":"Environmental Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; University of Toronto","funders":"National Natural Science Foundation of China","keywords":"China; Index (typography); Environmental science; Coastal plain; Geography; Remote sensing; Biology; Ecology; Archaeology; World Wide Web; Computer science","score_opus":0.02063176456647712,"score_gpt":0.2386361082318871,"score_spread":0.21800434366541,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386829292","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.34299442,0.0004621531,0.0053661265,0.00037085215,0.00015566662,0.00017358497,0.6402734,0.0015695571,0.008634183],"genre_scores_gemma":[0.30396575,0.00024287005,0.008198683,0.00012231365,0.000059243663,0.00030553245,0.68423307,0.000086966866,0.0027855497],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99973327,0.00003170019,0.000023163,0.00009431717,0.000073719,0.000043791086],"domain_scores_gemma":[0.9994842,0.000032925804,0.0000628718,0.000099204924,0.0002604155,0.00006023427],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00031192423,0.000615892,0.00034752832,0.0012082723,0.00039606722,0.00041429873,0.0006610309,0.0004333445,0.0023460903],"category_scores_gemma":[0.0006416393,0.00017175535,0.00043071495,0.0027283202,0.0002576635,0.00039020917,0.00049755105,0.00040532488,0.0013548351],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044009738,0.00048154657,0.46271855,0.00084899866,0.00044080222,0.0006906864,0.0008942226,0.03865127,0.010680668,0.001238773,0.36365715,0.11925724],"study_design_scores_gemma":[0.000059784717,0.0000424787,0.86698055,0.00010738787,0.0000562135,0.00016241409,0.0006901697,0.030171217,0.0036849214,0.0005009528,0.097460225,0.00008374326],"about_ca_topic_score_codex":0.10622824,"about_ca_topic_score_gemma":0.1891018,"teacher_disagreement_score":0.10622824,"about_ca_system_score_codex":0.00082095386,"about_ca_system_score_gemma":0.0011833683,"threshold_uncertainty_score":0.21121985},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4387014311","doi":"10.1038/s41597-023-02513-4","title":"Integrative GWAS and co-localisation analysis suggests novel genes associated with age-related multimorbidity","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Chronic Disease Management Strategies","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Discovery Centre","funders":"Medical Research Council; Wellcome Trust","keywords":"Genome-wide association study; Computational biology; Gene; Biology; Genetics; Evolutionary biology; Single-nucleotide polymorphism; Genotype","score_opus":0.13345698917719945,"score_gpt":0.3774755081471648,"score_spread":0.24401851896996535,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387014311","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9456268,0.004429103,0.027172418,0.0017848655,0.00012000831,0.00008240158,0.013826726,0.0013015387,0.0056561152],"genre_scores_gemma":[0.9743798,0.0012145382,0.01627221,0.00049551553,0.00009644881,0.000053549622,0.005941414,0.00013970646,0.0014069154],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99893457,0.0002537848,0.00008734449,0.0004372782,0.00018235207,0.000104788305],"domain_scores_gemma":[0.99665785,0.001661245,0.00075759954,0.00042221448,0.000165387,0.0003356809],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015641422,0.0005631402,0.00088434044,0.002661195,0.00051563047,0.0010148939,0.0004612463,0.0005771222,0.010604015],"category_scores_gemma":[0.0035580464,0.00026458045,0.0019225403,0.0040381597,0.00032370116,0.00029346542,0.0011307718,0.0005815277,0.00097366946],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014427471,0.00015258229,0.8691417,0.00053206953,0.0027455387,0.0014146124,0.0004040236,0.0012670072,0.034119938,0.0015362005,0.0045848745,0.08265874],"study_design_scores_gemma":[0.00010585258,0.00019297302,0.9805706,0.00006906586,0.0016971143,0.0014938622,0.00017881636,0.0035361995,0.00275945,0.0034186956,0.0059362263,0.000041161315],"about_ca_topic_score_codex":0.003903245,"about_ca_topic_score_gemma":0.008011104,"teacher_disagreement_score":0.010604015,"about_ca_system_score_codex":0.00022691849,"about_ca_system_score_gemma":0.00063145615,"threshold_uncertainty_score":0.035474002},"labels":[],"label_agreement":null},{"id":"W4387188565","doi":"10.1038/s41597-023-02579-0","title":"ecolo-zip: A global, rich and granular characterization of biogeophysical ecology for 1.5 million postal codes","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Land Use and Ecosystem Services","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Karl Landsteiner Privatuniversität für Gesundheitswissenschaften","keywords":"Geospatial analysis; Shuttle Radar Topography Mission; Granularity; Geography; Vegetation (pathology); Scale (ratio); Resource (disambiguation); Remote sensing; Climate change; Environmental resource management; Digital elevation model; Ecology; Physical geography; Data science; Computer science; Environmental science; Cartography; Biology","score_opus":0.021260797631806445,"score_gpt":0.2548493101242048,"score_spread":0.23358851249239837,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387188565","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10738388,0.00023070519,0.00796215,0.00058321253,0.00009704999,0.00012837788,0.873838,0.0023608531,0.0074159075],"genre_scores_gemma":[0.16964932,0.00023438907,0.02058537,0.00017968183,0.000065861255,0.0005143841,0.8061876,0.00048571988,0.0020975485],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9994816,0.000083955696,0.000071643386,0.00012633424,0.00014730595,0.00008929132],"domain_scores_gemma":[0.99750936,0.000550838,0.0006161582,0.00056373887,0.0005116297,0.00024821173],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004820796,0.0005007245,0.00036817454,0.0037083747,0.00043941592,0.0010234675,0.00068963063,0.0005819621,0.0056243925],"category_scores_gemma":[0.0055460758,0.00019721266,0.0004898146,0.005661425,0.00039461156,0.0012693136,0.002107668,0.00059263315,0.0036312446],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006260643,0.00025072455,0.3708952,0.0014609925,0.00020602482,0.00078550394,0.0020512426,0.012667807,0.0062979837,0.01637061,0.4913703,0.09701747],"study_design_scores_gemma":[0.0001360209,0.00010185096,0.37908554,0.00033348767,0.000049330876,0.00064284366,0.0025529487,0.021792635,0.003942605,0.011787471,0.5794501,0.00012513959],"about_ca_topic_score_codex":0.0143902935,"about_ca_topic_score_gemma":0.024146697,"teacher_disagreement_score":0.0143902935,"about_ca_system_score_codex":0.00054439093,"about_ca_system_score_gemma":0.00083672797,"threshold_uncertainty_score":0.02861303},"labels":[],"label_agreement":null},{"id":"W4387389354","doi":"10.1038/s41597-023-02588-z","title":"Transcriptome profiling of Fraxinus excelsior genotypes infested by emerald ash borer","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Forest Insect Ecology and Management","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada; Canadian Forest Service","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Fraxinus; Biology; Emerald ash borer; Transcriptome; Botany; Gene; Gene expression","score_opus":0.029734239144336885,"score_gpt":0.26384104144540427,"score_spread":0.2341068023010674,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387389354","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9944543,0.00055418455,0.00093446963,0.000041777603,0.000017734345,0.000010904524,0.0033266013,0.000030876163,0.00062901323],"genre_scores_gemma":[0.98310274,0.0007350444,0.002008696,0.000188786,0.000018525581,0.000055885568,0.010443453,0.000039093928,0.003407852],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9998685,0.000009159164,0.0000074594254,0.00005976992,0.000026517027,0.000028566745],"domain_scores_gemma":[0.9999056,0.000020649646,0.000027946538,0.00000800026,0.000019625413,0.000018211073],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00010917525,0.0001862936,0.0003323862,0.0003190329,0.00025711625,0.0004103874,0.00009039988,0.00021793162,0.0004948784],"category_scores_gemma":[0.000105146944,0.00011004738,0.00027443402,0.00038937866,0.00014782691,0.00019590282,0.00020917569,0.00037787773,0.00021990598],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000096574346,0.00001807504,0.00240763,0.000034850487,0.000011015645,0.000041759125,0.00008662071,0.000056389315,0.99607664,0.00002358347,0.00005083106,0.0010959491],"study_design_scores_gemma":[0.000018258212,0.00052846194,0.7992677,0.00004732429,0.00013384433,0.00046728918,0.0011130074,0.0023126858,0.19028579,0.0002087964,0.0055801356,0.00003671671],"about_ca_topic_score_codex":0.00087660464,"about_ca_topic_score_gemma":0.0015420369,"teacher_disagreement_score":0.00087660464,"about_ca_system_score_codex":0.000145873,"about_ca_system_score_gemma":0.00012234935,"threshold_uncertainty_score":0.0017430186},"labels":[],"label_agreement":null},{"id":"W4387502842","doi":"10.1038/s41597-023-02584-3","title":"The 10-m cotton maps in Xinjiang, China during 2018–2021","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Youth Innovation Promotion Association; Youth Innovation Promotion Association of the Chinese Academy of Sciences; Xinjiang Production and Construction Corps; National Natural Science Foundation of China","keywords":"China; Phenology; Remote sensing; Cartography; Geography; Mathematics; Forestry; Agronomy; Biology","score_opus":0.017986716668803852,"score_gpt":0.23649426800612575,"score_spread":0.2185075513373219,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387502842","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9976453,0.000059881517,0.00015313696,0.000029288198,0.0000062363165,0.00000808096,0.001503181,0.000022002389,0.0005728762],"genre_scores_gemma":[0.99726033,0.000060074464,0.00030863096,0.000006292169,0.0000043831096,0.000009504725,0.0018949813,0.0000019748163,0.00045378754],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9999205,0.000009325861,0.000006039136,0.000024410663,0.000018520723,0.000021180396],"domain_scores_gemma":[0.9998399,0.00002119451,0.000046541052,0.000011762147,0.00005108484,0.00002948235],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000290421,0.00037712813,0.00011836441,0.0011008836,0.00025688054,0.00027913027,0.00019743525,0.0001756504,0.00054534787],"category_scores_gemma":[0.00023834528,0.00011285214,0.0001383439,0.0010542605,0.00015148574,0.0002338108,0.00023685556,0.0000983963,0.00010178452],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005167184,0.00015882711,0.9032178,0.00016513265,0.00009396313,0.0014152933,0.0011001523,0.01339994,0.02011235,0.0003251278,0.0030091237,0.056485485],"study_design_scores_gemma":[0.00000383992,0.00001852003,0.9934208,0.0000054587035,0.000009379124,0.000039650266,0.00022540748,0.0043887775,0.000591594,0.000019231356,0.0012729542,0.0000044854564],"about_ca_topic_score_codex":0.048549816,"about_ca_topic_score_gemma":0.087905586,"teacher_disagreement_score":0.048549816,"about_ca_system_score_codex":0.00060176593,"about_ca_system_score_gemma":0.00031874803,"threshold_uncertainty_score":0.09653449},"labels":[],"label_agreement":null},{"id":"W4387642549","doi":"10.1038/s41597-023-02594-1","title":"A high-quality chromosome-level wild rice genome of Oryza coarctata","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Biotechnology Research Institute","funders":"Ministry of Science and Technology of the People's Republic of China","keywords":"Oryza; Biology; Oryza sativa; Genome; Chromosome; Botany; Genetics; Gene","score_opus":0.10197041917225738,"score_gpt":0.31836293382451614,"score_spread":0.21639251465225876,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387642549","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92377913,0.0016220148,0.035140943,0.00023883938,0.00012423693,0.00013345008,0.03479259,0.0014803612,0.0026883963],"genre_scores_gemma":[0.6953708,0.0017101207,0.109353796,0.00014430183,0.000045198085,0.00019673337,0.1832059,0.0009627662,0.009010513],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998529,0.000008694491,0.000012170539,0.00007002118,0.0000329144,0.000023359224],"domain_scores_gemma":[0.9998,0.000028666253,0.000042079755,0.00003854759,0.00005507609,0.000035675286],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022184536,0.00063332333,0.0005536363,0.0005976264,0.00051269657,0.00042132763,0.00031113715,0.00021627892,0.0015944738],"category_scores_gemma":[0.00032301215,0.00026662298,0.0005787776,0.0010579505,0.00017178975,0.00031440318,0.00044387064,0.0007221496,0.0010701917],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005100567,0.000064717555,0.0059411256,0.0003340672,0.000044682052,0.00034412678,0.00030209022,0.0006226133,0.9729462,0.00019057389,0.00060404424,0.0180957],"study_design_scores_gemma":[0.00014114127,0.00059416005,0.41927978,0.00015184068,0.00069147756,0.002170315,0.00064683025,0.00845117,0.4951715,0.0005469651,0.07203754,0.00011729199],"about_ca_topic_score_codex":0.0055757887,"about_ca_topic_score_gemma":0.008174669,"teacher_disagreement_score":0.0055757887,"about_ca_system_score_codex":0.00042416406,"about_ca_system_score_gemma":0.0009172016,"threshold_uncertainty_score":0.011086643},"labels":[],"label_agreement":null},{"id":"W4387692943","doi":"10.1038/s41597-023-02610-4","title":"Fully resolved assembly of Fusarium proliferatum DSM106835 genome","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Pathogens and Fungal Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Agriculture and Agri-Food Canada","funders":"United Arab Emirates University","keywords":"Fusarium proliferatum; Sequence assembly; Genome; Biology; Contig; Fusarium; Genetics; Gene; Reference genome; Genome size; Computational biology; Transcriptome","score_opus":0.04040575584233469,"score_gpt":0.2769645422456526,"score_spread":0.2365587864033179,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387692943","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.878918,0.003157361,0.059855677,0.00043315516,0.000278833,0.00030515768,0.04722067,0.0017704613,0.008060694],"genre_scores_gemma":[0.66991293,0.0021734878,0.13964014,0.000178411,0.000045230845,0.00030520474,0.18033755,0.00082157314,0.0065855198],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9997423,0.000035934092,0.000023688257,0.00009299197,0.00006879198,0.000036235066],"domain_scores_gemma":[0.9997826,0.00003807593,0.000037205256,0.000025946718,0.000081989965,0.000034180885],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045514884,0.00075171207,0.0005426047,0.0010187577,0.00068314647,0.000714702,0.00039491412,0.0005520646,0.0016155067],"category_scores_gemma":[0.0007225465,0.00028895115,0.00078388647,0.0013278503,0.00015382971,0.00042851715,0.000495251,0.00070604787,0.000952708],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00055530696,0.000066742956,0.0037952743,0.0005795895,0.000057557314,0.0007374204,0.00054127804,0.0021115995,0.96183044,0.0006486008,0.0018232098,0.027253086],"study_design_scores_gemma":[0.00017782643,0.0011278893,0.17514192,0.000505159,0.00059095025,0.0025188436,0.002059209,0.038402002,0.62430996,0.0021234592,0.15281665,0.00022627927],"about_ca_topic_score_codex":0.0050620367,"about_ca_topic_score_gemma":0.00808887,"teacher_disagreement_score":0.0050620367,"about_ca_system_score_codex":0.00052389654,"about_ca_system_score_gemma":0.00076157216,"threshold_uncertainty_score":0.010065138},"labels":[],"label_agreement":null},{"id":"W4387844510","doi":"10.1038/s41597-023-02638-6","title":"Non-pharmaceutical interventions to combat COVID-19 in the Americas described through daily sub-national data","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Inter-American Development Bank","keywords":"Coronavirus disease 2019 (COVID-19); 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Psychological intervention; Pandemic; Geography; Medicine; Virology; Outbreak; Nursing; Internal medicine; Infectious disease (medical specialty)","score_opus":0.8428128399512812,"score_gpt":0.6118075745123546,"score_spread":0.23100526543892663,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387844510","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006499064,0.0002721887,0.0003570048,0.0003038252,0.000039355018,0.00009943653,0.9887495,0.000064489584,0.003615131],"genre_scores_gemma":[0.015948202,0.0003056638,0.001460991,0.00023466718,0.000041555784,0.00060252374,0.9794194,0.000032930184,0.001954144],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9981013,0.00044263378,0.00033543448,0.00037104523,0.000511028,0.00023857498],"domain_scores_gemma":[0.992811,0.0018376097,0.0023906077,0.0008178061,0.0016293062,0.00051363115],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017966733,0.00066179835,0.000683689,0.0031786344,0.00048815625,0.0012330584,0.0014174565,0.0010319538,0.011703462],"category_scores_gemma":[0.009375847,0.0003669343,0.0007780307,0.007466026,0.0002757727,0.0005406257,0.001396276,0.0012974914,0.0045236666],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041407277,0.00033825912,0.124982916,0.0025223554,0.00040169497,0.00017684365,0.0005139888,0.0026700178,0.0006027338,0.005072854,0.8354381,0.026866125],"study_design_scores_gemma":[0.00026270718,0.00011109885,0.20106314,0.0011454897,0.000121973426,0.00014102232,0.00079929293,0.0014288265,0.00071326847,0.001598746,0.79255885,0.000055520224],"about_ca_topic_score_codex":0.046980105,"about_ca_topic_score_gemma":0.052519538,"teacher_disagreement_score":0.046980105,"about_ca_system_score_codex":0.0015247784,"about_ca_system_score_gemma":0.0031964155,"threshold_uncertainty_score":0.09341329},"labels":[],"label_agreement":null},{"id":"W4387875155","doi":"10.1038/s41597-023-02566-5","title":"1-km resolution rebound surfaces and paleotopography of glaciated North America since the Last Glacial Maximum","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Geological Survey of Canada; Natural Resources Canada","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Resources Canada; Natural Sciences and Engineering Research Council of Canada; Government of Canada; Université du Québec à Montréal","keywords":"Glacial period; Geology; Physical geography; Earth science; Geomorphology; Geography","score_opus":0.04252116280405769,"score_gpt":0.26417593035425047,"score_spread":0.22165476755019278,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387875155","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9547495,0.00020904058,0.0016565713,0.000100628546,0.000013858221,0.000012904158,0.03420587,0.00057662337,0.00847491],"genre_scores_gemma":[0.95717436,0.00017716862,0.0032536953,0.000011264643,0.000008197907,0.000024239263,0.037552413,0.00008682667,0.0017117349],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9999466,0.000007996347,0.0000034507411,0.00002133049,0.000012039562,0.000008558546],"domain_scores_gemma":[0.99983644,0.000027821992,0.000030242652,0.000026287631,0.00006633563,0.000012712933],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00013561451,0.00022609503,0.00011108292,0.0011898011,0.00017035805,0.0004492632,0.00022133872,0.00018934334,0.0025526807],"category_scores_gemma":[0.00038526396,0.00009795778,0.00017518844,0.0015558129,0.00010622315,0.00028485912,0.00023090179,0.00020072031,0.00054259825],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003273653,0.00014415898,0.74824244,0.00018567381,0.00014664704,0.00034401225,0.0012897896,0.14360368,0.0054412954,0.003033902,0.020199995,0.07704107],"study_design_scores_gemma":[0.000020420264,0.00001821503,0.9277551,0.000027714119,0.000024491934,0.000091431444,0.00043509816,0.050177235,0.0010488842,0.00061465794,0.019763723,0.000023004512],"about_ca_topic_score_codex":0.07720892,"about_ca_topic_score_gemma":0.10271605,"teacher_disagreement_score":0.92279106,"about_ca_system_score_codex":0.00047704627,"about_ca_system_score_gemma":0.00024823318,"threshold_uncertainty_score":0.15351903},"labels":[],"label_agreement":null},{"id":"W4387879092","doi":"10.1038/s41597-023-02549-6","title":"High-resolution (1 km) Köppen-Geiger maps for 1901–2099 based on constrained CMIP6 projections","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":718,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Centrum fÖr Personcentrerad Vård","keywords":"Climatology; Geiger counter; Coupled model intercomparison project; Climate model; Environmental science; Precipitation; Climate change; Representative Concentration Pathways; Climate system; Meteorology; Physical geography; Geography; Geology; Physics","score_opus":0.08465940337462678,"score_gpt":0.29728748149711537,"score_spread":0.21262807812248857,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387879092","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8325336,0.00032393466,0.020242345,0.00041437388,0.00005619188,0.00009481924,0.1241479,0.0013965326,0.02079027],"genre_scores_gemma":[0.90443254,0.00033839626,0.022247445,0.000053440395,0.000028121176,0.00011759113,0.07063848,0.00027114953,0.0018727516],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99982685,0.000037256967,0.000010284825,0.0000441287,0.000041535615,0.00004004999],"domain_scores_gemma":[0.9996265,0.00008302096,0.000068271125,0.00006378613,0.00013703766,0.000021276983],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00040722342,0.00054049277,0.00028513573,0.001657736,0.00032619064,0.0007747222,0.00044608375,0.00037911627,0.0042527975],"category_scores_gemma":[0.0012589145,0.00023987661,0.000554291,0.0023829762,0.00020247974,0.0006718382,0.00051644404,0.00044939894,0.00086643256],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004227794,0.000082642844,0.082843475,0.00014779174,0.00028438077,0.00041956233,0.00023659164,0.840175,0.003872257,0.004060494,0.021760792,0.04569422],"study_design_scores_gemma":[0.00014152876,0.000054933753,0.38891292,0.00014722136,0.00012807047,0.00019022811,0.0005073739,0.5629089,0.0060947607,0.004405861,0.03631186,0.00019627594],"about_ca_topic_score_codex":0.10510413,"about_ca_topic_score_gemma":0.08579079,"teacher_disagreement_score":0.10510413,"about_ca_system_score_codex":0.00094572623,"about_ca_system_score_gemma":0.00057753583,"threshold_uncertainty_score":0.20898473},"labels":[],"label_agreement":null},{"id":"W4387880964","doi":"10.1038/s41597-023-02651-9","title":"A geospatial database of close-to-reality travel times to obstetric emergency care in 15 Nigerian conurbations","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Global Maternal and Child Health","field":"Medicine","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; Women's College Hospital; University of Toronto","funders":"Fundação para a Ciência e a Tecnologia; Balliol College, University of Oxford; Fonds Wetenschappelijk Onderzoek; Nuffield College, University of Oxford; Bundesministerium für Gesundheit; Royal Society; Wellcome Trust; Google; Vlaamse regering; Bill and Melinda Gates Foundation","keywords":"Geospatial analysis; Geography; Database; Computer science; Cartography","score_opus":0.057429287489141036,"score_gpt":0.3576735155331557,"score_spread":0.3002442280440146,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387880964","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3347105,0.0006734027,0.007495958,0.00022141922,0.000053491345,0.00026571727,0.6454807,0.0012859509,0.00981282],"genre_scores_gemma":[0.4446136,0.00091184373,0.021658706,0.000038676728,0.000018581575,0.0005018095,0.5294273,0.0000969151,0.002732524],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9997402,0.000035951016,0.000062063074,0.00006693533,0.00006523524,0.00002962814],"domain_scores_gemma":[0.999281,0.00015087985,0.00013202413,0.00011972389,0.00023424331,0.00008219531],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003348607,0.0003600976,0.0004365195,0.003849431,0.00039098292,0.00078116433,0.0005296518,0.0003961535,0.003951573],"category_scores_gemma":[0.0014584422,0.00021653931,0.00030076428,0.0070352843,0.00017421396,0.0006570563,0.00047918534,0.00028440758,0.0014054239],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010260369,0.00075918995,0.44819435,0.0033522837,0.0003784053,0.0027472777,0.004605862,0.04814553,0.0055335164,0.008228104,0.16991939,0.30711007],"study_design_scores_gemma":[0.00012303125,0.00021536966,0.56300974,0.0010074491,0.00021785975,0.0018357033,0.012542568,0.05231641,0.0070807636,0.00305168,0.35836956,0.00022982986],"about_ca_topic_score_codex":0.041011035,"about_ca_topic_score_gemma":0.06978574,"teacher_disagreement_score":0.041011035,"about_ca_system_score_codex":0.0007167336,"about_ca_system_score_gemma":0.0014043387,"threshold_uncertainty_score":0.08154464},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4388077294","doi":"10.1038/s41597-023-02670-6","title":"Author Correction: Non-pharmaceutical interventions to combat COVID-19 in the Americas described through daily sub-national data","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Pharmaceutical Practices and Patient Outcomes","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Instituto Tecnológico y de Estudios Superiores de Monterrey; University of Miami","keywords":"Coronavirus disease 2019 (COVID-19); 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Psychological intervention; Pandemic; MEDLINE; Geography; Data science; Medicine; Virology; Computer science; Political science; Outbreak; Internal medicine; Nursing; Infectious disease (medical specialty)","score_opus":0.6615557178082387,"score_gpt":0.5739378310531165,"score_spread":0.08761788675512228,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388077294","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00017828449,0.00063751574,0.0008405639,0.11089618,0.8803917,0.00003242488,0.002480209,0.0004083542,0.004134827],"genre_scores_gemma":[0.027068842,0.010434503,0.012929089,0.27595985,0.38928473,0.0005283835,0.006612247,0.003677831,0.27350453],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9922438,0.0014067157,0.0016674937,0.000955895,0.0031821684,0.00054385635],"domain_scores_gemma":[0.9361144,0.023707435,0.003206656,0.00363804,0.031705074,0.0016284968],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0077259513,0.0019327344,0.001701129,0.0035585098,0.0034163683,0.0042854417,0.0033070857,0.006691992,0.038876716],"category_scores_gemma":[0.1256957,0.00125257,0.0014916415,0.003646538,0.0029073693,0.0024777905,0.0019417142,0.016235063,0.025605144],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014557784,0.0000026396503,0.000048652546,0.000053134198,0.0000053126523,0.000097899276,0.000033757056,0.000018644958,0.000009833564,0.0007429985,0.99692994,0.002042636],"study_design_scores_gemma":[0.000051786796,0.000013424973,0.0006054753,0.00059099187,0.000046078316,0.00062165497,0.00016509747,0.00026814456,0.00027054962,0.002431953,0.9948895,0.00004536844],"about_ca_topic_score_codex":0.03447685,"about_ca_topic_score_gemma":0.033961583,"teacher_disagreement_score":0.038876716,"about_ca_system_score_codex":0.005161496,"about_ca_system_score_gemma":0.010602244,"threshold_uncertainty_score":0.13005555},"labels":[],"label_agreement":null},{"id":"W4388305010","doi":"10.1038/s41597-023-02523-2","title":"CreelCat, a Catalog of United States Inland Creel and Angler Survey Data","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Archaeology and Natural History","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Memorial University of Newfoundland","funders":"U.S. Geological Survey; U.S. Department of the Interior","keywords":"Geography","score_opus":0.17752624405968445,"score_gpt":0.37486289223184516,"score_spread":0.1973366481721607,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388305010","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014690979,0.00042932533,0.0019291233,0.00017370067,0.00007178333,0.0004754787,0.9410867,0.00085147435,0.040291462],"genre_scores_gemma":[0.017975306,0.00046887525,0.0051993695,0.00021850968,0.000030445273,0.00064565294,0.9606423,0.00024919442,0.014570343],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991013,0.00010916324,0.00011991111,0.00019041728,0.00038614165,0.00009314956],"domain_scores_gemma":[0.99273527,0.0005340025,0.0010945449,0.0008150925,0.0043936134,0.00042751824],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010691615,0.0006042228,0.00036173884,0.009536131,0.0006114947,0.0011876735,0.0011190429,0.0003399262,0.038098387],"category_scores_gemma":[0.0053889393,0.00034533354,0.00021360222,0.017363945,0.00016394127,0.0012559814,0.00077118515,0.00044179545,0.021072332],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008397838,0.00007172011,0.0502043,0.0002908365,0.000048096997,0.000090383866,0.00030287812,0.0003040208,0.0006800314,0.0014579651,0.8919468,0.05451895],"study_design_scores_gemma":[0.000029632141,0.000021808402,0.13316531,0.00024399866,0.000022915901,0.00007229795,0.0006906431,0.000697063,0.0004737233,0.0003560623,0.86419415,0.000032548014],"about_ca_topic_score_codex":0.20275612,"about_ca_topic_score_gemma":0.28984833,"teacher_disagreement_score":0.20275612,"about_ca_system_score_codex":0.001239571,"about_ca_system_score_gemma":0.0045625065,"threshold_uncertainty_score":0.40315187},"labels":[],"label_agreement":null},{"id":"W4388490712","doi":"10.1038/s41597-023-02719-6","title":"Publisher Correction: 1-km resolution rebound surfaces and paleotopography of glaciated North America since the Last Glacial Maximum","year":2023,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université du Québec à Montréal; Geological Survey of Canada; Natural Resources Canada","funders":"","keywords":"Glacial period; Physical geography; Geology; Last Glacial Maximum; Geography; Geomorphology","score_opus":0.03695853775976452,"score_gpt":0.2562997780948711,"score_spread":0.21934124033510657,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388490712","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00027576688,0.00063093874,0.00079920003,0.015403861,0.9716114,0.000016127096,0.003684876,0.00042773283,0.0071501443],"genre_scores_gemma":[0.025635805,0.0077646333,0.011136769,0.03462009,0.20900254,0.0002085211,0.018034363,0.0065299403,0.68706733],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99801874,0.0002208844,0.00037633517,0.00031355972,0.00093721284,0.0001332719],"domain_scores_gemma":[0.9753975,0.0038932243,0.00080014736,0.0019720646,0.017235858,0.00070124277],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022052336,0.0024292725,0.0018705898,0.0047385097,0.003992441,0.0042166207,0.0033321243,0.003254944,0.07638612],"category_scores_gemma":[0.038174145,0.0014419291,0.0012511552,0.005125034,0.0015865352,0.0025863857,0.0017886977,0.007623652,0.045063585],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000107620135,0.0000026931448,0.00007166008,0.000041129697,0.0000040573586,0.000091631126,0.000032803964,0.000052140036,0.000022466334,0.0004517959,0.995854,0.0033649148],"study_design_scores_gemma":[0.000017749206,0.000006686324,0.0012617437,0.00021637419,0.00002156113,0.00026509303,0.00012648274,0.00024553644,0.00020340008,0.0010147122,0.99659747,0.000023145045],"about_ca_topic_score_codex":0.094865076,"about_ca_topic_score_gemma":0.119914785,"teacher_disagreement_score":0.9051349,"about_ca_system_score_codex":0.0037783864,"about_ca_system_score_gemma":0.005637945,"threshold_uncertainty_score":0.25553703},"labels":[],"label_agreement":null},{"id":"W4388592011","doi":"10.1038/s41597-023-02633-x","title":"Global dataset of soil organic carbon in tidal marshes","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Coastal wetland ecosystem dynamics","field":"Environmental Science","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; McGill University","funders":"Agencia Estatal de Investigación; Fundação para a Ciência e a Tecnologia; South African Agency for Science and Technology Advancement; Sight Research UK; Xunta de Galicia; Deutsche Forschungsgemeinschaft; National Research Foundation; Ministerio de Ciencia e Innovación; Higher Education Funding Council for Wales; Nelson Mandela University; U.S. Department of Defense; Department of Science and Innovation, South Africa; Fundación Séneca; Natural Environment Research Council","keywords":"Marsh; Soil carbon; Environmental science; Total organic carbon; Oceanography; Carbon fibers; Wetland; Ecology; Soil science; Geology; Soil water; Biology; Computer science","score_opus":0.02164637414739909,"score_gpt":0.25003930628751136,"score_spread":0.22839293214011228,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388592011","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06583134,0.00014402978,0.00072503893,0.00006819875,0.000018693196,0.00003021652,0.93167955,0.00037358145,0.0011293378],"genre_scores_gemma":[0.053299993,0.00007519904,0.0015423243,0.00003892159,0.000009546217,0.00008653606,0.9444244,0.000032958345,0.00049015385],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995402,0.00006110934,0.00007110535,0.00015051682,0.00011206323,0.000065021835],"domain_scores_gemma":[0.9981396,0.00023365086,0.00038952008,0.00042915996,0.0006060876,0.00020184576],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005612075,0.0005503154,0.00040580428,0.0021209528,0.00021606578,0.0005138918,0.00059703796,0.00052719214,0.0029930905],"category_scores_gemma":[0.001907772,0.00022676947,0.00059022807,0.0043546516,0.00021129182,0.0004687588,0.00097313855,0.00042490993,0.0025075283],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006592962,0.00028307,0.58033675,0.0019479207,0.0008730886,0.0005649586,0.0005040343,0.017354025,0.010799824,0.0017931826,0.32437715,0.060506694],"study_design_scores_gemma":[0.00014681656,0.000065281674,0.8159083,0.00013973146,0.00007722093,0.00014297118,0.00029885196,0.0056935353,0.002697334,0.00065640296,0.17411175,0.000061692255],"about_ca_topic_score_codex":0.025725327,"about_ca_topic_score_gemma":0.027847955,"teacher_disagreement_score":0.025725327,"about_ca_system_score_codex":0.00038154863,"about_ca_system_score_gemma":0.00067074003,"threshold_uncertainty_score":0.051151216},"labels":[],"label_agreement":null},{"id":"W4388729121","doi":"10.1038/s41597-023-02732-9","title":"Canada Source Watershed Polygons (Can-SWaP): A dataset for the protection of Canada’s municipal water supply","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Natural Resources Canada; Pacific Salmon Foundation; Ontario Forest Research Institute; Canadian Forest Service","funders":"Natural Resources Canada; U.S. Forest Service; Indigenous Services Canada","keywords":"Swap (finance); Watershed; Environmental science; Water supply; Surface water; Water resource management; STREAMS; Hydrology (agriculture); Environmental resource management; Business; Environmental engineering; Computer science; Engineering","score_opus":0.035135982471503416,"score_gpt":0.23212180834235696,"score_spread":0.19698582587085356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388729121","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003940727,0.000095617455,0.000624813,0.00012120157,0.000015015697,0.000042980242,0.9911601,0.000522683,0.0034769436],"genre_scores_gemma":[0.011646737,0.00010263023,0.0030543087,0.00003845603,0.0000048179754,0.000082910614,0.98355955,0.00013878956,0.0013718343],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99899834,0.0000640307,0.00006542273,0.00017359298,0.00047369578,0.00022488007],"domain_scores_gemma":[0.99593145,0.00021626237,0.00022020056,0.00037625324,0.0029109998,0.00034480213],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005740238,0.00080917735,0.00065450853,0.0048680734,0.0017422638,0.0018934347,0.0019357344,0.0005831226,0.012695401],"category_scores_gemma":[0.0046424186,0.00042613232,0.0007244812,0.013929751,0.0005787958,0.00064635236,0.0015512591,0.0011573562,0.004524723],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012298257,0.000055259992,0.023417609,0.00037835375,0.000110540066,0.00012769202,0.00042362054,0.0032822092,0.00048386326,0.004047709,0.95130616,0.016244093],"study_design_scores_gemma":[0.00013189937,0.000009952095,0.08895465,0.00029897087,0.00004590651,0.00008246946,0.0010779726,0.006758672,0.0009924178,0.0016094935,0.8999536,0.00008397917],"about_ca_topic_score_codex":0.9769845,"about_ca_topic_score_gemma":0.9894135,"teacher_disagreement_score":0.0230155,"about_ca_system_score_codex":0.014035411,"about_ca_system_score_gemma":0.035401814,"threshold_uncertainty_score":0.101834536},"labels":[],"label_agreement":null},{"id":"W4389224631","doi":"10.1038/s41597-023-02762-3","title":"EarSet: A Multi-Modal Dataset for Studying the Impact of Head and Facial Movements on In-Ear PPG Signals","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Non-Invasive Vital Sign Monitoring","field":"Engineering","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bell (Canada)","funders":"","keywords":"Computer science; Photoplethysmogram; Computer vision; Artificial intelligence; Motion (physics); Head (geology); Modal; Face (sociological concept); Acoustics; Geology; Materials science; Physics","score_opus":0.15494409441810533,"score_gpt":0.3776348428945414,"score_spread":0.22269074847643608,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389224631","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19852099,0.0052313237,0.04990807,0.0009330775,0.0012773778,0.00083289103,0.7210001,0.013119464,0.0091766985],"genre_scores_gemma":[0.21184474,0.0013947455,0.037232958,0.00047037212,0.0003328806,0.0012959812,0.7417721,0.00047729004,0.0051789563],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99946743,0.00009077664,0.000064148924,0.00014578516,0.00015915305,0.00007276349],"domain_scores_gemma":[0.99948466,0.00014516391,0.00006641203,0.00011416158,0.00013584434,0.000053853695],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00057248195,0.0018397849,0.0008568191,0.001523713,0.00039270832,0.0006145648,0.0009975899,0.0013935434,0.0034936154],"category_scores_gemma":[0.0017742983,0.00020045412,0.0008886416,0.00086178666,0.0002644836,0.0005070149,0.0012450686,0.00065420696,0.003042781],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0038895947,0.0012182563,0.05324171,0.004986417,0.001117242,0.0026090615,0.00058788987,0.012415256,0.1052379,0.0014629067,0.52944344,0.28379032],"study_design_scores_gemma":[0.0008838605,0.001946266,0.3800664,0.0009814493,0.0008466724,0.009244362,0.0013692784,0.11752093,0.06588422,0.0064840554,0.4141513,0.0006211999],"about_ca_topic_score_codex":0.0035101452,"about_ca_topic_score_gemma":0.011436913,"teacher_disagreement_score":0.0035101452,"about_ca_system_score_codex":0.00030639378,"about_ca_system_score_gemma":0.0004310283,"threshold_uncertainty_score":0.011687279},"labels":[],"label_agreement":null},{"id":"W4389306898","doi":"10.1038/s41597-023-02752-5","title":"Introducing MEG-MASC a high-quality magneto-encephalography dataset for evaluating natural speech processing","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Baycrest Hospital","funders":"Agence Nationale de la Recherche","keywords":"Computer science; Replicate; Magnetoencephalography; Metadata; Decoding methods; Natural language processing; Speech recognition; Active listening; Artificial intelligence; Electroencephalography; Psychology; Communication; World Wide Web","score_opus":0.20122518404499928,"score_gpt":0.4144129988181525,"score_spread":0.21318781477315324,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389306898","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.085711874,0.0012173825,0.17995809,0.00095530314,0.00063739857,0.0011802011,0.6709073,0.04788001,0.011552424],"genre_scores_gemma":[0.07564112,0.00033498704,0.10529476,0.0003970957,0.00019041452,0.0026333712,0.80880433,0.0032809891,0.0034229474],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992368,0.00015435826,0.00008218869,0.00023356866,0.000228145,0.00006492403],"domain_scores_gemma":[0.9988952,0.000276595,0.00008528436,0.00038235547,0.00027165835,0.00008900503],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011220216,0.0012986289,0.00063135626,0.0016549913,0.0006229173,0.0012895125,0.0016178555,0.00093924196,0.0067762015],"category_scores_gemma":[0.005196038,0.000333227,0.0008391744,0.00116427,0.00049348903,0.00089526834,0.002401685,0.0009973581,0.007602738],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010963483,0.0005969477,0.028769903,0.0028172522,0.0008016199,0.0014272066,0.000983634,0.019334208,0.098449655,0.0066824406,0.6535638,0.18547688],"study_design_scores_gemma":[0.00062137144,0.0005096762,0.1599682,0.0005502857,0.00033500596,0.003351432,0.00069849676,0.087495126,0.05988279,0.025005756,0.66116625,0.00041557717],"about_ca_topic_score_codex":0.004413355,"about_ca_topic_score_gemma":0.011703826,"teacher_disagreement_score":0.0067762015,"about_ca_system_score_codex":0.00040443867,"about_ca_system_score_gemma":0.0010432573,"threshold_uncertainty_score":0.02266872},"labels":[],"label_agreement":null},{"id":"W4389426225","doi":"10.1038/s41597-023-02809-5","title":"A dataset of proteomic changes during human heat stress and heat acclimation","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Thermoregulation and physiological responses","field":"Medicine","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Montreal Heart Institute","funders":"Fondation Institut de Cardiologie de Montréal; Institut de Cardiologie de Montréal; Canada Foundation for Innovation; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Acclimatization; Heat stress; Context (archaeology); Biology; Multiplex; Medicine; Bioinformatics; Ecology; Animal science","score_opus":0.1357171603890788,"score_gpt":0.381036359776381,"score_spread":0.2453191993873022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389426225","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019204943,0.0020284536,0.00090856117,0.0004075976,0.00014431177,0.0000672183,0.9750072,0.0005665876,0.0016651199],"genre_scores_gemma":[0.014615188,0.0006013164,0.0016446478,0.00022586534,0.000045697794,0.0001802025,0.98207825,0.000037630154,0.0005712315],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9989134,0.00017519404,0.00015319641,0.00042763262,0.00019208173,0.00013842426],"domain_scores_gemma":[0.9987311,0.0003980755,0.00020861485,0.00022662338,0.0002864138,0.0001491735],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00085909176,0.0012581103,0.0012282886,0.0022598042,0.00078125106,0.0012916299,0.001171062,0.001928774,0.004609734],"category_scores_gemma":[0.003070304,0.00024613217,0.0013883627,0.002920974,0.0003731469,0.0005480873,0.0017673764,0.00095385703,0.0054756515],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030634396,0.000542629,0.11617436,0.012371144,0.0013644261,0.0019975824,0.00060674886,0.0048343083,0.021356547,0.0017003958,0.76915574,0.066832766],"study_design_scores_gemma":[0.00036839876,0.00036479093,0.3052445,0.0015249047,0.00049820344,0.0019708502,0.0006507901,0.0047890223,0.006578542,0.0033919264,0.67442596,0.00019201906],"about_ca_topic_score_codex":0.005998715,"about_ca_topic_score_gemma":0.012577302,"teacher_disagreement_score":0.005998715,"about_ca_system_score_codex":0.00069847534,"about_ca_system_score_gemma":0.0010783374,"threshold_uncertainty_score":0.0154210925},"labels":[],"label_agreement":null},{"id":"W4389482760","doi":"10.1038/s41597-023-02759-y","title":"Canadian Covid-19 pandemic public health mitigation measures at the province level","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"York University","funders":"","keywords":"Pandemic; Public health; Coronavirus disease 2019 (COVID-19); Range (aeronautics); Implementation; Scale (ratio); Geography; 2019-20 coronavirus outbreak; Work (physics); Psychological intervention; Data science; Computer science; Medicine; Cartography; Outbreak; Nursing; Engineering; Virology","score_opus":0.7576683147232829,"score_gpt":0.49529430542492964,"score_spread":0.26237400929835325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389482760","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.059642885,0.0018902257,0.0064571593,0.0049866927,0.00022621389,0.00071212766,0.834232,0.0005015626,0.091351144],"genre_scores_gemma":[0.41115934,0.007849462,0.041746724,0.00181369,0.00016370758,0.0015770586,0.4787067,0.0003279381,0.056655522],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9973634,0.00020667973,0.00017599495,0.00026786985,0.0015894179,0.00039673428],"domain_scores_gemma":[0.98916954,0.0008724828,0.00086367485,0.0005029624,0.007809225,0.00078205904],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002215812,0.0005046425,0.0005866365,0.0051430846,0.002992745,0.0028987539,0.0018081219,0.0004344555,0.018897649],"category_scores_gemma":[0.013589044,0.00044101724,0.0007579052,0.011099479,0.0005499185,0.00076123077,0.0016735571,0.0011495097,0.0017633179],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035855317,0.00012184891,0.285667,0.0019807864,0.00033699922,0.00022627872,0.0027454218,0.005652252,0.0006867166,0.037068058,0.5490533,0.11610279],"study_design_scores_gemma":[0.000101586425,0.000071074144,0.42424715,0.0012132206,0.00015753864,0.000111106,0.0036135656,0.0051567536,0.0011958864,0.0037132325,0.5602587,0.00016021602],"about_ca_topic_score_codex":0.9964024,"about_ca_topic_score_gemma":0.9976464,"teacher_disagreement_score":0.04053258,"about_ca_system_score_codex":0.04053258,"about_ca_system_score_gemma":0.11052611,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"},{"model":"grok","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"},{"model":"opus","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4389628449","doi":"10.1038/s41597-023-02805-9","title":"Deep learning downscaled high-resolution daily near surface meteorological datasets over East Asia","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Education and Child Care","funders":"National Key Research and Development Program of China","keywords":"Shortwave radiation; Downscaling; Environmental science; Shortwave; Coupled model intercomparison project; Climatology; Precipitation; Wind speed; Climate change; East Asia; Relative humidity; Climate model; Meteorology; Humidity; Longwave; Atmospheric sciences; Geography; Radiation; Radiative transfer; Geology; China","score_opus":0.04810556784158295,"score_gpt":0.2743167713279881,"score_spread":0.22621120348640514,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389628449","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8792163,0.00049673225,0.027852174,0.00050014385,0.00019545367,0.0001372683,0.08119911,0.003869891,0.006532817],"genre_scores_gemma":[0.79381686,0.00036032475,0.030043343,0.00016127943,0.00006864972,0.00016966197,0.17165285,0.00015357634,0.0035735571],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998172,0.000019315794,0.000020732945,0.000062838735,0.0000438996,0.000036047786],"domain_scores_gemma":[0.9996842,0.000033798187,0.000033777986,0.000092253256,0.00012230012,0.0000337133],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00042576768,0.0007014982,0.0003552728,0.0008339285,0.00028627188,0.00038363689,0.0008408938,0.00039100976,0.0021844015],"category_scores_gemma":[0.00084020174,0.00028293574,0.00072972616,0.0014502178,0.00020361094,0.0006878269,0.00081045297,0.00076461985,0.0007203948],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00059536594,0.0009337159,0.16428818,0.00048054868,0.0008780053,0.00084361923,0.00036311653,0.47411728,0.02005866,0.0028665129,0.05723916,0.27733582],"study_design_scores_gemma":[0.00013277974,0.000105586776,0.1564408,0.00005301222,0.00010136687,0.000085836866,0.00020297992,0.81283915,0.010907335,0.0018410457,0.017216219,0.00007393893],"about_ca_topic_score_codex":0.055755664,"about_ca_topic_score_gemma":0.06377083,"teacher_disagreement_score":0.055755664,"about_ca_system_score_codex":0.00055415655,"about_ca_system_score_gemma":0.0011246465,"threshold_uncertainty_score":0.110862255},"labels":[],"label_agreement":null},{"id":"W4389894118","doi":"10.1038/s41597-023-02745-4","title":"Global inventory of species categorized by known underwater sonifery","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine animal studies overview","field":"Environmental Science","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University; University of Victoria; Simon Fraser University","funders":"Vlaamse regering; Fonds Wetenschappelijk Onderzoek; Vlaams Instituut voor de Zee; LifeWatch – Niclas Öberg Foundation; Richard Lounsbery Foundation","keywords":"Biodiversity; Invertebrate; Underwater; Marine life; Ecology; Subspecies; Marine biodiversity; Environmental resource management; Biology; Marine species; Geography; Environmental science","score_opus":0.06800758008801995,"score_gpt":0.2791835503204395,"score_spread":0.21117597023241957,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389894118","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08853126,0.0030226246,0.003315937,0.0001894665,0.000077032906,0.00015888219,0.88507044,0.0005555745,0.01907885],"genre_scores_gemma":[0.067604795,0.002251784,0.011126186,0.00012456738,0.000042837968,0.0004955768,0.9146502,0.00018883846,0.0035152603],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992256,0.00008614494,0.00015711793,0.00026178843,0.00019014136,0.00007919434],"domain_scores_gemma":[0.99625117,0.00071719877,0.0011965444,0.0004576526,0.0010143484,0.00036314994],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012219754,0.000685348,0.00066454645,0.021063317,0.00040539246,0.0008010219,0.0008779764,0.00024333864,0.013240516],"category_scores_gemma":[0.002743148,0.0003565763,0.00061072735,0.01630696,0.00034252895,0.0013330176,0.0020276902,0.00044683026,0.004760011],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00052851246,0.000083691506,0.3983645,0.009632949,0.00093267305,0.00046789547,0.0050778124,0.0018453954,0.011949998,0.0070105023,0.29639238,0.26771367],"study_design_scores_gemma":[0.000027891627,0.00006821316,0.5904321,0.00077430584,0.00021180225,0.00051818445,0.0014370292,0.0005865755,0.0011767595,0.0010691347,0.40364113,0.00005678026],"about_ca_topic_score_codex":0.011955284,"about_ca_topic_score_gemma":0.013751805,"teacher_disagreement_score":0.021063317,"about_ca_system_score_codex":0.0005127396,"about_ca_system_score_gemma":0.0010262791,"threshold_uncertainty_score":0.04429394},"labels":[],"label_agreement":null},{"id":"W4390007241","doi":"10.1038/s41597-023-02812-w","title":"A dataset of formulation compositions for self-emulsifying drug delivery systems","year":2023,"lang":"en","type":"article","venue":"Scientific Data","topic":"Analytical Chemistry and Chromatography","field":"Chemistry","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium; University of Toronto","funders":"Canada First Research Excellence Fund; University of Toronto; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Drug delivery; Drug; Computer science; Chemistry; Pharmacology; Medicine; Organic chemistry","score_opus":0.05732461586954081,"score_gpt":0.3054587720831645,"score_spread":0.24813415621362372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390007241","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11215216,0.016600607,0.020474432,0.0014494685,0.0005638036,0.0011126829,0.8322734,0.0051262425,0.010247221],"genre_scores_gemma":[0.0697047,0.004444412,0.037358902,0.0007228868,0.000105167965,0.0011004227,0.88353145,0.0004998047,0.0025322058],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99727386,0.0004295875,0.0004081435,0.00064492837,0.0011031767,0.00014037697],"domain_scores_gemma":[0.99668735,0.001124433,0.00048575067,0.00053310086,0.00097400986,0.00019541898],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025577194,0.0014329507,0.0012638939,0.0036019273,0.0008200821,0.0018008532,0.0016764938,0.0022755307,0.0031162256],"category_scores_gemma":[0.0068959603,0.00042262187,0.0017413574,0.0032833153,0.00066070305,0.0012649532,0.0013267575,0.0017473907,0.0035482412],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003664866,0.0030121366,0.059360147,0.025890447,0.0017481507,0.0025715542,0.0006030774,0.042591315,0.09221599,0.008042225,0.4554663,0.3048338],"study_design_scores_gemma":[0.0004638022,0.0010328253,0.04494383,0.001140026,0.0005217444,0.0015272251,0.0003546114,0.03279468,0.04657563,0.00474406,0.8656208,0.00028084812],"about_ca_topic_score_codex":0.0044430383,"about_ca_topic_score_gemma":0.007516217,"teacher_disagreement_score":0.0044430383,"about_ca_system_score_codex":0.0014357358,"about_ca_system_score_gemma":0.0021771404,"threshold_uncertainty_score":0.013526678},"labels":[],"label_agreement":null},{"id":"W4390545674","doi":"10.1038/s41597-023-02784-x","title":"Global fine-resolution data on springtail abundance and community structure","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Collembola Taxonomy and Ecology Studies","field":"Agricultural and Biological Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Forest Service; Western University; Université du Québec à Montréal; Université du Québec en Outaouais; Natural Resources Canada; Université Laval","funders":"Natural Resources Canada; Natural Sciences and Engineering Research Council of Canada; Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Agencia Estatal de Investigación; Higher Education Commision, Pakistan; Universität Leipzig; Russian Science Foundation; Agence Nationale pour la Gestion des Déchets Radioactifs; Office of Energy Research and Development; Latvijas Zinātnes Padome; Centre National de la Recherche Scientifique; Canadian Forest Service; Japan Society for the Promotion of Science; Vedecká Grantová Agentúra MŠVVaŠ SR a SAV; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Ministério da Ciência, Tecnologia e Inovação; Russian Foundation for Basic Research; National Natural Science Foundation of China; Deutsche Forschungsgemeinschaft; National Research Foundation; Austrian Science Fund; Ministerio de Ciencia e Innovación; Eesti Teadusfondi; Massey University; Royal Society Te Apārangi; Nanyang Technological University; Biodiversa+; U.S. Forest Service; Government of Alberta; Department of Science and Technology, Ministry of Science and Technology, India; Carl Tryggers Stiftelse för Vetenskaplig Forskning; Alberta Environment and Parks; Marsden Fund; Bundesministerium für Bildung und Forschung; Ural Branch, Russian Academy of Sciences; Agentúra na Podporu Výskumu a Vývoja; Alexander von Humboldt-Stiftung; Klima- und Energiefonds; St. John's University; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Russian Academy of Sciences; Agence Nationale de la Recherche; Ministry of Natural Resources; Conseil Régional, Île-de-France; American Association of University Women; National Science Foundation","keywords":"Springtail; Abundance (ecology); Ecology; Geography; Database; Woodland; Habitat; Biology","score_opus":0.09691553568951436,"score_gpt":0.2835592502376938,"score_spread":0.18664371454817943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390545674","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.46238038,0.0023444763,0.0068531125,0.00020626129,0.00006221508,0.0000951705,0.51975465,0.0006887317,0.0076149437],"genre_scores_gemma":[0.36976075,0.0013404355,0.015007199,0.00016158448,0.00008716397,0.00033808927,0.6103206,0.0001527606,0.0028314088],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992223,0.000083921586,0.00011029665,0.00028892583,0.00019977802,0.000094790485],"domain_scores_gemma":[0.99584204,0.0006928431,0.0012314187,0.00075072906,0.0010851743,0.00039781432],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011220707,0.00048248467,0.00057888904,0.005925162,0.00042589556,0.000745938,0.00053271727,0.00040345115,0.0049794335],"category_scores_gemma":[0.003107423,0.00027678392,0.0005070074,0.00882603,0.0002360615,0.0009300063,0.0014174066,0.00034189198,0.0022473864],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019393925,0.00007417451,0.87715584,0.0013920426,0.00053018145,0.00023653838,0.0016422212,0.0008611917,0.020835452,0.0006098151,0.025597144,0.07087152],"study_design_scores_gemma":[0.000014468762,0.000033736265,0.96630013,0.00007800285,0.00007830426,0.0001336687,0.0004116778,0.00021101322,0.0007881616,0.00025273144,0.031680062,0.000017926599],"about_ca_topic_score_codex":0.012683262,"about_ca_topic_score_gemma":0.028879043,"teacher_disagreement_score":0.012683262,"about_ca_system_score_codex":0.00029670354,"about_ca_system_score_gemma":0.0006662768,"threshold_uncertainty_score":0.025218904},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4390583429","doi":"10.1038/s41597-023-02866-w","title":"FISHGLOB_data: an integrated dataset of fish biodiversity sampled with scientific bottom-trawl surveys","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and fisheries research","field":"Environmental Science","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Fisheries and Oceans Canada","funders":"National Marine Fisheries Service; Fondation pour la Recherche sur la Biodiversite; European Commission; Cooperative Institute for the North Atlantic Region; National Oceanic and Atmospheric Administration; Villum Fonden","keywords":"Geography; Biodiversity; Context (archaeology); Fishery; Abundance (ecology); Marine protected area; Fishing; Marine ecosystem; Environmental resource management; Ecosystem; Oceanography; Ecology; Environmental science; Habitat; Biology; Geology","score_opus":0.0694030363355991,"score_gpt":0.29348390382763295,"score_spread":0.22408086749203385,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390583429","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032253729,0.00006813159,0.00042801903,0.00005112982,0.000018060704,0.000024121588,0.9952734,0.0003091697,0.00060257805],"genre_scores_gemma":[0.0036387385,0.000047756934,0.0014575651,0.000036719037,0.000007799778,0.00013096303,0.9943136,0.000056809113,0.00031002186],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987029,0.00017554458,0.00025403194,0.00038435895,0.00030270615,0.00018033656],"domain_scores_gemma":[0.996229,0.00080673455,0.00091443735,0.0006071483,0.0011068597,0.00033583722],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001371667,0.0011172642,0.0008332585,0.0046557905,0.00045066967,0.0011224343,0.001526446,0.0010355188,0.009372685],"category_scores_gemma":[0.0058593857,0.0005441308,0.0009137178,0.007720127,0.00033008316,0.0010155791,0.0020568997,0.0011488808,0.009725403],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025308155,0.0001233054,0.094104975,0.0019534072,0.00037364863,0.00017025998,0.0004122975,0.0023242668,0.0017774537,0.001587747,0.8722177,0.024702003],"study_design_scores_gemma":[0.00023462907,0.00006100694,0.24282984,0.00052432646,0.0001352057,0.00021615061,0.00071964617,0.0025891627,0.0016507297,0.0013304793,0.74959797,0.00011091403],"about_ca_topic_score_codex":0.03314832,"about_ca_topic_score_gemma":0.050320126,"teacher_disagreement_score":0.03314832,"about_ca_system_score_codex":0.0010007918,"about_ca_system_score_gemma":0.0019456827,"threshold_uncertainty_score":0.06591076},"labels":[],"label_agreement":null},{"id":"W4390672990","doi":"10.1038/s41597-023-02824-6","title":"A database of mapped global fishing activity 1950–2017","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and fisheries research","field":"Environmental Science","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Fishing; Tonnage; Marine conservation; Computer science; Range (aeronautics); Database; Identification (biology); Automatic Identification System; Environmental science; Fishery; Environmental resource management; Oceanography; Data mining; Ecology; Engineering; Geology","score_opus":0.08081955254806912,"score_gpt":0.33877929819720226,"score_spread":0.25795974564913315,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390672990","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012199436,0.00032124162,0.00092627277,0.000038031976,0.0000438011,0.000042250802,0.9821547,0.0002789673,0.0039952965],"genre_scores_gemma":[0.010733046,0.00034397433,0.0015384094,0.000018229974,0.000018568266,0.0001509973,0.9856367,0.000056632955,0.0015034718],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99937016,0.000049629427,0.00014757115,0.00019377339,0.00016561264,0.000073264004],"domain_scores_gemma":[0.9974389,0.0003076975,0.000662211,0.00037996456,0.000996201,0.00021503642],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005813116,0.0008354783,0.00073615735,0.009883458,0.000372379,0.0010553448,0.0007733558,0.00040486627,0.01567134],"category_scores_gemma":[0.0026669367,0.00031314383,0.0006681828,0.013439289,0.0002369685,0.00092060654,0.0012001405,0.00064466515,0.009729877],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006754042,0.0002239562,0.1745192,0.0047867694,0.0006963648,0.00097518775,0.001154014,0.008291392,0.004141928,0.006640221,0.6399648,0.15793072],"study_design_scores_gemma":[0.00005600368,0.0000562778,0.2691974,0.000668904,0.00012867518,0.00040364818,0.0006875234,0.0014724914,0.001530578,0.0012595867,0.7244624,0.00007658066],"about_ca_topic_score_codex":0.029765498,"about_ca_topic_score_gemma":0.03031833,"teacher_disagreement_score":0.029765498,"about_ca_system_score_codex":0.00082823465,"about_ca_system_score_gemma":0.0020935594,"threshold_uncertainty_score":0.05918449},"labels":[],"label_agreement":null},{"id":"W4390673179","doi":"10.1038/s41597-023-02796-7","title":"Measuring the presence and incidence of cholera in Hindustan: New data from primary sources for the colonial era","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Vibrio bacteria research studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Center for Interuniversity Research and Analysis on Organizations","funders":"Agence Nationale de la Recherche","keywords":"Harmonization; Cholera; Data collection; Colonialism; Geography; Incidence (geometry); Data science; Regional science; Computer science; Statistics; Mathematics; Biology","score_opus":0.10411196234134029,"score_gpt":0.33325454022855044,"score_spread":0.22914257788721015,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390673179","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8279194,0.0005928607,0.001082762,0.0003822884,0.000057948422,0.00012001524,0.16677023,0.00006775038,0.003006716],"genre_scores_gemma":[0.6596338,0.00048990577,0.0067028957,0.00015925722,0.000107331034,0.0003898148,0.3299796,0.00003355509,0.0025039164],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9991554,0.00021718103,0.0000926275,0.00015958722,0.00020323698,0.00017201698],"domain_scores_gemma":[0.9962941,0.0009138894,0.0009720669,0.00062448863,0.00085720816,0.00033811294],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007512437,0.00041399355,0.00041345344,0.005043114,0.00056188455,0.0008791068,0.000657575,0.0004693857,0.0024419113],"category_scores_gemma":[0.0036905801,0.00022839493,0.00039817474,0.006390191,0.00038107208,0.00030284259,0.0012783691,0.0005653592,0.0010905806],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002810969,0.00024262592,0.9249954,0.00063340185,0.0003581923,0.0013582973,0.003018725,0.0032852027,0.0022064915,0.000798233,0.025359062,0.037463147],"study_design_scores_gemma":[0.000025151423,0.00005770163,0.9648022,0.000105789906,0.00007695256,0.0003675104,0.0013643564,0.00086422334,0.0008501086,0.00011714969,0.03133972,0.000029067514],"about_ca_topic_score_codex":0.08796861,"about_ca_topic_score_gemma":0.101084776,"teacher_disagreement_score":0.08796861,"about_ca_system_score_codex":0.0010311676,"about_ca_system_score_gemma":0.0014879218,"threshold_uncertainty_score":0.17491317},"labels":[],"label_agreement":null},{"id":"W4390740994","doi":"10.1038/s41597-023-02855-z","title":"An ensemble of bias-adjusted CMIP6 climate simulations based on a high-resolution North American reanalysis","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ouranos","funders":"Environment and Climate Change Canada","keywords":"Quantile; Climate model; Coupled model intercomparison project; Climatology; Climate change; Computer science; Environmental science; Multivariate statistics; Precipitation; Econometrics; Statistics; Meteorology; Mathematics; Machine learning; Geography","score_opus":0.07410897584124482,"score_gpt":0.30279054523550447,"score_spread":0.22868156939425965,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390740994","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2814891,0.00051888445,0.030882483,0.00063138205,0.00039442416,0.00025255847,0.6671183,0.008818827,0.009894064],"genre_scores_gemma":[0.2521684,0.00029871447,0.04716574,0.00017684058,0.00008480955,0.00041637494,0.69673955,0.0010619597,0.0018877048],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999648,0.00007592723,0.000029492887,0.000110291454,0.00009831403,0.00003803716],"domain_scores_gemma":[0.9990565,0.00017219757,0.00008643936,0.00034263084,0.00028226653,0.000059979717],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011276288,0.000632336,0.0005708982,0.00093259243,0.0003803096,0.0006183555,0.0014597873,0.00071550615,0.0041948534],"category_scores_gemma":[0.0022316745,0.00045536496,0.0010078882,0.0024483895,0.00025932657,0.0006405718,0.0006422027,0.0011761452,0.0018548243],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00054572854,0.00048897706,0.09469801,0.0005867669,0.0014360903,0.00021285325,0.00030845634,0.5420204,0.007762668,0.006061986,0.253366,0.09251197],"study_design_scores_gemma":[0.00080374203,0.00014422667,0.11908563,0.00013873345,0.00028663018,0.00017995301,0.00019317928,0.63990325,0.01298676,0.01061724,0.2153869,0.00027379935],"about_ca_topic_score_codex":0.026689809,"about_ca_topic_score_gemma":0.03741917,"teacher_disagreement_score":0.026689809,"about_ca_system_score_codex":0.0006127701,"about_ca_system_score_gemma":0.0012826212,"threshold_uncertainty_score":0.053068876},"labels":[],"label_agreement":null},{"id":"W4390782248","doi":"10.1038/s41597-023-02902-9","title":"Enhanced U-Pb detrital zircon, Lu-Hf zircon, δ18O zircon, and Sm-Nd whole rock global databases","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geological and Geochemical Analysis","field":"Earth and Planetary Sciences","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"Natural Environment Research Council; Sight Research UK","keywords":"Zircon; Database; Geology; Geochemistry; Computer science","score_opus":0.03478651441941164,"score_gpt":0.25313005438975256,"score_spread":0.21834353997034092,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390782248","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.81398976,0.001262967,0.11454347,0.00023170323,0.00010713199,0.0008967216,0.052147776,0.0012285457,0.015592024],"genre_scores_gemma":[0.73178935,0.00052472745,0.20546293,0.00013617815,0.00004677032,0.00085128844,0.056251977,0.00033191295,0.004604906],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99643075,0.000675916,0.0007009377,0.00058702816,0.0014185539,0.00018677016],"domain_scores_gemma":[0.9874386,0.0016482854,0.0014471372,0.0034297423,0.005822493,0.00021369083],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008359746,0.00042140274,0.0006017324,0.003481359,0.000688302,0.0017108147,0.001380594,0.00042506083,0.002440209],"category_scores_gemma":[0.009272178,0.0002675277,0.00048478163,0.0041652494,0.0003433941,0.0011958805,0.0017529055,0.0003446908,0.0007326702],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015460077,0.0003128436,0.4600458,0.00082803605,0.00070112944,0.00031567246,0.00061076053,0.0066280426,0.195325,0.003915721,0.006617757,0.3231532],"study_design_scores_gemma":[0.00016154109,0.00042956497,0.53681195,0.00014166001,0.0005014287,0.0005557574,0.0007186416,0.008411429,0.37489578,0.0014050876,0.07585007,0.00011706534],"about_ca_topic_score_codex":0.006656043,"about_ca_topic_score_gemma":0.01725906,"teacher_disagreement_score":0.008359746,"about_ca_system_score_codex":0.000640606,"about_ca_system_score_gemma":0.0013463745,"threshold_uncertainty_score":0.04421109},"labels":[],"label_agreement":null},{"id":"W4390815635","doi":"10.1038/s41597-023-02616-y","title":"Plant macrofossil data for 48-0 ka in the USGS North American Packrat Midden Database, version 5.0","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"U.S. Geological Survey; National Oceanic and Atmospheric Administration; U.S. Department of the Interior","keywords":"Macrofossil; Midden; Archaeology; Radiocarbon dating; Geography; Physical geography; Holocene","score_opus":0.08891087125449199,"score_gpt":0.3149814654156725,"score_spread":0.22607059416118053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390815635","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0056271753,0.00014625586,0.00063297333,0.000026851523,0.000016945853,0.000024778028,0.98931867,0.0005886093,0.0036177759],"genre_scores_gemma":[0.008106989,0.00015028693,0.0022042303,0.000034841247,0.000009250757,0.0001009249,0.9874113,0.00016560013,0.0018165678],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996785,0.000023241388,0.00007554637,0.000085066094,0.00009776166,0.00003990604],"domain_scores_gemma":[0.9982881,0.0001813246,0.00045149066,0.00029157544,0.00065431587,0.00013317863],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00062832783,0.00053274876,0.00047200755,0.0054467297,0.00042919867,0.00079014257,0.00075271813,0.00025847668,0.036803193],"category_scores_gemma":[0.0020052572,0.00037099991,0.00022209955,0.0076139793,0.00010753378,0.00080694014,0.00064428925,0.00040414964,0.009656731],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034522,0.000071829505,0.09597194,0.0014511169,0.00017482815,0.00020668449,0.00061460386,0.00088101224,0.0026624484,0.002293311,0.82927954,0.066047385],"study_design_scores_gemma":[0.00006277834,0.000020897862,0.19301726,0.0002603373,0.000080250924,0.0001484672,0.0003392181,0.0004899976,0.0013429181,0.0009920566,0.80321616,0.000029574996],"about_ca_topic_score_codex":0.048544135,"about_ca_topic_score_gemma":0.07296196,"teacher_disagreement_score":0.048544135,"about_ca_system_score_codex":0.00053906994,"about_ca_system_score_gemma":0.0010785218,"threshold_uncertainty_score":0.12311894},"labels":[],"label_agreement":null},{"id":"W4390817161","doi":"10.1038/s41597-023-02689-9","title":"The Pelagic Species Trait Database, an open data resource to support trait-based ocean research","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and fisheries research","field":"Environmental Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Government of Canada; Marine Environmental Observation Prediction and Response Network","keywords":"Pelagic zone; Trait; Food web; Ecology; Biology; Habitat; Predation; Apex predator; Population; Invertebrate; Computer science","score_opus":0.2983281128450758,"score_gpt":0.41775939093007336,"score_spread":0.11943127808499754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390817161","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012301053,0.00025836666,0.0013210912,0.000064218715,0.000025405403,0.000041611864,0.9947805,0.0010712699,0.0012074576],"genre_scores_gemma":[0.0026193338,0.00020662592,0.0038334075,0.00006451305,0.0000101156265,0.00016088589,0.9925184,0.00025872048,0.00032797156],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980286,0.00022755511,0.0006283292,0.00049229735,0.00047759025,0.00014572752],"domain_scores_gemma":[0.9900372,0.0031270247,0.001984459,0.001763738,0.001947891,0.0011396293],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0026110266,0.0014909968,0.0015008014,0.0121413125,0.00076135056,0.0022528514,0.0024619645,0.0015555538,0.018984975],"category_scores_gemma":[0.016202688,0.00065159664,0.00090278586,0.014107444,0.00041356916,0.002638421,0.00437892,0.0015734492,0.01802951],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062067993,0.00013001985,0.042357408,0.00894867,0.00054951577,0.00053643854,0.00058894034,0.0016826516,0.005262182,0.007573778,0.8761247,0.055625066],"study_design_scores_gemma":[0.00019725985,0.00006285862,0.04938746,0.0011930593,0.00019093408,0.00027735924,0.0003404143,0.001991402,0.0019787173,0.005161374,0.9390916,0.00012746333],"about_ca_topic_score_codex":0.010048881,"about_ca_topic_score_gemma":0.019602256,"teacher_disagreement_score":0.99753803,"about_ca_system_score_codex":0.0009515451,"about_ca_system_score_gemma":0.0032348623,"threshold_uncertainty_score":0.06351107},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4391138155","doi":"10.1038/s41597-024-02930-z","title":"Urban water and electricity demand data for understanding climate change impacts on the water-energy nexus","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Water-Energy-Food Nexus Studies","field":"Environmental Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"United Nations University Institute for Water, Environment, and Health","funders":"Purdue Climate Change Research Center, Purdue University; National Socio-Environmental Synthesis Center; Purdue University; National Science Foundation","keywords":"Nexus (standard); Electricity; Water-energy nexus; Climate change; Electricity demand; Natural resource economics; Environmental science; Consumption (sociology); Electricity system; Electricity generation; Economics; Ecology; Computer science; Engineering","score_opus":0.15815554639560336,"score_gpt":0.28292850732302405,"score_spread":0.12477296092742068,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391138155","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.29183906,0.0016770171,0.014405567,0.0039423043,0.00022932884,0.00022515096,0.65321,0.00065248605,0.03381909],"genre_scores_gemma":[0.5622904,0.0012638549,0.02398803,0.00036538695,0.00014219289,0.00042719368,0.40844858,0.00015644907,0.0029180138],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99946624,0.00017364297,0.00005555497,0.00010729313,0.00013807412,0.000059275502],"domain_scores_gemma":[0.99799687,0.0005781812,0.00042661122,0.00037085026,0.00049733365,0.00013012829],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00070338603,0.00040599215,0.00033549467,0.002869863,0.0003884162,0.00094042363,0.0005624959,0.0004416411,0.0057376027],"category_scores_gemma":[0.0036645965,0.00020326914,0.00049485744,0.006389201,0.00024078447,0.0011238968,0.0011488934,0.00068982615,0.0014000495],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019522486,0.0004221859,0.64808685,0.0011627154,0.0004975799,0.0004056294,0.0007157976,0.059205953,0.0018370548,0.027249573,0.19946516,0.06075622],"study_design_scores_gemma":[0.000086568885,0.00007306876,0.42681807,0.0006368417,0.00017211723,0.00013004678,0.004467729,0.0978262,0.004001714,0.022537528,0.44315118,0.00009883762],"about_ca_topic_score_codex":0.056804027,"about_ca_topic_score_gemma":0.10403036,"teacher_disagreement_score":0.056804027,"about_ca_system_score_codex":0.00085058477,"about_ca_system_score_gemma":0.0014582416,"threshold_uncertainty_score":0.11294681},"labels":[],"label_agreement":null},{"id":"W4391229880","doi":"10.1038/s41597-024-02948-3","title":"Global seasonal prediction of fire danger","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Joint Research Centre; European Commission","keywords":"Environmental science; Climatology; Geography; Geology","score_opus":0.019007074035609677,"score_gpt":0.24858858525618993,"score_spread":0.22958151122058026,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391229880","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93717444,0.00059800234,0.0076214084,0.0006580856,0.00012800541,0.000031881824,0.044757996,0.0006603051,0.008369832],"genre_scores_gemma":[0.96457356,0.00029101464,0.0038707412,0.0000480326,0.000033294382,0.000021370963,0.03037619,0.00005690847,0.0007289055],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99992657,0.000010307061,0.000004310482,0.0000278327,0.000019640695,0.000011359901],"domain_scores_gemma":[0.99973315,0.000051944666,0.000054718188,0.000040712617,0.00008386782,0.000035528534],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00042432765,0.00037088816,0.00019068312,0.0006693942,0.00011624533,0.00040290385,0.00021766787,0.0003705199,0.0012168295],"category_scores_gemma":[0.0009369084,0.000115984134,0.00030827575,0.0009290693,0.000117225674,0.00034129326,0.00027582294,0.0003278949,0.00032233386],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033615317,0.00009450337,0.32926723,0.00017992435,0.00025564685,0.00025581967,0.00016942463,0.54975766,0.008469298,0.0027830657,0.02624609,0.08218517],"study_design_scores_gemma":[0.000048163573,0.0000602341,0.4589682,0.00005737129,0.000049444083,0.00007752451,0.00009174711,0.520349,0.0022764537,0.0022414725,0.015738076,0.000042238033],"about_ca_topic_score_codex":0.020923985,"about_ca_topic_score_gemma":0.017351005,"teacher_disagreement_score":0.020923985,"about_ca_system_score_codex":0.00046200032,"about_ca_system_score_gemma":0.00026823775,"threshold_uncertainty_score":0.04160434},"labels":[],"label_agreement":null},{"id":"W4391308957","doi":"10.1038/s41597-023-02788-7","title":"Multidimensional well-being of US households at a fine spatial scale using fused household surveys","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Urban, Neighborhood, and Segregation Studies","field":"Social Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute of Child Health and Human Development; National Science Foundation; RTI International; Canadian Institute for Advanced Research; University of Pennsylvania; U.S. Environmental Protection Agency","keywords":"Microdata (statistics); Survey data collection; Limiting; Scale (ratio); American Community Survey; General Social Survey; Government (linguistics); Consumer Expenditure Survey; Data science; Geography; Business; Environmental economics; Computer science; Public economics; Economics; Environmental health; Statistics; Engineering; Cartography; Medicine","score_opus":0.10187748787511886,"score_gpt":0.3263173078517201,"score_spread":0.22443981997660123,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391308957","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71760833,0.0011416263,0.10547368,0.001281383,0.00011627598,0.00046272148,0.1607857,0.001099647,0.012030657],"genre_scores_gemma":[0.88902617,0.00049380393,0.045058645,0.00023498025,0.00005406231,0.0008597663,0.062659785,0.00013051907,0.0014822591],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99743855,0.0011010391,0.00020262769,0.00061756244,0.0004749173,0.00016532186],"domain_scores_gemma":[0.99559253,0.0011738767,0.0010687327,0.0011852513,0.0007994383,0.00018013154],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035111278,0.00051860785,0.00056751905,0.0039819144,0.0004859558,0.0014392073,0.0008756659,0.0004917179,0.0048868246],"category_scores_gemma":[0.015624363,0.0002946665,0.0012118367,0.0069120717,0.00043293886,0.0011185036,0.0034508808,0.00093773805,0.00095818],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030487616,0.0002354162,0.82250786,0.0006013324,0.001720768,0.00018714966,0.0042536575,0.015325818,0.0012261578,0.0122813145,0.031846225,0.109509416],"study_design_scores_gemma":[0.00003976247,0.00013923593,0.90976083,0.00033572968,0.00029943825,0.00017433794,0.0028573498,0.026597127,0.0014111015,0.020108167,0.038176358,0.00010054414],"about_ca_topic_score_codex":0.0279369,"about_ca_topic_score_gemma":0.024069099,"teacher_disagreement_score":0.0279369,"about_ca_system_score_codex":0.00068258384,"about_ca_system_score_gemma":0.0007416578,"threshold_uncertainty_score":0.05554861},"labels":[],"label_agreement":null},{"id":"W4391351012","doi":"10.1038/s41597-024-02969-y","title":"TrackdAT, an acoustic telemetry metadata dataset to support aquatic animal tracking research","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fish Ecology and Management Studies","field":"Environmental Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Windsor; Dalhousie University","funders":"Great Lakes Fishery Commission","keywords":"Metadata; Computer science; Data science; Stakeholder; Telemetry; Key (lock); Ecology; Abiotic component; Data sharing; Citizen science; World Wide Web; Environmental resource management; Biology; Environmental science","score_opus":0.20699486369023426,"score_gpt":0.42036780401303503,"score_spread":0.21337294032280077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391351012","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019392471,0.00028817586,0.0023461198,0.00016612842,0.0001153586,0.000078819874,0.9914932,0.0012534275,0.0023194018],"genre_scores_gemma":[0.0032804268,0.00024784514,0.0065814336,0.00009860177,0.00004001406,0.000188217,0.9883122,0.00022740487,0.0010238705],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982644,0.0002451061,0.00050259085,0.0004078081,0.00045414345,0.00012605109],"domain_scores_gemma":[0.991486,0.0021910532,0.001912895,0.0016056275,0.0020612387,0.0007433124],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023575053,0.0008901296,0.00067226443,0.010009118,0.0007250047,0.0019357548,0.0016356783,0.0014652279,0.013172198],"category_scores_gemma":[0.012407379,0.00042543255,0.00075379806,0.009266359,0.00041123157,0.0028044493,0.0025986566,0.0010129411,0.012629109],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000492672,0.00017007951,0.042170104,0.005629838,0.00029022098,0.0005068792,0.0010279808,0.0023992711,0.010212263,0.0085589895,0.85684645,0.07169522],"study_design_scores_gemma":[0.00010009413,0.00004884529,0.031200735,0.0007095547,0.000088023626,0.00020607296,0.00047683515,0.0015282333,0.0021331604,0.003312213,0.96010774,0.000088507644],"about_ca_topic_score_codex":0.017433971,"about_ca_topic_score_gemma":0.045581475,"teacher_disagreement_score":0.017433971,"about_ca_system_score_codex":0.0011903668,"about_ca_system_score_gemma":0.003567484,"threshold_uncertainty_score":0.044065416},"labels":[],"label_agreement":null},{"id":"W4391566575","doi":"10.1038/s41597-024-02981-2","title":"Preoperative CT and survival data for patients undergoing resection of colorectal liver metastases","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hepatocellular Carcinoma Treatment and Prognosis","field":"Medicine","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"National Cancer Institute; National Institutes of Health; Memorial Sloan-Kettering Cancer Center; Division of Cancer Epidemiology and Genetics, National Cancer Institute; U.S. Department of Health and Human Services","keywords":"Medicine; Colorectal cancer; Resection; Radiology; Hepatectomy; Overall survival; Curative treatment; Internal medicine; Oncology; Cancer; Surgery; Disease","score_opus":0.15467646260189458,"score_gpt":0.32486564361019016,"score_spread":0.17018918100829558,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391566575","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90347695,0.0003223148,0.0005715417,0.00011736577,0.0000117640375,0.000049294107,0.09395067,0.000078729856,0.0014214301],"genre_scores_gemma":[0.8377554,0.00028280498,0.0014411379,0.000049581657,0.000024825333,0.000112593254,0.15980478,0.000026702315,0.00050217495],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997061,0.00003903772,0.00009381717,0.000073965646,0.00004523836,0.00004177003],"domain_scores_gemma":[0.9968592,0.0008630227,0.0010071886,0.00063969655,0.0003606872,0.00027012857],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036273996,0.00025593516,0.00034677307,0.0020718006,0.00019888367,0.00042054575,0.00020827811,0.0003382539,0.0022658135],"category_scores_gemma":[0.0033608256,0.00015858735,0.0003773637,0.0024784647,0.0002777782,0.0002801458,0.00033694977,0.00036979737,0.00064135616],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00097335706,0.00006829482,0.98358625,0.00008913611,0.0001392979,0.00033993364,0.00008362758,0.0012867253,0.0011037462,0.00013063458,0.004136444,0.00806248],"study_design_scores_gemma":[0.000046533092,0.0001513752,0.9904565,0.000027586648,0.00008181382,0.0017362274,0.0001717917,0.0015770824,0.0007802666,0.00012137732,0.00482572,0.000023828872],"about_ca_topic_score_codex":0.0069171013,"about_ca_topic_score_gemma":0.008932841,"teacher_disagreement_score":0.0069171013,"about_ca_system_score_codex":0.00037686148,"about_ca_system_score_gemma":0.00058385363,"threshold_uncertainty_score":0.013753653},"labels":[],"label_agreement":null},{"id":"W4391606134","doi":"10.1038/s41597-024-03011-x","title":"Chromosome level genome assembly of the Etruscan shrew Suncus etruscus","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"National Institute of General Medical Sciences; Rockefeller University; National Science Foundation; National Institutes of Health; U.S. Department of Health and Human Services; Howard Hughes Medical Institute","keywords":"Biology; Genome; Sequence assembly; Genetics; Genomics; Gene; Chromosome; Reference genome; Genome project; Computational biology; Transcriptome","score_opus":0.05750344690567299,"score_gpt":0.28238233709022753,"score_spread":0.22487889018455454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391606134","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.39464015,0.0046357964,0.10399966,0.00092490733,0.00054440775,0.00087667693,0.4650268,0.010822376,0.018529234],"genre_scores_gemma":[0.14565715,0.0019875693,0.13193417,0.00034474887,0.000065341184,0.0005944686,0.70399135,0.0024584318,0.0129668],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998062,0.000010738753,0.00001371412,0.00009131623,0.00005711917,0.000020993959],"domain_scores_gemma":[0.9997484,0.000025501202,0.00002802044,0.00002342627,0.00014174666,0.000032902175],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00018303678,0.0007141533,0.0006615625,0.0017353467,0.000933489,0.0006937988,0.0005842238,0.00049690716,0.0070790662],"category_scores_gemma":[0.00055497955,0.00033325684,0.0007496304,0.0018662633,0.00017970883,0.00029848787,0.00047262752,0.00093335286,0.0041814754],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009709056,0.00011453562,0.0059208293,0.001617242,0.00017962775,0.000975497,0.0006583393,0.0038100688,0.9020786,0.0018159399,0.02454373,0.057314645],"study_design_scores_gemma":[0.0003623137,0.000506881,0.276258,0.00053980673,0.0010128212,0.0013842728,0.0006085853,0.02505054,0.26144713,0.0028424556,0.42969504,0.000292107],"about_ca_topic_score_codex":0.029172597,"about_ca_topic_score_gemma":0.035231918,"teacher_disagreement_score":0.029172597,"about_ca_system_score_codex":0.0007374219,"about_ca_system_score_gemma":0.0016693136,"threshold_uncertainty_score":0.05800557},"labels":[],"label_agreement":null},{"id":"W4391670666","doi":"10.1038/s41597-024-03739-6","title":"MAPLES-DR: MESSIDOR Anatomical and Pathological Labels for Explainable Screening of Diabetic Retinopathy","year":2024,"lang":"en","type":"preprint","venue":"Scientific Data","topic":"Retinal Imaging and Analysis","field":"Medicine","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Université de Montréal; Hôpital Maisonneuve-Rosemont; Polytechnique Montréal","funders":"Université de Montréal; Diabetes Action Research and Education Foundation","keywords":"Pathological; Diabetic retinopathy; Medicine; Ophthalmology; Pathology; Diabetes mellitus; Endocrinology","score_opus":0.10410165174014922,"score_gpt":0.3696656813168617,"score_spread":0.2655640295767125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391670666","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07161333,0.0027240287,0.46768913,0.0030326287,0.0007040526,0.0010066702,0.24590835,0.19665802,0.0106637115],"genre_scores_gemma":[0.17181464,0.0008598281,0.46871415,0.0008879586,0.0002777982,0.0013949329,0.34328946,0.007560533,0.00520076],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99763346,0.0007254632,0.00015980417,0.00073332555,0.0005707824,0.00017719522],"domain_scores_gemma":[0.9941723,0.002474499,0.00052583526,0.0018666282,0.00070323347,0.00025747446],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037767193,0.0020021447,0.00078882574,0.0032833074,0.0008375081,0.0023328431,0.0026297998,0.0030364448,0.008315805],"category_scores_gemma":[0.017364172,0.000678493,0.0015588698,0.0015479671,0.0008599853,0.0018069089,0.0035078994,0.002043843,0.0050505325],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023915858,0.00045428294,0.019402241,0.0025224034,0.0006005277,0.00076961546,0.00096864626,0.055229925,0.019813504,0.016378004,0.5312249,0.3502443],"study_design_scores_gemma":[0.00089307805,0.0006006316,0.042725507,0.00069971196,0.00029190248,0.0018499035,0.00062263175,0.4573526,0.0492444,0.073287666,0.37206557,0.00036633713],"about_ca_topic_score_codex":0.0061881496,"about_ca_topic_score_gemma":0.010284388,"teacher_disagreement_score":0.008315805,"about_ca_system_score_codex":0.0012462883,"about_ca_system_score_gemma":0.0015967869,"threshold_uncertainty_score":0.027819157},"labels":[],"label_agreement":null},{"id":"W4391685687","doi":"10.1038/s41597-023-02872-y","title":"PLAS-20k: Extended Dataset of Protein-Ligand Affinities from MD Simulations for Machine Learning Applications","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Science and Engineering Research Board; International Institute of Information Technology, Hyderabad; Indian Institute of Technology Delhi; Department of Science and Technology, Ministry of Science and Technology, India","keywords":"Binding affinities; Affinities; Drug discovery; Computer science; Docking (animal); Pipeline (software); Artificial intelligence; Machine learning; Computational biology; Chemistry; Stereochemistry; Biology; Biochemistry","score_opus":0.0871256239288214,"score_gpt":0.3650666403697066,"score_spread":0.2779410164408852,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391685687","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18757418,0.004014333,0.02720374,0.0014748382,0.00069776044,0.00040924584,0.7395621,0.025641076,0.013422776],"genre_scores_gemma":[0.14035559,0.0009277819,0.027290286,0.00046940683,0.00007195094,0.00077084644,0.8271172,0.0015245747,0.0014723238],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992501,0.00019081616,0.000087320856,0.00016990892,0.00021914559,0.00008261225],"domain_scores_gemma":[0.9987948,0.00044672226,0.00011603293,0.00025871993,0.00024448926,0.00013915735],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012726597,0.0022079991,0.0015685428,0.0014457729,0.00086508744,0.0012150268,0.00390818,0.0027758041,0.007827664],"category_scores_gemma":[0.003913065,0.0007388199,0.0018373171,0.0021449418,0.00058712706,0.001399407,0.0013435506,0.0027391112,0.0038921912],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001976291,0.0014362646,0.018109102,0.0060135704,0.0011207,0.00079955603,0.00016981557,0.45317945,0.013051326,0.008473172,0.4485694,0.047101274],"study_design_scores_gemma":[0.0012787583,0.0007200654,0.016274119,0.00029375526,0.00023411996,0.00053672347,0.00016165288,0.76905864,0.016333012,0.0124510825,0.18240535,0.00025270527],"about_ca_topic_score_codex":0.008546884,"about_ca_topic_score_gemma":0.013839379,"teacher_disagreement_score":0.008546884,"about_ca_system_score_codex":0.0011387613,"about_ca_system_score_gemma":0.0022384843,"threshold_uncertainty_score":0.026186168},"labels":[],"label_agreement":null},{"id":"W4391743207","doi":"10.1038/s41597-023-02832-6","title":"The bii4africa dataset of faunal and floral population intactness estimates across Africa’s major land uses","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ecology and Vegetation Dynamics Studies","field":"Environmental Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary; Vancouver Island University","funders":"Natural Environment Research Council; Sight Research UK","keywords":"Geography; Population; Ecology; Biology; Demography; Sociology","score_opus":0.03607999813467442,"score_gpt":0.3171929596664338,"score_spread":0.28111296153175935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391743207","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008840671,0.00022201567,0.0006558901,0.00011566168,0.000017036993,0.0001180704,0.988165,0.00017351337,0.0016920605],"genre_scores_gemma":[0.014232598,0.0001508139,0.0029403835,0.000058330108,0.000013825815,0.0009950676,0.9808719,0.000050120532,0.00068697456],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9988531,0.00024739254,0.0002243342,0.00024131945,0.00027111894,0.0001627318],"domain_scores_gemma":[0.9954465,0.0010480416,0.0012320686,0.00053792284,0.0014181166,0.00031731784],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015997873,0.00095712696,0.00095643103,0.005580388,0.00057325914,0.0010547262,0.0014088576,0.00079788093,0.013943169],"category_scores_gemma":[0.0077012205,0.00042640985,0.00077544345,0.008077874,0.00026156264,0.00076695246,0.0017582957,0.0009856819,0.006699283],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000520889,0.00016663289,0.09822136,0.0067873946,0.00046213256,0.00049779494,0.002047165,0.0022182586,0.0031153397,0.0033517291,0.81721145,0.06539987],"study_design_scores_gemma":[0.00023003027,0.00006509498,0.27656823,0.0012039972,0.00010730424,0.0003353256,0.0013841776,0.0016314561,0.0014057138,0.0014394481,0.7155439,0.00008532991],"about_ca_topic_score_codex":0.032939047,"about_ca_topic_score_gemma":0.044381578,"teacher_disagreement_score":0.032939047,"about_ca_system_score_codex":0.0012255079,"about_ca_system_score_gemma":0.0014410315,"threshold_uncertainty_score":0.06549466},"labels":[],"label_agreement":null},{"id":"W4391778775","doi":"10.1038/s41597-024-02996-9","title":"U.S. cereal rye winter cover crop growth database","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Carbon and Nitrogen Dynamics","field":"Agricultural and Biological Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Natural Resources Conservation Service; National Institute of Food and Agriculture; U.S. Department of Agriculture","keywords":"Cover crop; Secale; Biomass (ecology); Agronomy; Hardiness (plants); Environmental science; Crop; Hordeum vulgare; Biology; Poaceae; Cultivar","score_opus":0.042711798139851403,"score_gpt":0.2603312968689096,"score_spread":0.2176194987290582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391778775","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031909791,0.0001005418,0.00019775353,0.000032470165,0.0000075546423,0.000015792855,0.99521554,0.00026989722,0.0009694966],"genre_scores_gemma":[0.005223759,0.00007759827,0.00059526286,0.000026097056,0.000003993912,0.00005443224,0.99356925,0.000024175617,0.00042542067],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9997236,0.000034069,0.00005444346,0.00009191948,0.000064721564,0.000031198688],"domain_scores_gemma":[0.9987909,0.00021652968,0.00022954546,0.00019688965,0.00046020673,0.00010593985],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004172106,0.00072264456,0.0005856245,0.002359754,0.00023354165,0.00067123753,0.0006899389,0.00044433275,0.012983322],"category_scores_gemma":[0.0017758624,0.00025126824,0.00038128227,0.0055533918,0.00009581655,0.00047262033,0.0005249172,0.00039626364,0.010243975],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033857534,0.00010814172,0.052051146,0.001171706,0.00021486828,0.00017992118,0.00010411588,0.0027297477,0.0012522303,0.0012565284,0.9127098,0.027883139],"study_design_scores_gemma":[0.00041550313,0.00006861473,0.21636972,0.00028763997,0.000101143094,0.00016493634,0.0002229321,0.008465806,0.0017445348,0.0015488967,0.7705461,0.00006416331],"about_ca_topic_score_codex":0.031998143,"about_ca_topic_score_gemma":0.041393027,"teacher_disagreement_score":0.031998143,"about_ca_system_score_codex":0.00055578473,"about_ca_system_score_gemma":0.0008927721,"threshold_uncertainty_score":0.063623786},"labels":[],"label_agreement":null},{"id":"W4391998772","doi":"10.1038/s41597-024-02980-3","title":"Plant trait and vegetation data along a 1314 m elevation gradient with fire history in Puna grasslands, Perú","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ecology and Vegetation Dynamics Studies","field":"Environmental Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Université de Montréal; University of British Columbia","funders":"Senter for Internasjonalisering av Utdanning; Universitetet i Bergen","keywords":"Ecosystem; Vegetation (pathology); Grassland; Biodiversity; Biomass (ecology); Ecology; Trait; Specific leaf area; Growing season; Environmental science; Plant functional type; Climate change; Geography; Threatened species; Biology; Agronomy; Botany","score_opus":0.03511588098095743,"score_gpt":0.24053489864654767,"score_spread":0.20541901766559023,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391998772","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9787578,0.0001687494,0.0002877173,0.0000572034,0.0000020353295,0.000029245211,0.020012397,0.000046011282,0.0006387894],"genre_scores_gemma":[0.9318052,0.00022074352,0.0017630387,0.00006177573,0.000010411242,0.00028603408,0.0650919,0.00002336559,0.0007374082],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996294,0.000093830735,0.00003437989,0.00012487311,0.000059542137,0.00005787463],"domain_scores_gemma":[0.9989477,0.00020442957,0.00034933127,0.00016528109,0.00020276848,0.00013056146],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00048514124,0.00030763776,0.0003296963,0.0023683333,0.0005548457,0.0004954658,0.00042339612,0.0003231132,0.0020844045],"category_scores_gemma":[0.0014359799,0.0002355669,0.00035569185,0.004310922,0.00033559493,0.00038226333,0.0010820203,0.00029339828,0.00059078645],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010044231,0.00006680138,0.9847105,0.0001411376,0.00014170527,0.00023895635,0.0010667411,0.0004391883,0.0024459106,0.0000780892,0.0021004588,0.008470057],"study_design_scores_gemma":[0.000008836325,0.000015560514,0.9974038,0.0000113986325,0.0000180877,0.000097993456,0.00039700963,0.00021490177,0.00009117771,0.000020154284,0.0017151992,0.0000058393957],"about_ca_topic_score_codex":0.037165012,"about_ca_topic_score_gemma":0.051006023,"teacher_disagreement_score":0.037165012,"about_ca_system_score_codex":0.00037224535,"about_ca_system_score_gemma":0.00038770167,"threshold_uncertainty_score":0.07389736},"labels":[],"label_agreement":null},{"id":"W4392044107","doi":"10.1038/s41597-024-02967-0","title":"Nature’s contribution to poverty alleviation, human wellbeing and the SDGs","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Sustainable Development and Environmental Policy","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Svenska Forskningsrådet Formas; Department of Biotechnology, Ministry of Science and Technology, India; Natural Environment Research Council; Research Councils UK; Sight Research UK","keywords":"Poverty; Natural resource economics; Political science; Development economics; Geography; Environmental planning; Economic growth; Economics","score_opus":0.009768203093206035,"score_gpt":0.2580561098142185,"score_spread":0.24828790672101247,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392044107","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3584613,0.0063550454,0.004160775,0.009996693,0.00043223248,0.00012469254,0.56225216,0.00017129208,0.058045834],"genre_scores_gemma":[0.79049486,0.0047826096,0.004857663,0.00091019366,0.00015236065,0.00041882982,0.19143085,0.000117414835,0.0068351524],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99928075,0.0002852444,0.000067187255,0.00009242186,0.00017180345,0.00010268357],"domain_scores_gemma":[0.9982147,0.0005601775,0.00039290654,0.00028740315,0.0003495633,0.00019520173],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011249455,0.0002419877,0.00032016332,0.0019291221,0.0003319305,0.001493566,0.00034716786,0.000312606,0.006637531],"category_scores_gemma":[0.004007026,0.00012371551,0.0004340103,0.0076187295,0.00040461318,0.0010874335,0.0023551346,0.00056728243,0.0006761144],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002920521,0.000091354406,0.5350803,0.0033134324,0.0011877422,0.00038077007,0.004381707,0.010034747,0.0006727106,0.049753867,0.23580119,0.1590102],"study_design_scores_gemma":[0.000026994856,0.000042770578,0.5332992,0.0011814297,0.00017257614,0.00012955295,0.0039973375,0.0016880961,0.00045083213,0.011546037,0.44741502,0.000050075643],"about_ca_topic_score_codex":0.029137338,"about_ca_topic_score_gemma":0.032989018,"teacher_disagreement_score":0.029137338,"about_ca_system_score_codex":0.00085479056,"about_ca_system_score_gemma":0.0012243289,"threshold_uncertainty_score":0.057935476},"labels":[],"label_agreement":null},{"id":"W4392107466","doi":"10.1038/s41597-024-03064-y","title":"A deep learning dataset for sample preparation artefacts detection in multispectral high-content microscopy","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Universität Zürich; Sächsisches Staatsministerium für Wissenschaft und Kunst; Bundesministerium für Bildung und Forschung","keywords":"Multispectral image; Microscopy; Content (measure theory); Computer science; Artificial intelligence; Sample (material); Deep learning; Remote sensing; Chemistry; Chromatography; Geology; Optics; Mathematics; Physics","score_opus":0.03355106710513036,"score_gpt":0.33986287162740697,"score_spread":0.30631180452227663,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392107466","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2432755,0.0030841331,0.08913116,0.0018733895,0.00056595344,0.0009860975,0.63758636,0.016208008,0.007289415],"genre_scores_gemma":[0.13423927,0.00078400545,0.09723917,0.00038763668,0.00006784902,0.0010238562,0.76139647,0.00042483176,0.0044368594],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99911815,0.0001148761,0.00008079291,0.0002468717,0.00032244046,0.00011695798],"domain_scores_gemma":[0.9985689,0.00033063747,0.000175745,0.00040760954,0.00037688392,0.00014020875],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001177685,0.0013289662,0.00065557106,0.0017266021,0.0006677395,0.00087622885,0.0022295392,0.0021366808,0.0026036236],"category_scores_gemma":[0.002257176,0.00036130476,0.0011316049,0.0017845419,0.00068040256,0.00059564854,0.0012171338,0.0015812877,0.002725273],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015817918,0.0030705857,0.02931136,0.0035304124,0.0005532655,0.0019536584,0.00037398597,0.08376948,0.13465586,0.005605434,0.4806905,0.25490364],"study_design_scores_gemma":[0.0005357426,0.0010561589,0.11118871,0.00045335415,0.00024654655,0.0033662866,0.00046747585,0.29667792,0.18426801,0.012862969,0.38848087,0.0003959871],"about_ca_topic_score_codex":0.007582302,"about_ca_topic_score_gemma":0.016467296,"teacher_disagreement_score":0.007582302,"about_ca_system_score_codex":0.0013209525,"about_ca_system_score_gemma":0.0013759539,"threshold_uncertainty_score":0.015076339},"labels":[],"label_agreement":null},{"id":"W4392132169","doi":"10.1038/s41597-024-03080-y","title":"Band-by-band spectral radiative kernels based on the ERA5 reanalysis","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Radiative transfer; Radiation; Spectral bands; Environmental science; Radiative flux; Kernel (algebra); Flux (metallurgy); Broadband; Atmosphere (unit); Remote sensing; Atmospheric sciences; Computational physics; Physics; Meteorology; Mathematics; Optics; Materials science; Geology","score_opus":0.043987350714291835,"score_gpt":0.27181954459097796,"score_spread":0.22783219387668613,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392132169","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.692695,0.00037215377,0.29367468,0.00011339732,0.00014858223,0.00016554534,0.004328454,0.003727352,0.0047748154],"genre_scores_gemma":[0.9412011,0.00013514482,0.054688588,0.00002116655,0.000018400273,0.000064147236,0.0030440616,0.00027205373,0.0005552379],"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","domain_scores_codex":[0.99971515,0.000041513234,0.000020976579,0.0000834041,0.000101282356,0.000037640206],"domain_scores_gemma":[0.9994293,0.00012021066,0.000092099384,0.00015120259,0.00018191645,0.0000252925],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005492056,0.000505778,0.00029973147,0.0007327369,0.00027463346,0.00056442316,0.00063215615,0.00034118004,0.0010567653],"category_scores_gemma":[0.002400544,0.00019865263,0.0008748065,0.00089360535,0.0002748184,0.0010161743,0.00044778024,0.0005863344,0.00031452373],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031381563,0.00024322057,0.032852795,0.00023225737,0.00037599957,0.00014513491,0.0001837018,0.827923,0.047839906,0.009386482,0.0037959765,0.07670772],"study_design_scores_gemma":[0.000020935287,0.00003503033,0.019126082,0.000009462822,0.000058000765,0.00004721328,0.000024642288,0.9643671,0.011976963,0.0018758175,0.0024152724,0.000043566855],"about_ca_topic_score_codex":0.009487022,"about_ca_topic_score_gemma":0.00594056,"teacher_disagreement_score":0.009487022,"about_ca_system_score_codex":0.0004393872,"about_ca_system_score_gemma":0.00054093753,"threshold_uncertainty_score":0.018863618},"labels":[],"label_agreement":null},{"id":"W4392231253","doi":"10.1038/s41597-024-03059-9","title":"GEOWEALTH-US: Spatial wealth inequality data for the United States, 1960–2020","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Spatial and Panel Data Analysis","field":"Economics, Econometrics and Finance","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Microdata (statistics); Inequality; Economic inequality; Survey data collection; Spatial inequality; Disadvantage; Distribution (mathematics); Geography; Imputation (statistics); Social inequality; Demographic economics; Economics; Development economics; Population; Census; Political science; Demography; Missing data; Sociology","score_opus":0.17710370419728813,"score_gpt":0.32480729677186704,"score_spread":0.1477035925745789,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392231253","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0071964134,0.00015700111,0.00027050867,0.00019719733,0.000035762914,0.000025014895,0.988636,0.00012639274,0.0033557182],"genre_scores_gemma":[0.018581452,0.00025338097,0.0007456908,0.00008916149,0.000022593811,0.00015444426,0.9783207,0.000028688659,0.0018039524],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9995963,0.000038512586,0.00008366367,0.000074665506,0.00015001207,0.00005685704],"domain_scores_gemma":[0.998389,0.00013168668,0.00038840427,0.00017524876,0.0007656738,0.0001500418],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00048923184,0.0005116496,0.00045577012,0.0031300131,0.00044302587,0.00085699663,0.00065150164,0.0004291778,0.008270808],"category_scores_gemma":[0.00361724,0.00027132296,0.00028201955,0.008343054,0.00019366913,0.00070581445,0.001107281,0.0007432869,0.0058711492],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011276057,0.000060738286,0.07781341,0.0002521951,0.000070865884,0.000084910855,0.00021004998,0.00087945786,0.00017434926,0.0023185017,0.90054005,0.017482612],"study_design_scores_gemma":[0.00009404944,0.00003816342,0.39799204,0.00036031217,0.00005165786,0.0001348942,0.0006595588,0.0015137388,0.00062349107,0.0013199482,0.59716403,0.000048074846],"about_ca_topic_score_codex":0.16821459,"about_ca_topic_score_gemma":0.15663584,"teacher_disagreement_score":0.16821459,"about_ca_system_score_codex":0.0009559578,"about_ca_system_score_gemma":0.0015187748,"threshold_uncertainty_score":0.33447093},"labels":[],"label_agreement":null},{"id":"W4392864726","doi":"10.1038/s41597-024-03109-2","title":"HYADES - A Global Archive of Annual Maxima Daily Precipitation","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Global Institute for Water Security; University of Saskatchewan; University of Calgary","funders":"Grantová Agentura České Republiky","keywords":"Maxima; Precipitation; Climatology; Environmental science; Climate change; Global warming; Series (stratigraphy); Physical geography; Meteorology; Geography; Geology; History; Oceanography","score_opus":0.03728824735855933,"score_gpt":0.29464886289976727,"score_spread":0.2573606155412079,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392864726","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.049215168,0.0005947719,0.014005035,0.0002295344,0.0001419272,0.00015365232,0.92058176,0.0084206015,0.0066576363],"genre_scores_gemma":[0.05752623,0.0003279103,0.020581339,0.00003632846,0.00006404729,0.00019571147,0.91929567,0.00052829256,0.0014444484],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999647,0.000050349165,0.000040963965,0.00010502177,0.00012466803,0.00003190928],"domain_scores_gemma":[0.9990533,0.00008598293,0.00015156185,0.00030980303,0.00032741073,0.00007193556],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006476991,0.00068528915,0.000538012,0.0020398214,0.00032384932,0.0007650943,0.0010377656,0.00043308764,0.0048054205],"category_scores_gemma":[0.0017595724,0.00028551221,0.000466903,0.0032206452,0.0001903202,0.0008604501,0.00066781486,0.000769865,0.0047729816],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00065202836,0.00024962085,0.08728115,0.001275504,0.00054608635,0.00035304113,0.000592388,0.028510205,0.011891058,0.0048312405,0.70763963,0.15617804],"study_design_scores_gemma":[0.00031114434,0.00007864658,0.2477854,0.00024343011,0.00018108587,0.00040073667,0.00035462744,0.05118261,0.012294279,0.004635109,0.6823753,0.0001576012],"about_ca_topic_score_codex":0.010834728,"about_ca_topic_score_gemma":0.015860517,"teacher_disagreement_score":0.010834728,"about_ca_system_score_codex":0.00027753305,"about_ca_system_score_gemma":0.0010146163,"threshold_uncertainty_score":0.021543324},"labels":[],"label_agreement":null},{"id":"W4393007704","doi":"10.1038/s41597-024-03139-w","title":"The Arctic Plant Aboveground Biomass Synthesis Dataset","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate change and permafrost","field":"Earth and Planetary Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; Université Laval; Queen's University; Alberta Biodiversity Monitoring Institute; Nova Scotia Community College; Université du Québec à Trois-Rivières; Carleton University","funders":"Biological and Environmental Research; Division of Environmental Biology; Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Office of Science; Societas pro Fauna et Flora Fennica; Nordenskiöld-samfundet; Academy of Finland; Queen's University; U.S. Department of Energy; European Commission; Office of Polar Programs; Polar Knowledge Canada; Nuclear Safety and Security Commission; Université Laval; ArcticNet; Natural Resources Canada; National Aeronautics and Space Administration; Vetenskapsrådet; National Science Foundation","keywords":"Biomass (ecology); Environmental science; The arctic; Arctic; Ecology; Oceanography; Biology; Geology","score_opus":0.10829667691575451,"score_gpt":0.28661073285618044,"score_spread":0.17831405594042593,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393007704","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0059407484,0.00020449085,0.0003559352,0.000044253655,0.000021022299,0.000025680387,0.99213964,0.00039819934,0.0008701613],"genre_scores_gemma":[0.0036513866,0.00005650041,0.000860511,0.000022095915,0.000006467276,0.00006896717,0.9950245,0.00002856202,0.00028090063],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999141,0.0001728189,0.0001009685,0.00030109638,0.00017710775,0.000106942396],"domain_scores_gemma":[0.9985494,0.00031702034,0.0001794345,0.00026827867,0.00054185645,0.00014401505],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009382988,0.0011456945,0.0007745539,0.0028219793,0.00076042564,0.0011062382,0.001194767,0.0008806482,0.007815016],"category_scores_gemma":[0.0022662415,0.0003848585,0.0011091549,0.004072042,0.00026109722,0.0005180402,0.0012728337,0.0007773804,0.009585398],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006628459,0.00020475892,0.053205643,0.0024310194,0.0006633436,0.0003719594,0.000301315,0.009614703,0.0042338097,0.0022122802,0.89569,0.030408321],"study_design_scores_gemma":[0.00038889123,0.00008001144,0.106850475,0.0003936405,0.00016975132,0.00022619883,0.00036256772,0.0054446612,0.002156316,0.0012774513,0.88256025,0.00008985468],"about_ca_topic_score_codex":0.04127043,"about_ca_topic_score_gemma":0.076171234,"teacher_disagreement_score":0.04127043,"about_ca_system_score_codex":0.0009053762,"about_ca_system_score_gemma":0.0014919654,"threshold_uncertainty_score":0.08206046},"labels":[],"label_agreement":null},{"id":"W4393054394","doi":"10.1038/s41597-024-03116-3","title":"A comprehensive historical and geolocalized database of mining activities in Canada","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Mining and Resource Management","field":"Engineering","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Lunds Universitet","keywords":"Database; Geographic coordinate system; Computer science; Matching (statistics); Principal (computer security); Credibility; Standardization; Hierarchy; Geography; Selection (genetic algorithm); Data mining; Information retrieval; Cartography; Statistics; Political science; Mathematics","score_opus":0.035226128438713304,"score_gpt":0.228593163905945,"score_spread":0.1933670354672317,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393054394","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022302473,0.0014122593,0.0029579776,0.0002517706,0.00003930674,0.00017390944,0.94999266,0.0008849505,0.021984728],"genre_scores_gemma":[0.089692116,0.0040289536,0.013542182,0.00014021035,0.000030785974,0.00030441006,0.87553686,0.00025619686,0.016468378],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984011,0.000049462386,0.00016761977,0.00020102081,0.0009183059,0.00026247828],"domain_scores_gemma":[0.9918824,0.00039605747,0.0005681602,0.00040101373,0.00607891,0.0006734537],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00061226427,0.00084472325,0.0008021867,0.019056559,0.0028183642,0.0029881273,0.0018759595,0.00045284908,0.013113643],"category_scores_gemma":[0.0049627796,0.0004255583,0.0004952146,0.052566778,0.0005281644,0.0010732813,0.0013927275,0.0005722334,0.0045611877],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041686674,0.00012422421,0.1299055,0.0026978853,0.00020897217,0.00075699645,0.0028008537,0.0078205215,0.0026395994,0.015442168,0.5784869,0.25869948],"study_design_scores_gemma":[0.000024312776,0.000020418716,0.19868052,0.00059014134,0.00009614853,0.00026774564,0.0020466764,0.003438717,0.0015908559,0.0011945965,0.79191,0.0001399225],"about_ca_topic_score_codex":0.98809993,"about_ca_topic_score_gemma":0.99221885,"teacher_disagreement_score":0.025945485,"about_ca_system_score_codex":0.025945485,"about_ca_system_score_gemma":0.08395255,"threshold_uncertainty_score":0.18824857},"labels":[],"label_agreement":null},{"id":"W4393258334","doi":"10.1038/s41597-024-03142-1","title":"Genome assembly of a symbiotic balantidia (Balantidium ctenopharyngodoni) in fish hindgut","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium","funders":"National Key Research and Development Program of China; Earmarked Fund for China Agriculture Research System; China Postdoctoral Science Foundation; National Natural Science Foundation of China","keywords":"Hindgut; Biology; Genome; Gene; Comparative genomics; Synteny; Genetics; Genomics; Evolutionary biology; Botany; Midgut","score_opus":0.024093684600011166,"score_gpt":0.2647571905508476,"score_spread":0.24066350595083646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393258334","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91236186,0.0028323468,0.022810474,0.00030967518,0.00010760576,0.00010750637,0.05680523,0.0011947624,0.003470522],"genre_scores_gemma":[0.69666064,0.0021079208,0.09160169,0.00020426509,0.000037523172,0.0002534026,0.20340802,0.0006647039,0.0050618574],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99987996,0.000009894044,0.000007644241,0.00006426513,0.000023243456,0.0000149691805],"domain_scores_gemma":[0.99987674,0.000024108009,0.00002749747,0.000012475241,0.000034854696,0.000024326397],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001603457,0.00045999358,0.00039372474,0.0010584757,0.0007320773,0.00050099683,0.00024454133,0.0003556647,0.0014341105],"category_scores_gemma":[0.00038004626,0.0003068966,0.0006301804,0.0010617216,0.00012590579,0.00030249686,0.00041495764,0.00047370212,0.000638801],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035718692,0.000046619047,0.015133425,0.0008370156,0.00009643611,0.0003862432,0.000476358,0.0024698158,0.9562709,0.00039005128,0.0019114673,0.021624595],"study_design_scores_gemma":[0.000088135195,0.00056531036,0.68841493,0.00041750187,0.0006828384,0.0016718638,0.0013697117,0.03856217,0.18894704,0.0015260228,0.07761328,0.00014125138],"about_ca_topic_score_codex":0.005655576,"about_ca_topic_score_gemma":0.010569946,"teacher_disagreement_score":0.005655576,"about_ca_system_score_codex":0.00036798874,"about_ca_system_score_gemma":0.00069529837,"threshold_uncertainty_score":0.01124531},"labels":[],"label_agreement":null},{"id":"W4394569876","doi":"10.1038/s41597-024-03197-0","title":"A large normative connectome for exploring the tractographic correlates of focal brain interventions","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Krembil Foundation; University of Toronto; University Health Network","funders":"National Institute of Neurological Disorders and Stroke; National Institute of Mental Health; Deutsche Forschungsgemeinschaft; National Institutes of Health; Government of Canada; Canadian Institutes of Health Research; EU Joint Programme – Neurodegenerative Disease Research; New Venture Fund; Deutsches Zentrum für Luft- und Raumfahrt; Foundation for the National Institutes of Health","keywords":"Connectome; Connectomics; Human Connectome Project; Diffusion MRI; Neuroimaging; Computer science; White matter; Context (archaeology); Resting state fMRI; Neuroscience; Psychology; Artificial intelligence; Functional connectivity; Medicine; Magnetic resonance imaging; Biology; Radiology","score_opus":0.2653553452847806,"score_gpt":0.3686404338187473,"score_spread":0.10328508853396667,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394569876","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4818488,0.0044340524,0.22044252,0.0021151714,0.00046631106,0.0009966827,0.2667961,0.0057865647,0.017113809],"genre_scores_gemma":[0.5382656,0.0027735492,0.15998559,0.00069930125,0.0003130347,0.0048747803,0.28628474,0.0021323427,0.004671192],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9993148,0.00015420286,0.00006950441,0.00024103919,0.0001662648,0.00005417409],"domain_scores_gemma":[0.99605453,0.0011571972,0.0005595781,0.0009373923,0.0010647025,0.00022650372],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017939082,0.00059844914,0.0005968818,0.0042782305,0.0009535485,0.00088992866,0.000724738,0.0009808242,0.008596365],"category_scores_gemma":[0.0116230305,0.000310878,0.0006775538,0.0035563412,0.00066867465,0.0012265812,0.0016350953,0.00074699114,0.0027359738],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014352287,0.00047543683,0.13538577,0.0042705135,0.0015134243,0.0042831637,0.0039285463,0.022282632,0.19572829,0.057891227,0.22243348,0.3503722],"study_design_scores_gemma":[0.00026626376,0.00050638017,0.5123848,0.0013574092,0.00087273814,0.011637033,0.0012757877,0.03690864,0.033245288,0.0907453,0.31039345,0.00040692635],"about_ca_topic_score_codex":0.0035458717,"about_ca_topic_score_gemma":0.008550291,"teacher_disagreement_score":0.008596365,"about_ca_system_score_codex":0.0005255543,"about_ca_system_score_gemma":0.0013668285,"threshold_uncertainty_score":0.028757691},"labels":[],"label_agreement":null},{"id":"W4394570919","doi":"10.1038/s41597-024-03141-2","title":"A monthly gridded burned area database of national wildland fire data","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Natural Resources Canada; Canadian Forest Service","funders":"Fondo de Financiamiento de Centros de Investigación en Áreas Prioritarias; Agencia Estatal de Investigación; Fundação para a Ciência e a Tecnologia; Natural Resources Canada; Horizon 2020 Framework Programme; New South Wales Government; Ministerio de Educación, Cultura y Deporte; National Fisheries Development Board; Ministerio de Ciencia, Innovación y Universidades; Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Ministério da Ciência, Tecnologia e Ensino Superior; European Regional Development Fund; European Commission; Agencia Nacional de Investigación y Desarrollo; European Space Agency","keywords":"Database; Government (linguistics); Resource (disambiguation); Data source; Grid cell; Metadata; Grid; Geography; Environmental science; Computer science; Environmental resource management; Meteorology; World Wide Web","score_opus":0.05093371852271682,"score_gpt":0.2793830591394934,"score_spread":0.22844934061677657,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394570919","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017215868,0.00015823304,0.0028302602,0.000096395415,0.00007297886,0.00016985822,0.9738152,0.0004919109,0.0051492634],"genre_scores_gemma":[0.015015985,0.00011611534,0.0078656105,0.000052034793,0.000022805796,0.00038892656,0.97489923,0.00013569459,0.0015037006],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987721,0.0001337925,0.00025805196,0.00031725652,0.00039514704,0.0001236634],"domain_scores_gemma":[0.9968881,0.00041028886,0.0005617694,0.0006241106,0.001252634,0.00026308923],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012922075,0.0004886392,0.0006725316,0.0039720265,0.00046808284,0.0009863221,0.0010034005,0.0004675934,0.009039643],"category_scores_gemma":[0.004136967,0.00028279336,0.0005724514,0.007126936,0.0001944533,0.0007787443,0.00091659534,0.000983909,0.0070006857],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005108145,0.00046123605,0.12488743,0.0015016708,0.00042369173,0.0003561373,0.0008843321,0.008423081,0.003813055,0.006424821,0.7661711,0.08614272],"study_design_scores_gemma":[0.00014756151,0.00007339131,0.26330543,0.00038602634,0.000111891306,0.00024183464,0.00066794275,0.006617567,0.0030744637,0.002259464,0.7230189,0.00009551732],"about_ca_topic_score_codex":0.035341714,"about_ca_topic_score_gemma":0.05391243,"teacher_disagreement_score":0.035341714,"about_ca_system_score_codex":0.00093655224,"about_ca_system_score_gemma":0.0024427578,"threshold_uncertainty_score":0.07027197},"labels":[],"label_agreement":null},{"id":"W4394728517","doi":"10.1038/s41597-024-03165-8","title":"High-resolution freshwater dissolved calcium and pH data layers for Canada and the United States","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and coastal ecosystems","field":"Earth and Planetary Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Fisheries and Oceans Canada","funders":"Fisheries and Oceans Canada; Crown-Indigenous Relations and Northern Affairs Canada; Government of Canada; Ministry of Natural Resources","keywords":"Calcium; Oceanography; Environmental science; Resolution (logic); Environmental chemistry; Chemistry; Mineralogy; Geology; Computer science","score_opus":0.036334851093883944,"score_gpt":0.23434452646199344,"score_spread":0.19800967536810948,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394728517","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05443055,0.00030480316,0.004193315,0.00040119188,0.00005567191,0.0001931313,0.91501683,0.0025153796,0.022889197],"genre_scores_gemma":[0.15880702,0.0006017016,0.031500828,0.00015822722,0.000012299214,0.00024672694,0.79764354,0.0003925848,0.010637147],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99966156,0.000010789981,0.000023917853,0.00004544459,0.00018529397,0.00007290364],"domain_scores_gemma":[0.99793124,0.00005129277,0.000080984806,0.00007184209,0.0017510286,0.000113634764],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00033543154,0.0004562252,0.0002885022,0.0026638906,0.0013934213,0.001273796,0.0008554571,0.00031983995,0.008963672],"category_scores_gemma":[0.0016733811,0.00026379206,0.00041849999,0.0066772066,0.00018657428,0.00050602516,0.00051702675,0.0005975452,0.002308625],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035839787,0.00019168278,0.23703068,0.00053869304,0.0002725697,0.00034553444,0.001003961,0.0173356,0.003660134,0.0057084374,0.5790895,0.15446486],"study_design_scores_gemma":[0.00012350784,0.00001569832,0.56145906,0.0002785847,0.00008721053,0.00009499886,0.0015589199,0.019968638,0.004008787,0.0010929506,0.41116846,0.00014333543],"about_ca_topic_score_codex":0.9893036,"about_ca_topic_score_gemma":0.993616,"teacher_disagreement_score":0.010696411,"about_ca_system_score_codex":0.0103111565,"about_ca_system_score_gemma":0.020324253,"threshold_uncertainty_score":0.07481301},"labels":[],"label_agreement":null},{"id":"W4394891764","doi":"10.1038/s41597-024-03235-x","title":"Towards Gender Harmony Dataset: Gender Beliefs and Gender Stereotypes in 62 Countries","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Gender Roles and Identity Studies","field":"Social Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Femininity; Harmony (color); Masculinity; Perception; Psychology; Social psychology; Qualitative property; Gender studies; Sociology","score_opus":0.1683779190793405,"score_gpt":0.38049659109328043,"score_spread":0.21211867201393994,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394891764","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0152649125,0.00013378952,0.00028252264,0.0002274372,0.000045332963,0.00011055013,0.9806192,0.00017267166,0.0031435895],"genre_scores_gemma":[0.017777072,0.00013450184,0.0014706095,0.00015608051,0.000029591165,0.00079407525,0.9776432,0.00006311758,0.001931809],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99890137,0.0002754724,0.0001474288,0.00018473982,0.00029595278,0.00019506589],"domain_scores_gemma":[0.9963529,0.00083361205,0.00066889887,0.00061519677,0.0010378476,0.0004914468],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015315558,0.0005208225,0.00057145185,0.003428299,0.00086553383,0.0015407915,0.0011064184,0.0011567781,0.015045766],"category_scores_gemma":[0.005488923,0.00031270823,0.0005335583,0.0055831783,0.00036420047,0.0009803001,0.0030171862,0.0010305282,0.009410674],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023997352,0.00021361606,0.07841271,0.0016346859,0.00012287933,0.0002252198,0.0023275595,0.0005764032,0.0007709098,0.00316348,0.8840486,0.028263891],"study_design_scores_gemma":[0.00015791973,0.000062908744,0.24468057,0.00061215134,0.000047075962,0.00016390465,0.004710159,0.00047520106,0.00069854624,0.0011041119,0.7472225,0.00006495739],"about_ca_topic_score_codex":0.020767566,"about_ca_topic_score_gemma":0.027522683,"teacher_disagreement_score":0.020767566,"about_ca_system_score_codex":0.0010218263,"about_ca_system_score_gemma":0.0015912564,"threshold_uncertainty_score":0.050333083},"labels":[],"label_agreement":null},{"id":"W4394987905","doi":"10.1038/s41597-024-03238-8","title":"Chromosome-level genome assembly of marine diatom Skeletonema tropicum","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Chinese Academy of Sciences; Institute of Oceanology, Chinese Academy of Sciences; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China","keywords":"Genome; Biology; Gene; Polyploid; Genetics; Evolutionary biology; Diatom; Computational biology; Ecology","score_opus":0.038496672889657385,"score_gpt":0.2769623112537986,"score_spread":0.23846563836414122,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394987905","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8189853,0.0045406627,0.07466539,0.00049076095,0.0001668004,0.00044421136,0.092902996,0.0018418796,0.00596214],"genre_scores_gemma":[0.53422666,0.0038644238,0.1796898,0.0001867842,0.00004076821,0.0005382479,0.27338496,0.00091832894,0.0071500437],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998429,0.000013197112,0.000014064243,0.000073391406,0.000036609697,0.000019905161],"domain_scores_gemma":[0.99979824,0.00003976398,0.000032559823,0.000022565106,0.0000731546,0.000033708526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019279782,0.0005081752,0.0006195454,0.0011455745,0.0004973173,0.00061311125,0.00040695912,0.00044090016,0.0015492071],"category_scores_gemma":[0.0005663311,0.00038284407,0.00074947573,0.0014252261,0.00014156042,0.0003306646,0.0004948711,0.00072380767,0.0012631393],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003270997,0.00003887346,0.004229323,0.00046575716,0.000056430377,0.00025342926,0.00026408426,0.0018722538,0.9747229,0.00032788786,0.0011488157,0.016293082],"study_design_scores_gemma":[0.00024096794,0.00078056904,0.30899802,0.00029288806,0.0006967411,0.0013842608,0.0007553585,0.04533688,0.5408882,0.0014590343,0.09895511,0.00021200266],"about_ca_topic_score_codex":0.007518622,"about_ca_topic_score_gemma":0.011437356,"teacher_disagreement_score":0.007518622,"about_ca_system_score_codex":0.0005108081,"about_ca_system_score_gemma":0.00070565264,"threshold_uncertainty_score":0.014949739},"labels":[],"label_agreement":null},{"id":"W4395038839","doi":"10.1038/s41597-024-03190-7","title":"RASMI: Global ranges of building material intensities differentiated by region, structure, and function","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Environmental Impact and Sustainability","field":"Environmental Science","cited_by":55,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"HORIZON EUROPE Framework Programme; International Institute for Applied Systems Analysis; Israel Science Foundation; Ministry of Economy, Trade and Industry; Canada Research Chairs; European Commission","keywords":"Demolition; Resource (disambiguation); Metric (unit); Environmental science; Function (biology); Computer science; Environmental resource management; Civil engineering; Engineering; Operations management","score_opus":0.014555602065107615,"score_gpt":0.24681710996722409,"score_spread":0.23226150790211647,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395038839","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10576997,0.0004930549,0.01183474,0.00020733073,0.00007044855,0.00014078972,0.86748475,0.003365382,0.010633496],"genre_scores_gemma":[0.11517321,0.000281483,0.018746404,0.000062976906,0.000030252377,0.0003781904,0.86376595,0.00030319038,0.0012582823],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994035,0.000086823035,0.000064440355,0.00018857418,0.00017878925,0.00007780971],"domain_scores_gemma":[0.9989925,0.00013806211,0.00021675935,0.00027481394,0.00028767806,0.000090207795],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046810534,0.0009117481,0.00046456518,0.0039907577,0.00031268678,0.00088957767,0.0009458065,0.0006558654,0.0039373836],"category_scores_gemma":[0.0021815463,0.00031657482,0.00090979727,0.005584309,0.00035863026,0.0009350544,0.0013450711,0.000704604,0.0037915623],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006974458,0.00039529862,0.28722695,0.002744372,0.00059253647,0.0004884085,0.0014596441,0.05462614,0.014130016,0.009631134,0.4662368,0.16177128],"study_design_scores_gemma":[0.00012235678,0.00014418637,0.4570746,0.00028613367,0.0001060975,0.0005010451,0.0013890152,0.029269291,0.008236698,0.0044472315,0.49826592,0.00015736173],"about_ca_topic_score_codex":0.017164977,"about_ca_topic_score_gemma":0.028870163,"teacher_disagreement_score":0.017164977,"about_ca_system_score_codex":0.00058132235,"about_ca_system_score_gemma":0.00064702815,"threshold_uncertainty_score":0.034130156},"labels":[],"label_agreement":null},{"id":"W4395081664","doi":"10.1038/s41597-024-03265-5","title":"Author Correction: Neurocognitive aging data release with behavioral, structural and multi-echo functional MRI measures","year":2024,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Brain Tumor Detection and Classification","field":"Neuroscience","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; McGill University; Douglas Mental Health University Institute; McGill Genome Centre; University of Toronto; Toronto Metropolitan University; Montreal Neurological Institute and Hospital","funders":"","keywords":"Neurocognitive; Echo (communications protocol); Neuroscience; Psychology; Computer science; Cognition; Computer security","score_opus":0.21425257817395657,"score_gpt":0.35676817750834694,"score_spread":0.14251559933439037,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395081664","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00072926213,0.0008141902,0.009218311,0.049214553,0.90080684,0.00014682375,0.022653384,0.004507003,0.011909612],"genre_scores_gemma":[0.037156317,0.0063320817,0.069670446,0.06558704,0.14063002,0.0014107873,0.063013956,0.02515376,0.59104556],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98882896,0.0014982689,0.0026217764,0.0014757883,0.0051740385,0.00040110864],"domain_scores_gemma":[0.8852497,0.024393348,0.0048826197,0.012493761,0.07122982,0.0017507985],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0090993,0.0021347357,0.0016407969,0.0060053887,0.0036101202,0.004286464,0.0035141208,0.0035313363,0.1343595],"category_scores_gemma":[0.16943638,0.0012883125,0.0013538883,0.004921777,0.0020080348,0.0026911357,0.0035001473,0.007962515,0.09393219],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022032827,0.000003995415,0.00010340499,0.00008174518,0.000006945623,0.00011597941,0.00004798142,0.00002994776,0.000042003478,0.0006576411,0.99230474,0.006583568],"study_design_scores_gemma":[0.000024858997,0.000009411685,0.00053254183,0.00041285725,0.000027587452,0.0005117028,0.000113596834,0.0002212896,0.0004954774,0.002263841,0.99535155,0.00003530811],"about_ca_topic_score_codex":0.01495391,"about_ca_topic_score_gemma":0.02268147,"teacher_disagreement_score":0.1343595,"about_ca_system_score_codex":0.003028552,"about_ca_system_score_gemma":0.007431727,"threshold_uncertainty_score":0.44947726},"labels":[],"label_agreement":null},{"id":"W4395081723","doi":"10.1038/s41597-024-03230-2","title":"HiMIC-Monthly: A 1 km high-resolution atmospheric moisture index collection over China, 2003–2020","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; University of Toronto","funders":"Natural Science Foundation of Guangdong Province; National Natural Science Foundation of China","keywords":"Environmental science; Land cover; Water content; Moisture; Climatology; Atmospheric sciences; Scale (ratio); Meteorology; Geography; Land use; Cartography; Geology","score_opus":0.015646973276605015,"score_gpt":0.24264317704420568,"score_spread":0.22699620376760066,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395081723","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5878934,0.0004046823,0.0030681223,0.00023150878,0.00009446112,0.00017177132,0.40458453,0.0015178968,0.0020335533],"genre_scores_gemma":[0.45104903,0.00019532075,0.006602246,0.00008533753,0.000071339586,0.00020985396,0.5404515,0.00008461798,0.0012506543],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99971074,0.000034934594,0.000027191216,0.00008740127,0.00008611003,0.000053507076],"domain_scores_gemma":[0.9993249,0.000036289155,0.00009600637,0.00014440577,0.0002776225,0.000120879326],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007349076,0.00077072345,0.00054899626,0.0016813084,0.00037684527,0.00042789525,0.0008390169,0.00039987758,0.0014806053],"category_scores_gemma":[0.0007588942,0.00021602644,0.00045906036,0.0026997214,0.00026655436,0.00040367295,0.000552938,0.00034461264,0.00080473244],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010400561,0.00069747824,0.6418625,0.00078506884,0.0007457541,0.0007265514,0.00041691388,0.05967582,0.018775579,0.0010103481,0.17751285,0.096751],"study_design_scores_gemma":[0.00013288131,0.000063404426,0.91366154,0.000045656914,0.00007250957,0.00007993864,0.00018503734,0.05120519,0.0047112578,0.00019836487,0.029580379,0.000063804895],"about_ca_topic_score_codex":0.11418121,"about_ca_topic_score_gemma":0.12473648,"teacher_disagreement_score":0.11418121,"about_ca_system_score_codex":0.0008891645,"about_ca_system_score_gemma":0.0015739714,"threshold_uncertainty_score":0.2270332},"labels":[],"label_agreement":null},{"id":"W4395112800","doi":"10.1038/s41597-024-03275-3","title":"A new commercial boundary dataset for metropolitan areas in the USA and Canada, built from open data","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Human Mobility and Location-Based Analysis","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"University of Toronto","keywords":"Metropolitan area; Boundary (topology); Geography; Regional science; Cluster analysis; Data collection; Data science; Segmentation; Economic geography; Cartography; Computer science; Statistics","score_opus":0.16116477062218248,"score_gpt":0.41821989254241393,"score_spread":0.25705512192023144,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395112800","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023251802,0.00028358636,0.0019564966,0.00015474949,0.000028578042,0.000111661044,0.96624094,0.0011008632,0.006871298],"genre_scores_gemma":[0.02741274,0.00016408094,0.0048436895,0.000044914894,0.000006866599,0.00017774249,0.96588844,0.00014038937,0.001321139],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992149,0.000040663504,0.000059364786,0.00019092155,0.00031532024,0.00017889126],"domain_scores_gemma":[0.9974396,0.00017573853,0.00018593296,0.00027335456,0.0015808566,0.00034442244],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003685466,0.0007961467,0.0007767107,0.0064626536,0.002146125,0.0018030267,0.0019197854,0.00073265244,0.004966723],"category_scores_gemma":[0.0038079093,0.0003741099,0.00083949347,0.015612809,0.00061038404,0.001066199,0.002253637,0.0010661285,0.00332236],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002960811,0.00017439888,0.10998959,0.0010481371,0.00022857326,0.0005255889,0.0018802074,0.008077192,0.0014510459,0.0069818757,0.81138927,0.05795796],"study_design_scores_gemma":[0.00012359452,0.000024680177,0.30787507,0.00050173356,0.00009484431,0.00032812895,0.004557607,0.016608404,0.0019298805,0.0033464392,0.66444474,0.00016481806],"about_ca_topic_score_codex":0.9607536,"about_ca_topic_score_gemma":0.98037726,"teacher_disagreement_score":0.03924638,"about_ca_system_score_codex":0.009385867,"about_ca_system_score_gemma":0.018953258,"threshold_uncertainty_score":0.078954995},"labels":[],"label_agreement":null},{"id":"W4395112828","doi":"10.1038/s41597-024-03262-8","title":"Chromosome-level genome assembly of the yellow-cheek carp Elopichthys bambusa","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"","keywords":"Genome; Carp; Biology; Chromosome; Sequence assembly; Genome size; Gene; Genetics; Evolutionary biology; Fish <Actinopterygii>; Transcriptome; Fishery","score_opus":0.04989781057585496,"score_gpt":0.273890589885207,"score_spread":0.22399277930935202,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395112828","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7755482,0.0035866094,0.06284157,0.00048968487,0.00021477297,0.00039195683,0.14296992,0.0024096724,0.011547646],"genre_scores_gemma":[0.53461415,0.0027201453,0.13056003,0.00018863496,0.000051817504,0.0005715478,0.3157409,0.0012959271,0.014256864],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998783,0.000007616031,0.000007636629,0.00005944342,0.000030126412,0.00001682578],"domain_scores_gemma":[0.99986005,0.000021583166,0.0000262284,0.000013872447,0.000049659488,0.000028581298],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001799518,0.0005155705,0.0004945537,0.0014581195,0.00062672916,0.00061460404,0.00036726572,0.0003833806,0.0036588886],"category_scores_gemma":[0.00047294755,0.00030282614,0.00076207495,0.0014442517,0.00016665913,0.00031365472,0.00045320106,0.0006857217,0.0018407204],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008167619,0.00005651044,0.008314351,0.0010269913,0.00009884163,0.00076934596,0.00054199604,0.0037725258,0.9479798,0.0009412059,0.0032043033,0.0324773],"study_design_scores_gemma":[0.00030911702,0.0007949172,0.49519616,0.00070129463,0.0008859743,0.0023736113,0.0014437605,0.036443736,0.29427254,0.0026635574,0.16457562,0.00033968713],"about_ca_topic_score_codex":0.01126345,"about_ca_topic_score_gemma":0.0152004,"teacher_disagreement_score":0.01126345,"about_ca_system_score_codex":0.00053703494,"about_ca_system_score_gemma":0.001122878,"threshold_uncertainty_score":0.02239579},"labels":[],"label_agreement":null},{"id":"W4396618987","doi":"10.1038/s41597-024-03258-4","title":"A regionally refined quarter-degree global atmospheric rivers database based on ERA5","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of California, San Diego; Jet Propulsion Laboratory; Nuclear Safety and Security Commission; National Aeronautics and Space Administration; California Institute of Technology; Department of Water Resources; Environment and Climate Change Canada","keywords":"Thresholding; Environmental science; Database; Polar; Block (permutation group theory); Computer science; Meteorology; Water cycle; Water vapor; Tracking (education); Degree (music); Geography; Mathematics; Artificial intelligence; Image (mathematics)","score_opus":0.0682208969313054,"score_gpt":0.2872415749504418,"score_spread":0.21902067801913638,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396618987","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.30253646,0.0007923492,0.038802434,0.00032949468,0.00018728952,0.0002584265,0.63222075,0.01258529,0.0122875795],"genre_scores_gemma":[0.294277,0.0003372209,0.06312885,0.00010251735,0.000054784377,0.00017478314,0.63960564,0.00061125105,0.001707883],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99952674,0.0000507607,0.000090213965,0.00017650677,0.00011889939,0.0000369429],"domain_scores_gemma":[0.9987066,0.0001249939,0.00014944567,0.00047674112,0.000480194,0.00006211655],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007259674,0.00042078496,0.00054952974,0.0016799018,0.00036187697,0.00097060186,0.0010873745,0.00056946697,0.0039291],"category_scores_gemma":[0.0016712267,0.00028178302,0.0008349394,0.0030137314,0.00014681478,0.0010235528,0.00057958794,0.0005420142,0.0018282823],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012643792,0.0005221039,0.26251167,0.001526084,0.0014100831,0.0007546721,0.0005798806,0.24303393,0.046109755,0.007616022,0.16298959,0.27168187],"study_design_scores_gemma":[0.00042655363,0.00018386303,0.49597812,0.00013815073,0.00051985634,0.00036124868,0.0004968029,0.22420768,0.02226373,0.0024185874,0.25270635,0.000298969],"about_ca_topic_score_codex":0.048578296,"about_ca_topic_score_gemma":0.048095886,"teacher_disagreement_score":0.048578296,"about_ca_system_score_codex":0.00041322256,"about_ca_system_score_gemma":0.0012651947,"threshold_uncertainty_score":0.096591055},"labels":[],"label_agreement":null},{"id":"W4396624749","doi":"10.1038/s41597-024-03282-4","title":"High resolution Tibetan Plateau regional reanalysis 1961-present","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cryospheric studies and observations","field":"Earth and Planetary Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Education and Child Care","funders":"National Key Research and Development Program of China; Jiangsu Collaborative Innovation Center for Climate Change; National Natural Science Foundation of China","keywords":"Plateau (mathematics); Climatology; Precipitation; Environmental science; Weather Research and Forecasting Model; Glacier; Permafrost; Global warming; Initialization; Climate change; Atmospheric sciences; Physical geography; Meteorology; Geology; Geography","score_opus":0.08346350337635322,"score_gpt":0.26992819772547094,"score_spread":0.18646469434911772,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396624749","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6572127,0.002497361,0.01360479,0.0006167391,0.0007224882,0.00019357417,0.29826102,0.0028741627,0.02401721],"genre_scores_gemma":[0.783523,0.00085523893,0.012419445,0.000082984036,0.00017664455,0.00018087288,0.19361924,0.00018493309,0.008957658],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9999075,0.000013565054,0.000008095075,0.000026755575,0.000029435514,0.00001459539],"domain_scores_gemma":[0.9998522,0.0000069298912,0.000012983253,0.000017968601,0.00009841686,0.000011476852],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039851668,0.00045481144,0.00029262,0.0007503365,0.00032132838,0.00061911513,0.00042663893,0.00024596794,0.0031768915],"category_scores_gemma":[0.00037091944,0.00016474766,0.000377065,0.00233742,0.000105212544,0.0002627967,0.00016768888,0.00029779543,0.00070198637],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009222501,0.00039580118,0.31788015,0.0011318598,0.00181801,0.00092730974,0.0009826457,0.24983427,0.022718966,0.0040113972,0.16345471,0.2359227],"study_design_scores_gemma":[0.0003811168,0.000085387604,0.653924,0.00011148247,0.0002815262,0.000117777825,0.00034146215,0.20182866,0.004543681,0.0011110458,0.13716744,0.00010649587],"about_ca_topic_score_codex":0.18944113,"about_ca_topic_score_gemma":0.1728029,"teacher_disagreement_score":0.18944113,"about_ca_system_score_codex":0.0009102242,"about_ca_system_score_gemma":0.0012469746,"threshold_uncertainty_score":0.37667698},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4396646579","doi":"10.1038/s41597-024-03300-5","title":"The phased Solanum okadae genome and Petota pangenome analysis of 23 other potato wild relatives and hybrids","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Pathogens and Resistance","field":"Agricultural and Biological Sciences","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of New Brunswick; Agriculture and Agri-Food Canada; McGill University","funders":"Agriculture and Agri-Food Canada; Government of Canada; Compute Canada; Direktoratet for Utviklingssamarbeid; Génome Québec; U.S. Department of Agriculture","keywords":"Polyploid; Genome; Biology; Introgression; Ploidy; Germplasm; Solanum tuberosum; Hybrid; Genetics; Solanum; Phylogenetic tree; Gene; Botany","score_opus":0.03343120371711719,"score_gpt":0.24376960080837823,"score_spread":0.21033839709126104,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396646579","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92363477,0.00076321024,0.0043878416,0.00007831372,0.000035564488,0.000058146008,0.068346575,0.00026405664,0.0024315382],"genre_scores_gemma":[0.613094,0.0007359975,0.021651598,0.00014543012,0.000015996377,0.00025414725,0.36093256,0.00047847876,0.002691846],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99983096,0.000011931634,0.00001399429,0.00009752661,0.000023510953,0.000022073165],"domain_scores_gemma":[0.99974674,0.00006080313,0.00004429529,0.00004220117,0.000052019037,0.000054002587],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00021139179,0.00048525986,0.00039957155,0.0010587021,0.0005959154,0.00054757216,0.00026789168,0.00034580976,0.0022993684],"category_scores_gemma":[0.00053658953,0.00020213619,0.0006837278,0.0015911979,0.00015527895,0.00030702743,0.00067441346,0.00052902143,0.0007835474],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009371433,0.000088064626,0.01814286,0.00070602464,0.0002139082,0.0007793215,0.0010691403,0.0009362129,0.9543787,0.00033017405,0.0013996417,0.021018859],"study_design_scores_gemma":[0.00013207068,0.00060434634,0.7330778,0.00018911975,0.0007381945,0.0034675,0.0023627023,0.0042335372,0.10546173,0.00072807306,0.1488831,0.000121824676],"about_ca_topic_score_codex":0.0019891805,"about_ca_topic_score_gemma":0.0053701224,"teacher_disagreement_score":0.0022993684,"about_ca_system_score_codex":0.00020885143,"about_ca_system_score_gemma":0.00027733337,"threshold_uncertainty_score":0.007692158},"labels":[],"label_agreement":null},{"id":"W4396736647","doi":"10.1038/s41597-024-03280-6","title":"Getting your DUCs in a row - standardising the representation of Digital Use Conditions","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Research Data Management Practices","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University Health Centre; Ontario Brain Institute","funders":"Institut National de la Santé et de la Recherche Médicale; European Commission","keywords":"Computer science; Context (archaeology); Ambiguity; Asset (computer security); Data science; Granularity; Representation (politics); Task (project management); Diversity (politics); Knowledge management; Risk analysis (engineering); Business; Systems engineering","score_opus":0.3149498459064001,"score_gpt":0.45547181828544675,"score_spread":0.14052197237904662,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396736647","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020525502,0.0015414576,0.8708756,0.02338485,0.0011341132,0.0008013211,0.0025255973,0.0033802742,0.07583134],"genre_scores_gemma":[0.23718478,0.002142118,0.74139535,0.00266714,0.00045083446,0.0010216058,0.003114146,0.0012774264,0.010746616],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9732006,0.012150369,0.005274548,0.0029273047,0.005188927,0.0012581561],"domain_scores_gemma":[0.9291704,0.023295084,0.004584493,0.028612616,0.011156313,0.003181073],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.03025257,0.001224074,0.0012961846,0.008497762,0.0037290673,0.02392035,0.0030740954,0.004479668,0.005912913],"category_scores_gemma":[0.07247039,0.0012446684,0.001632039,0.0063543604,0.014771833,0.03870075,0.012506884,0.007705025,0.0030019858],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009008485,0.00006953826,0.0034819855,0.00035016373,0.00003623175,0.0003952517,0.0077617797,0.0025142324,0.0018225595,0.8716067,0.01295265,0.098918885],"study_design_scores_gemma":[0.000022078266,0.000073165655,0.0014708531,0.0014084831,0.00005489788,0.000945224,0.009730507,0.009769272,0.0046039377,0.5316194,0.44012544,0.00017671552],"about_ca_topic_score_codex":0.0063024675,"about_ca_topic_score_gemma":0.0062613455,"teacher_disagreement_score":0.9697474,"about_ca_system_score_codex":0.004770067,"about_ca_system_score_gemma":0.0076997373,"threshold_uncertainty_score":0.15999281},"labels":[],"label_agreement":null},{"id":"W4396749763","doi":"10.1038/s41597-024-03279-z","title":"Common conditions of use elements. Atomic concepts for consistent and effective information governance","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Research Data Management Practices","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Ontario Brain Institute","funders":"Institut National de la Santé et de la Recherche Médicale","keywords":"Corporate governance; Information retrieval; Business; Computer science","score_opus":0.09817008775534806,"score_gpt":0.40416528018941256,"score_spread":0.3059951924340645,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396749763","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055024855,0.005894881,0.7218681,0.028268412,0.0014345254,0.00395462,0.005829175,0.0021925487,0.17553289],"genre_scores_gemma":[0.50379634,0.00278553,0.46793228,0.0036385637,0.00050533604,0.005389596,0.0054108654,0.00082666,0.009714863],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.95717305,0.023040257,0.00851013,0.0033354259,0.0063402764,0.0016008967],"domain_scores_gemma":[0.93928975,0.03647773,0.0057864035,0.010296632,0.0065016295,0.0016478082],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.023723235,0.0012095233,0.0011447404,0.008615699,0.004534784,0.012865165,0.002202502,0.005434183,0.0063192816],"category_scores_gemma":[0.057021048,0.0013143058,0.0016636361,0.0074127647,0.020390287,0.02695195,0.009165088,0.004202533,0.0018864314],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000076171615,0.000025731237,0.0024426451,0.00087311084,0.000033149583,0.0003443156,0.027260462,0.0005122282,0.0016926647,0.9330679,0.007538288,0.026133299],"study_design_scores_gemma":[0.00003567427,0.00005596973,0.0024958856,0.0018627039,0.00007118268,0.0013453774,0.019749053,0.0018333419,0.0015214079,0.5566112,0.41429076,0.00012739545],"about_ca_topic_score_codex":0.0060322043,"about_ca_topic_score_gemma":0.004453605,"teacher_disagreement_score":0.97627676,"about_ca_system_score_codex":0.0053261365,"about_ca_system_score_gemma":0.009412389,"threshold_uncertainty_score":0.12546194},"labels":[],"label_agreement":null},{"id":"W4396799260","doi":"10.1038/s41597-024-03306-z","title":"How accurate are existing land cover maps for agriculture in Sub-Saharan Africa?","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Land Use and Ecosystem Services","field":"Environmental Science","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"National Aeronautics and Space Administration","keywords":"Land cover; Agriculture; Land use; Agricultural land; Environmental resource management; Satellite; Work (physics); Monitoring and evaluation; Remote sensing; Geography; Environmental science; Ecology; Economic growth","score_opus":0.05920618791904884,"score_gpt":0.2583645116580236,"score_spread":0.19915832373897477,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396799260","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90968484,0.012127412,0.015427238,0.00909362,0.0003413065,0.00019878057,0.029734507,0.00050734484,0.022884868],"genre_scores_gemma":[0.97698885,0.0030008154,0.013497567,0.00032239905,0.00006642714,0.00007129338,0.0055977744,0.00007811952,0.0003767408],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99696285,0.001310737,0.00028740618,0.0004050152,0.00075491215,0.0002790606],"domain_scores_gemma":[0.99001557,0.0031939365,0.0021837035,0.0014421325,0.0028620707,0.00030258318],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005426576,0.00048385066,0.00039278078,0.004581571,0.0004693951,0.0022307106,0.00079800794,0.00070326135,0.0016039456],"category_scores_gemma":[0.028634137,0.00022683304,0.00040174575,0.006804305,0.00057466235,0.0044116587,0.00065402716,0.000419309,0.00067703356],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012533745,0.00009508301,0.6700046,0.0006520308,0.00041717873,0.00026954227,0.00209386,0.009014886,0.0010682413,0.0027513036,0.011384445,0.3021234],"study_design_scores_gemma":[0.00002154304,0.000066920424,0.9332472,0.0011253849,0.00017994262,0.00026977013,0.006669269,0.011962497,0.0012211163,0.002951847,0.042220794,0.00006367397],"about_ca_topic_score_codex":0.027477425,"about_ca_topic_score_gemma":0.035371125,"teacher_disagreement_score":0.027477425,"about_ca_system_score_codex":0.0009382273,"about_ca_system_score_gemma":0.00084582315,"threshold_uncertainty_score":0.05463493},"labels":[],"label_agreement":null},{"id":"W4396890157","doi":"10.1038/s41597-024-03327-8","title":"Analysis of AlphaMissense data in different protein groups and structural context","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":120,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"National Research, Development and Innovation Office; Semmelweis Egyetem","keywords":"Context (archaeology); Computer science; Information retrieval; Computational biology; Biology; Paleontology","score_opus":0.026816436164746913,"score_gpt":0.29135570214338696,"score_spread":0.26453926597864,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396890157","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8824672,0.0099743465,0.017176738,0.0012312804,0.000364188,0.00015117812,0.07342781,0.0107533615,0.004453845],"genre_scores_gemma":[0.7490057,0.0016706031,0.02575763,0.0004658478,0.00013840293,0.00011103894,0.22111881,0.0007311753,0.0010008288],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9962334,0.00078590837,0.0004876718,0.0011713072,0.0010819563,0.00023983097],"domain_scores_gemma":[0.9901349,0.0040306714,0.00125607,0.0024499027,0.0016015344,0.0005269935],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005967494,0.0013451134,0.0013517226,0.0058635734,0.0010658599,0.0020385096,0.001051927,0.001502694,0.0013242831],"category_scores_gemma":[0.013366633,0.00029413376,0.0013674776,0.0037881883,0.00073580036,0.0013037161,0.0016669688,0.0012859072,0.0014199778],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00534899,0.00082780357,0.5917246,0.0039655566,0.002788197,0.004651378,0.0009153343,0.030642089,0.11146824,0.0038765771,0.06732495,0.17646633],"study_design_scores_gemma":[0.00053018384,0.002714896,0.46132883,0.00094670744,0.0011944813,0.01688306,0.0016869371,0.24092093,0.10935184,0.022838172,0.1412199,0.00038400295],"about_ca_topic_score_codex":0.0019814002,"about_ca_topic_score_gemma":0.0050668446,"teacher_disagreement_score":0.005967494,"about_ca_system_score_codex":0.00063585053,"about_ca_system_score_gemma":0.0009174275,"threshold_uncertainty_score":0.031559527},"labels":[],"label_agreement":null},{"id":"W4396890326","doi":"10.1038/s41597-024-03295-z","title":"ReMIND: The Brain Resection Multimodal Imaging Database","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Radiomics and Machine Learning in Medical Imaging","field":"Medicine","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"National Institutes of Health; Brigham and Women's Hospital; National Institute of Biomedical Imaging and Bioengineering; U.S. Department of Health and Human Services","keywords":"Neuronavigation; Magnetic resonance imaging; Interventional magnetic resonance imaging; Medicine; Intraoperative MRI; Workflow; Database; Radiology; Brain tumor; Computer science; Pathology","score_opus":0.03263658663287678,"score_gpt":0.35931578249153895,"score_spread":0.3266791958586622,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396890326","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025942829,0.0005632774,0.0012204153,0.0001682055,0.000039874274,0.00010241582,0.9909156,0.0027121943,0.0016837916],"genre_scores_gemma":[0.004966802,0.00029216663,0.0014189768,0.00010140201,0.00003000627,0.00018707367,0.99237436,0.00019602124,0.00043316482],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992455,0.000076774755,0.00022215174,0.00022113333,0.00015630251,0.000078131474],"domain_scores_gemma":[0.9967235,0.0006534847,0.0005596865,0.0009658238,0.0007198534,0.00037782334],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009153821,0.0014937466,0.0013596851,0.0054558143,0.0004268989,0.002032921,0.0022388552,0.001537073,0.038013794],"category_scores_gemma":[0.0069010993,0.00052191206,0.0006864998,0.0061161746,0.0002897023,0.0013996104,0.0021327862,0.0010689003,0.0412924],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007892993,0.00010749011,0.01008163,0.0018872814,0.00018348914,0.0005705271,0.00009989802,0.0011807528,0.0023663323,0.0012613455,0.932905,0.04856706],"study_design_scores_gemma":[0.00067651906,0.00012069373,0.042113464,0.00055681536,0.0002312709,0.0018604408,0.00026742372,0.0039991383,0.0038908436,0.003912501,0.9422221,0.00014880126],"about_ca_topic_score_codex":0.00512024,"about_ca_topic_score_gemma":0.0058890027,"teacher_disagreement_score":0.038013794,"about_ca_system_score_codex":0.00073276303,"about_ca_system_score_gemma":0.0017783771,"threshold_uncertainty_score":0.12716877},"labels":[],"label_agreement":null},{"id":"W4396930028","doi":"10.1038/s41597-024-03350-9","title":"A multi-institutional meningioma MRI dataset for automated multi-sequence image segmentation","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Meningioma and schwannoma management","field":"Medicine","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba; University of Toronto; University of Ottawa","funders":"National Cancer Institute; National Institutes of Health; Perelman School of Medicine, University of Pennsylvania; Instituto Tecnológico de Costa Rica; American Society of Neuroradiology; University of Pennsylvania; Foundation for the National Institutes of Health","keywords":"Meningioma; Fluid-attenuated inversion recovery; Segmentation; Sequence (biology); Contrast (vision); Radiology; Medicine; Computer science; Magnetic resonance imaging; Medical physics; Artificial intelligence","score_opus":0.1666244615822458,"score_gpt":0.4157804038618527,"score_spread":0.2491559422796069,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396930028","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09006181,0.0031014173,0.022404736,0.0010737599,0.0003655938,0.0010923285,0.8637607,0.013711528,0.004428124],"genre_scores_gemma":[0.037527554,0.0005004795,0.02761465,0.00014734341,0.000057829857,0.0005794446,0.9320419,0.00038904132,0.0011417595],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991968,0.000113232665,0.00011513401,0.00027611127,0.00020884932,0.00008984185],"domain_scores_gemma":[0.9988937,0.00021500776,0.00012578136,0.00032840532,0.00031112367,0.00012609243],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001324994,0.0019198138,0.0010517987,0.0026584666,0.0008173803,0.0011147291,0.0026537282,0.0019494884,0.004455049],"category_scores_gemma":[0.0035172878,0.00050106965,0.0016296632,0.0018413389,0.00045845838,0.0006108132,0.0016263772,0.0013502757,0.0060197148],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026217687,0.0013612236,0.027711553,0.0038716127,0.0014937234,0.0031251628,0.0005039713,0.027668197,0.042463187,0.0026868181,0.7210288,0.16546398],"study_design_scores_gemma":[0.0018544242,0.001041851,0.1329844,0.0011959551,0.0008951244,0.010109833,0.0008819589,0.12295949,0.047053333,0.011380805,0.669077,0.0005658119],"about_ca_topic_score_codex":0.01428341,"about_ca_topic_score_gemma":0.030468859,"teacher_disagreement_score":0.01428341,"about_ca_system_score_codex":0.0011277959,"about_ca_system_score_gemma":0.0017896753,"threshold_uncertainty_score":0.02840054},"labels":[],"label_agreement":null},{"id":"W4397005444","doi":"10.1038/s41597-024-03345-6","title":"LNDb v4: pulmonary nodule annotation from medical reports","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Lung Cancer Diagnosis and Treatment","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Fundação para a Ciência e a Tecnologia; Centre hospitalier universitaire Sainte-Justine","keywords":"Annotation; Nodule (geology); Computer science; Medicine; Information retrieval; Computational biology; Biology; Artificial intelligence; Paleontology","score_opus":0.040594619729269886,"score_gpt":0.3490781918351679,"score_spread":0.30848357210589805,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4397005444","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006930795,0.00059854233,0.022074504,0.0006074785,0.00015546852,0.00068026286,0.88238126,0.080677554,0.0058940835],"genre_scores_gemma":[0.014056527,0.0005160577,0.040943116,0.00020647317,0.00008313781,0.0007016955,0.93750685,0.0038262166,0.0021599578],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.995453,0.0007915795,0.001548496,0.0007632256,0.0011707168,0.00027294623],"domain_scores_gemma":[0.9812164,0.008152224,0.0024158782,0.0041809687,0.0033564712,0.0006781173],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0049908794,0.0018151022,0.0012455472,0.0142420735,0.0007629214,0.0044606593,0.0024942637,0.0017780978,0.039545663],"category_scores_gemma":[0.030969303,0.0010764217,0.0012147588,0.008960496,0.0005658846,0.0033360096,0.003080882,0.0011286535,0.028598087],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012569951,0.00033333764,0.016661698,0.008129547,0.00027022904,0.0012803699,0.0013711439,0.004498919,0.01357283,0.0053086616,0.80671537,0.14060092],"study_design_scores_gemma":[0.00040609814,0.00023984663,0.033319358,0.0012611187,0.0001442158,0.001344282,0.00088831026,0.013545083,0.019536845,0.0042812102,0.924776,0.00025753293],"about_ca_topic_score_codex":0.010961836,"about_ca_topic_score_gemma":0.006952422,"teacher_disagreement_score":0.039545663,"about_ca_system_score_codex":0.0015715011,"about_ca_system_score_gemma":0.003510981,"threshold_uncertainty_score":0.1322934},"labels":[],"label_agreement":null},{"id":"W4398238439","doi":"10.1038/s41597-024-03319-8","title":"Typical and extreme weather datasets for studying the resilience of buildings to climate change and heatwaves","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Wind and Air Flow Studies","field":"Environmental Science","cited_by":87,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; National Research Council Canada","funders":"Energiteknologisk udviklings- og demonstrationsprogram; Fraunhofer-Gesellschaft; European Commission; U.S. Department of Energy","keywords":"Resilience (materials science); Extreme weather; Climate change; Climate extremes; Environmental science; Climatology; Extreme heat; Geography; Ecology; Geology; Biology","score_opus":0.1085341305823821,"score_gpt":0.3079705183688564,"score_spread":0.19943638778647432,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4398238439","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020381073,0.000083444946,0.0067852787,0.00024469342,0.00006673616,0.00012177463,0.96591103,0.0020558056,0.0043501947],"genre_scores_gemma":[0.04151398,0.00012625357,0.012923089,0.000061729086,0.000033012853,0.0005119481,0.94395137,0.00029059747,0.00058790576],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990615,0.00020886923,0.0001535673,0.0001733381,0.0003015733,0.000101210535],"domain_scores_gemma":[0.9973465,0.0004948411,0.00036469827,0.00094943674,0.0006275753,0.00021706031],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012432692,0.0008569905,0.0006275761,0.0014794905,0.00045300173,0.0008434107,0.0016320149,0.00095773133,0.00767183],"category_scores_gemma":[0.004131009,0.00037687385,0.0009020633,0.0045429,0.0002821411,0.001552003,0.0012246029,0.0010938976,0.003406554],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006177008,0.00045213962,0.0643553,0.0014701288,0.00041199417,0.00026303393,0.00047533048,0.19151819,0.0039192652,0.013892253,0.68075037,0.041874304],"study_design_scores_gemma":[0.00087671046,0.00024152586,0.18351728,0.00048490718,0.0001557698,0.00036026738,0.000832763,0.19503582,0.012052935,0.014788193,0.5913738,0.00027997096],"about_ca_topic_score_codex":0.0148526,"about_ca_topic_score_gemma":0.017221257,"teacher_disagreement_score":0.0148526,"about_ca_system_score_codex":0.00067689165,"about_ca_system_score_gemma":0.0010184495,"threshold_uncertainty_score":0.029532313},"labels":[],"label_agreement":null},{"id":"W4399179273","doi":"10.1038/s41597-024-03410-0","title":"Global Spore Sampling Project: A global, standardized dataset of airborne fungal DNA","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Indoor Air Quality and Microbial Exposure","field":"Environmental Science","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Center for Northern Studies; Université de Moncton; University of Guelph; Saint Mary's University; MacEwan University","funders":"Biotechnology and Biological Sciences Research Council; Natural Sciences and Engineering Research Council of Canada; Nature; Natural Environment Research Council; Mushroom Research Foundation; Vetenskapsrådet; Eesti Teadusagentuur; Svenska Forskningsrådet Formas; Novo Nordisk Fonden; Norges Forskningsråd; Academy of Finland; National Natural Science Foundation of China; Fundação de Amparo à Pesquisa do Estado de São Paulo; European Commission; Sight Research UK; Hrvatska Zaklada za Znanost; Novo Nordisk; Polar Knowledge Canada; Jane ja Aatos Erkon Säätiö; National Science Foundation","keywords":"Sampling (signal processing); Spore; Biology; Environmental science; Geography; Botany; Computer science","score_opus":0.05739459153731757,"score_gpt":0.33950596873200195,"score_spread":0.2821113771946844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399179273","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1568672,0.00044082777,0.014287016,0.0001373934,0.000051898656,0.00038442845,0.825135,0.0008706251,0.0018255457],"genre_scores_gemma":[0.074659094,0.00012062445,0.022071201,0.0000695736,0.000024053601,0.00071617484,0.90173036,0.00018843544,0.00042041962],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99790204,0.0004686077,0.00025470363,0.0007021428,0.00046161044,0.00021090916],"domain_scores_gemma":[0.99751943,0.00044468313,0.00065343553,0.00054979656,0.0005101318,0.00032250467],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023306422,0.00095282384,0.000732982,0.003175818,0.0005795211,0.0008201403,0.0008086479,0.0008270922,0.0026691926],"category_scores_gemma":[0.004453719,0.00039595246,0.0006094414,0.0044363593,0.00043899074,0.00047597996,0.002256447,0.00088597037,0.0021235582],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017535612,0.00068387657,0.5673607,0.0032079117,0.0010737445,0.0008920529,0.0015942742,0.012576007,0.085059546,0.0024522278,0.1986177,0.124728434],"study_design_scores_gemma":[0.00039078327,0.00042069773,0.77577525,0.00019453588,0.00015865524,0.00057341426,0.00065567583,0.005578535,0.010192059,0.001355579,0.20460625,0.000098615805],"about_ca_topic_score_codex":0.007976214,"about_ca_topic_score_gemma":0.0088736275,"teacher_disagreement_score":0.007976214,"about_ca_system_score_codex":0.00035637285,"about_ca_system_score_gemma":0.0014005617,"threshold_uncertainty_score":0.015859544},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4399241163","doi":"10.1038/s41597-024-03384-z","title":"A labelled dataset to classify direct deforestation drivers from Earth Observation imagery in Cameroon","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"International Institute for Sustainable Development","funders":"Natural Environment Research Council; Research Councils UK","keywords":"Deforestation (computer science); Earth (classical element); Remote sensing; Geography; Geology; Environmental science; Earth science; Physical geography; Computer science; Mathematics","score_opus":0.036656666155668864,"score_gpt":0.2699228151220721,"score_spread":0.23326614896640324,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399241163","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19981243,0.0012496568,0.0023186284,0.0006266566,0.00022105366,0.0004646243,0.79154336,0.0011202183,0.0026434076],"genre_scores_gemma":[0.10716263,0.00028709244,0.00586806,0.00011465667,0.00003821155,0.00034787133,0.88519937,0.000044914763,0.0009371952],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995541,0.00008895602,0.000046866022,0.00014084816,0.000055810935,0.00011348962],"domain_scores_gemma":[0.9993247,0.00017798718,0.000090379996,0.00011300237,0.0001985136,0.000095418065],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006701592,0.0012787113,0.00040931941,0.003080441,0.00064166007,0.00059674296,0.0007826858,0.0008818242,0.0031514],"category_scores_gemma":[0.0017894879,0.00020784188,0.0005922209,0.003260546,0.00048830727,0.0006579,0.0007947528,0.0007624483,0.0015488039],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014055511,0.001132528,0.20785145,0.0028375168,0.00036163558,0.0031382283,0.0010808466,0.026749842,0.019180924,0.0026166807,0.586331,0.14731371],"study_design_scores_gemma":[0.00053454435,0.00028893433,0.5855343,0.0010158168,0.00019461189,0.0009963588,0.0036995765,0.06236294,0.0098921945,0.0015956296,0.3336738,0.00021130887],"about_ca_topic_score_codex":0.08890255,"about_ca_topic_score_gemma":0.12276396,"teacher_disagreement_score":0.08890255,"about_ca_system_score_codex":0.001609649,"about_ca_system_score_gemma":0.0013923462,"threshold_uncertainty_score":0.17677015},"labels":[],"label_agreement":null},{"id":"W4399335878","doi":"10.1038/s41597-024-03416-8","title":"Publication, funding, and experimental data in support of Human Reference Atlas construction and usage","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Common Fund; NIH Office of the Director; National Institute of Diabetes and Digestive and Kidney Diseases; National Cancer Institute; National Institutes of Health; Canadian Institute for Advanced Research","keywords":"Metadata; Computer science; Atlas (anatomy); Information retrieval; Data science; World Wide Web; Digital library; Medicine","score_opus":0.08935916613375858,"score_gpt":0.33606086493482396,"score_spread":0.24670169880106538,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399335878","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012803715,0.003462931,0.024466746,0.0012338986,0.00052702497,0.0008410008,0.9287378,0.002676414,0.0252505],"genre_scores_gemma":[0.023957092,0.002844973,0.06269388,0.00028891893,0.0001502481,0.002005443,0.90329987,0.000899262,0.0038603283],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9715973,0.00528473,0.009823312,0.0045266533,0.0074884174,0.0012795083],"domain_scores_gemma":[0.88570416,0.048243422,0.014101875,0.028575638,0.019817099,0.0035577577],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.028475301,0.0013235032,0.0019302306,0.05505793,0.0021439265,0.006972698,0.00253948,0.0010455885,0.030637814],"category_scores_gemma":[0.09092993,0.00090225105,0.0016314834,0.075355716,0.001326105,0.0041920063,0.005494905,0.0013394131,0.014368289],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012035891,0.00019567272,0.06930183,0.023216005,0.0010706985,0.0016485037,0.0061349543,0.0036264088,0.015050254,0.051272538,0.4979475,0.32933205],"study_design_scores_gemma":[0.00008183716,0.000050386687,0.03175026,0.0018202408,0.00048674017,0.00039799727,0.0011055279,0.0006099224,0.006303772,0.005190519,0.9521023,0.00010056605],"about_ca_topic_score_codex":0.014711448,"about_ca_topic_score_gemma":0.01735969,"teacher_disagreement_score":0.99746054,"about_ca_system_score_codex":0.0036268374,"about_ca_system_score_gemma":0.020134343,"threshold_uncertainty_score":0.15059358},"labels":[],"label_agreement":null},{"id":"W4399362984","doi":"10.1038/s41597-024-03390-1","title":"Individual Brain Charting dataset extension, third release for movie watching and retinotopy data","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Canada First Research Excellence Fund; European Commission; University of Minnesota","keywords":"Retinotopy; Computer science; Functional magnetic resonance imaging; Neuroscience; Visual cortex; Task (project management); Artificial intelligence; Psychology","score_opus":0.20730925246454832,"score_gpt":0.36652595026983376,"score_spread":0.15921669780528544,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399362984","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018309868,0.00015045947,0.0015138336,0.00008038024,0.000062891544,0.00010080603,0.99025285,0.0049368427,0.001071004],"genre_scores_gemma":[0.0035721692,0.000056265108,0.0037250465,0.000054008065,0.000019519166,0.00039954766,0.990721,0.0006375858,0.0008148736],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994542,0.0000776154,0.000056090063,0.00021778242,0.000108062515,0.00008621742],"domain_scores_gemma":[0.99889755,0.00022604596,0.00008963285,0.00043931557,0.00022294163,0.0001244734],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010598273,0.0018507588,0.0011547656,0.0019235215,0.0007280497,0.0015154002,0.002189141,0.0017410467,0.034442708],"category_scores_gemma":[0.0044623343,0.000554786,0.0016681023,0.0022238132,0.0004640927,0.00082211214,0.0017056974,0.001623396,0.034381885],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004603089,0.00012446671,0.003554582,0.0013171228,0.00028514283,0.00020529215,0.00014173718,0.0017299356,0.003936468,0.001018113,0.97285235,0.014374484],"study_design_scores_gemma":[0.0008356543,0.00012728943,0.029038006,0.0002963618,0.00018568484,0.000518094,0.0001551477,0.005090565,0.0055174073,0.005011887,0.95307946,0.0001444173],"about_ca_topic_score_codex":0.015717918,"about_ca_topic_score_gemma":0.03703334,"teacher_disagreement_score":0.034442708,"about_ca_system_score_codex":0.0006696535,"about_ca_system_score_gemma":0.0014740503,"threshold_uncertainty_score":0.115222275},"labels":[],"label_agreement":null},{"id":"W4399390779","doi":"10.1038/s41597-024-03443-5","title":"The Pan-Canadian Chemical Library: A Mechanism to Open Academic Chemistry to High-Throughput Virtual Screening","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Winnipeg; University of Alberta; Structural Genomics Consortium; University of Toronto","funders":"National Institute of General Medical Sciences; Genentech; European Federation of Pharmaceutical Industries and Associations; University of Toronto; Ontario Genomics Institute; Merck KGaA; Alliance de recherche numérique du Canada; National Institutes of Health; Ontario Genomics; Genome Canada; McGill University; Bayer; Pfizer; Bristol-Myers Squibb","keywords":"Chemical space; Drug discovery; Workflow; Virtual screening; Computer science; Chemical library; Data science; Cheminformatics; Chemical database; Process (computing); Mechanism (biology); Chemistry; Chemical biology; Resource (disambiguation); Nanotechnology; Small molecule; Database; Physics; Materials science; Computational chemistry; Organic chemistry","score_opus":0.06922418333834772,"score_gpt":0.3512207175254694,"score_spread":0.2819965341871217,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399390779","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.059777386,0.0046266234,0.6669582,0.0056391796,0.0009583761,0.0068510245,0.04047036,0.06443249,0.15028642],"genre_scores_gemma":[0.135583,0.004342275,0.7690748,0.0025715856,0.00012887575,0.0029512602,0.03862142,0.0032329408,0.0434938],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981862,0.00023552966,0.00006848502,0.00024763177,0.0010404144,0.00022180218],"domain_scores_gemma":[0.9983789,0.0002402283,0.000086367545,0.00042840437,0.0005594063,0.00030674724],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0019952205,0.0010435092,0.0009690819,0.0031210065,0.0018222859,0.0020837192,0.00296503,0.00084665,0.01122046],"category_scores_gemma":[0.004194284,0.0006553793,0.0010656756,0.0045972923,0.0015534298,0.0014775919,0.003858282,0.0019102031,0.0048129284],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018173648,0.0008281171,0.0038416474,0.0013110221,0.00032146214,0.0008059817,0.0006500719,0.022623237,0.13282844,0.11331404,0.22304715,0.49861145],"study_design_scores_gemma":[0.0005718056,0.0005198029,0.0027207506,0.00012735986,0.00017698007,0.000600948,0.00015474857,0.040460702,0.10229088,0.02007047,0.8320172,0.0002883824],"about_ca_topic_score_codex":0.096684575,"about_ca_topic_score_gemma":0.20846032,"teacher_disagreement_score":0.99703497,"about_ca_system_score_codex":0.003262876,"about_ca_system_score_gemma":0.014673689,"threshold_uncertainty_score":0.19224364},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":true,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W4399437290","doi":"10.1038/s41597-024-03396-9","title":"Three-dimensional reconstruction of high latitude bamboo coral via X-ray microfocus Computed Tomography","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Bamboo properties and applications","field":"Agricultural and Biological Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval; ArcticNet","funders":"Fisheries and Oceans Canada; Memorial University of Newfoundland; Research Councils UK; Université Laval","keywords":"Bamboo; Coral; Computed tomography; Tomography; Latitude; Geology; Radiology; Biology; Geodesy; Medicine; Oceanography; Botany","score_opus":0.03660706357985551,"score_gpt":0.2273520553715746,"score_spread":0.19074499179171908,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399437290","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9660266,0.00034992414,0.028636677,0.00015769384,0.000007896724,0.00004545777,0.001673153,0.00035195737,0.0027505588],"genre_scores_gemma":[0.9763757,0.00027840675,0.021585414,0.0000229415,0.0000052992577,0.000026859367,0.0010525907,0.00004923506,0.0006034826],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9999294,0.000007082271,0.000005046731,0.000014740519,0.00003072695,0.000012943853],"domain_scores_gemma":[0.99986744,0.000029938772,0.000034972807,0.000017686312,0.00003588208,0.000014150141],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00015358173,0.00040931202,0.00012978123,0.0012198478,0.00021057382,0.0005370344,0.00021149118,0.00027247265,0.0017550731],"category_scores_gemma":[0.0004151624,0.00035032822,0.00017133384,0.00085181394,0.00026442562,0.00021104963,0.0003119403,0.00022804439,0.00027445648],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051264215,0.0001962513,0.21685582,0.00049694994,0.00023910547,0.002385834,0.0017578611,0.10728004,0.56959873,0.002138451,0.0023886308,0.09614973],"study_design_scores_gemma":[0.000047840953,0.00009797845,0.7172402,0.00009068719,0.000105540224,0.0025996268,0.000992817,0.23761185,0.034986526,0.0006795898,0.005437981,0.00010924763],"about_ca_topic_score_codex":0.015815645,"about_ca_topic_score_gemma":0.03562133,"teacher_disagreement_score":0.015815645,"about_ca_system_score_codex":0.00027687868,"about_ca_system_score_gemma":0.0005252831,"threshold_uncertainty_score":0.031447172},"labels":[],"label_agreement":null},{"id":"W4399514622","doi":"10.1038/s41597-024-03395-w","title":"A Global Multi-Source Tropical Cyclone Precipitation (MSTCP) Dataset","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Tropical and Extratropical Cyclones Research","field":"Earth and Planetary Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Nuclear Safety and Security Commission; Université du Québec à Montréal; National Aeronautics and Space Administration","keywords":"Tropical cyclone; Precipitation; Storm; Environmental science; Climatology; Tropical cyclone scales; Quantitative precipitation estimation; Meteorology; Cyclone (programming language); Geography; Geology; Computer science","score_opus":0.07639561662460964,"score_gpt":0.33238279257025183,"score_spread":0.2559871759456422,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399514622","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017272667,0.00003581243,0.0001888431,0.000046544268,0.000022594199,0.000026404894,0.9970523,0.00022711625,0.0006730991],"genre_scores_gemma":[0.0022967537,0.000027475384,0.0005013011,0.000017880484,0.000008980512,0.000062945066,0.9968238,0.000026962674,0.00023391814],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994943,0.00006781011,0.00008615125,0.00013723414,0.0001443091,0.00007016884],"domain_scores_gemma":[0.9988907,0.0001699016,0.00014481942,0.00021093177,0.00049078994,0.00009285305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005189345,0.000845154,0.00065101846,0.0019222421,0.00037519942,0.0007310389,0.0011783992,0.0007464916,0.012702503],"category_scores_gemma":[0.0020176251,0.00021258958,0.00065484736,0.0038875903,0.00016140893,0.0006372386,0.0008107291,0.0008959374,0.012621543],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001811721,0.0000994776,0.01505977,0.0007044475,0.0001200579,0.00014501,0.000073405114,0.0046178116,0.0012997662,0.0011114072,0.9630352,0.013552391],"study_design_scores_gemma":[0.00031132437,0.00006188168,0.07356267,0.00025536775,0.00006455342,0.00015876786,0.00024249687,0.009317192,0.0022385938,0.0013560236,0.91236216,0.0000690136],"about_ca_topic_score_codex":0.020806307,"about_ca_topic_score_gemma":0.025336385,"teacher_disagreement_score":0.020806307,"about_ca_system_score_codex":0.0007807342,"about_ca_system_score_gemma":0.0012564678,"threshold_uncertainty_score":0.04249406},"labels":[],"label_agreement":null},{"id":"W4399694734","doi":"10.1038/s41597-024-03455-1","title":"3D atlas of the human fetal chondrocranium in the middle trimester","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Bone Tumor Diagnosis and Treatments","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Vysoké Učení Technické v Brně; Grantová Agentura České Republiky; Ministerstvo Školství, Mládeže a Tělovýchovy","keywords":"Endochondral ossification; Anatomy; Biology; Craniofacial; Skull; Atlas (anatomy); Cartilage","score_opus":0.10400742827335764,"score_gpt":0.32918538283790905,"score_spread":0.22517795456455142,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399694734","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3118315,0.012876973,0.5416081,0.0015592317,0.00038358849,0.0005645267,0.0327232,0.0046174284,0.09383545],"genre_scores_gemma":[0.63343996,0.0070158644,0.32477015,0.00023429778,0.000059880753,0.0004922585,0.00880429,0.00078570726,0.024397656],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99989223,0.000014992638,0.000009680695,0.000019583256,0.00004976501,0.0000137502275],"domain_scores_gemma":[0.9998467,0.000045672266,0.000020135287,0.000027969689,0.00004517061,0.000014272664],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002391584,0.00022652745,0.00015976424,0.0015992336,0.00022501007,0.00058953406,0.00018658905,0.0004916443,0.0058224928],"category_scores_gemma":[0.0004081433,0.00027631328,0.0002953477,0.00084852043,0.00019946459,0.0001724066,0.0003519744,0.00034377322,0.001415461],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00071706495,0.0000612146,0.024187014,0.0009870806,0.00009122236,0.006396694,0.0017761862,0.023012431,0.50483114,0.020624083,0.029389927,0.387926],"study_design_scores_gemma":[0.000051879833,0.00030913035,0.1325317,0.0004975729,0.00019021101,0.050535835,0.0007226556,0.04643464,0.27118284,0.0046251235,0.49273533,0.00018312398],"about_ca_topic_score_codex":0.0063760923,"about_ca_topic_score_gemma":0.011036577,"teacher_disagreement_score":0.0063760923,"about_ca_system_score_codex":0.0003626433,"about_ca_system_score_gemma":0.0010268507,"threshold_uncertainty_score":0.019478142},"labels":[],"label_agreement":null},{"id":"W4399886311","doi":"10.1038/s41597-024-03414-w","title":"SARS-CoV-2 viral titer measurements in Ontario, Canada wastewaters throughout the COVID-19 pandemic","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"SARS-CoV-2 detection and testing","field":"Medicine","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Children's Hospital of Eastern Ontario; Ministry of the Environment, Conservation and Parks; University of Windsor; Trent University; University of Guelph; Ontario Tech University; McMaster University; Queen's University; Western University; University of Toronto; Health Sciences North; Ottawa Hospital; Toronto Metropolitan University; Carleton University; University of Waterloo; University of Ottawa","funders":"Ministère de l’Environnement, de la Protection de la nature et des Parcs; Canadian Institutes of Health Research; University of Waterloo; University of Toronto; University of Ontario Institute of Technology; Trent University; Queen's University; McMaster University; University of Windsor; University of Ottawa","keywords":"Pandemic; Coronavirus disease 2019 (COVID-19); Virology; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); 2019-20 coronavirus outbreak; Titer; Betacoronavirus; Biology; Medicine; Virus; Outbreak; Infectious disease (medical specialty)","score_opus":0.2744851924528665,"score_gpt":0.3861850209734499,"score_spread":0.11169982852058341,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399886311","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5436605,0.002081101,0.0025354486,0.0029470136,0.00011831612,0.00033139726,0.42879307,0.00038587174,0.019147303],"genre_scores_gemma":[0.75352454,0.002309735,0.0053582704,0.00078539085,0.00005626091,0.00028083925,0.22564131,0.00010963924,0.011934051],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.998283,0.000117222975,0.00012619213,0.000303368,0.00077055255,0.0003996762],"domain_scores_gemma":[0.9960556,0.0001975418,0.00040517177,0.00019424713,0.0028622383,0.00028529807],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009621837,0.00044767692,0.00045670648,0.002015017,0.002306086,0.0014830522,0.0009862211,0.0004549534,0.0016237716],"category_scores_gemma":[0.004042913,0.00028480456,0.0005035237,0.0064241113,0.00076718343,0.0004731944,0.0010452404,0.0004115773,0.00053995394],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006400846,0.00008992533,0.8506202,0.000961142,0.00022011745,0.00043832735,0.003338502,0.0054851566,0.0037010615,0.0014101281,0.0866292,0.046466228],"study_design_scores_gemma":[0.00005231301,0.000049796272,0.91061765,0.00027785552,0.00007526625,0.00010482452,0.0038274936,0.0033862353,0.001552993,0.0003066302,0.07967061,0.00007833777],"about_ca_topic_score_codex":0.9949639,"about_ca_topic_score_gemma":0.99648654,"teacher_disagreement_score":0.031659607,"about_ca_system_score_codex":0.031659607,"about_ca_system_score_gemma":0.045978494,"threshold_uncertainty_score":0.2297076},"labels":[],"label_agreement":null},{"id":"W4399920911","doi":"10.1038/s41597-024-03477-9","title":"Spatially distributed atmospheric boundary layer properties in Houston – A value-added observational dataset","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric aerosols and clouds","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Texas Commission on Environmental Quality; U.S. Department of Energy; National Science Foundation","keywords":"Planetary boundary layer; Wind speed; Meteorology; Environmental science; Convective Boundary Layer; Relative humidity; Shortwave; RADIUS; Boundary layer; Atmospheric sciences; Geology; Geography; Computer science; Turbulence; Radiative transfer; Physics; Mechanics","score_opus":0.07452173068592607,"score_gpt":0.28003448562847194,"score_spread":0.20551275494254587,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399920911","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21013603,0.00017437009,0.0007534351,0.00020789987,0.000045247765,0.00006251938,0.7861831,0.00041434445,0.0020229858],"genre_scores_gemma":[0.17308618,0.00009778965,0.002412886,0.0000684049,0.000038711925,0.00017004034,0.8232517,0.000033440134,0.0008408625],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996774,0.00004877607,0.00003349093,0.0000859253,0.00009191076,0.000062480096],"domain_scores_gemma":[0.9989243,0.00011525211,0.00022843217,0.0002038431,0.00036527627,0.00016296393],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00043774713,0.00053044077,0.00039750495,0.0013452081,0.00032466947,0.00055279606,0.0007528669,0.00050364056,0.0026899355],"category_scores_gemma":[0.0009999422,0.00018415874,0.00038210646,0.002781376,0.00020316741,0.00039645782,0.0007527465,0.0004979477,0.0016899334],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005430603,0.00042768338,0.6695788,0.0005566094,0.0004290912,0.00078651914,0.0005247027,0.016347589,0.005643905,0.001067653,0.27068266,0.0334117],"study_design_scores_gemma":[0.000120634955,0.000062457715,0.9133795,0.00007723531,0.000036019428,0.00012569678,0.00063324434,0.009931597,0.0016634389,0.00034145173,0.073572874,0.00005575214],"about_ca_topic_score_codex":0.053172674,"about_ca_topic_score_gemma":0.06592418,"teacher_disagreement_score":0.053172674,"about_ca_system_score_codex":0.0009180694,"about_ca_system_score_gemma":0.00089230435,"threshold_uncertainty_score":0.10572636},"labels":[],"label_agreement":null},{"id":"W4400041829","doi":"10.1038/s41597-024-03557-w","title":"Computed tomography reconstructions of burrow networks for the Opheliid polychaete, Armandia cirrhosa","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine Biology and Ecology Research","field":"Earth and Planetary Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"Engineering and Physical Sciences Research Council; Research Councils UK; Natural England","keywords":"Burrow; Polychaete; Intraspecific competition; Ecology; Context (archaeology); Biology; Replicate; Range (aeronautics); Habitat; Ecosystem; Paleontology","score_opus":0.0543216792617203,"score_gpt":0.28617384848020994,"score_spread":0.23185216921848964,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400041829","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99376374,0.00007158322,0.0045620617,0.000034984292,0.000003200931,0.000008886049,0.0008586833,0.00012959415,0.00056717114],"genre_scores_gemma":[0.99400043,0.000054261982,0.0048655737,0.000006539376,0.0000015181009,0.000008355005,0.00086399744,0.000022322292,0.00017708515],"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9999044,0.0000104365445,0.000009143678,0.000035827874,0.000025325697,0.000014913945],"domain_scores_gemma":[0.9996891,0.00006759061,0.00008166742,0.000042768785,0.000084886844,0.00003412245],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00014422726,0.00027824205,0.00014468057,0.0009485761,0.00026270252,0.00054940797,0.0003444909,0.0003958996,0.0014960022],"category_scores_gemma":[0.00082587823,0.00036964178,0.00028340958,0.00077025173,0.00031068482,0.00032806495,0.00035640533,0.00029971264,0.00022316002],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006032873,0.0002508844,0.45353267,0.0002507514,0.00025581263,0.0010978169,0.0011861755,0.29035527,0.20768355,0.0015740055,0.00138776,0.041822072],"study_design_scores_gemma":[0.000042463587,0.000111685724,0.67687875,0.00002846727,0.00007398989,0.000787932,0.00072992826,0.31238055,0.007231426,0.0005669257,0.0011077601,0.00006005242],"about_ca_topic_score_codex":0.01424561,"about_ca_topic_score_gemma":0.02948629,"teacher_disagreement_score":0.01424561,"about_ca_system_score_codex":0.00038133792,"about_ca_system_score_gemma":0.00047793405,"threshold_uncertainty_score":0.028325379},"labels":[],"label_agreement":null},{"id":"W4400229728","doi":"10.1038/s41597-024-03570-z","title":"1 Million Segmented Red Blood Cells With 240 K Classified in 9 Shapes and 47 K Patches of 25 Manual Blood Smears","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Digital Imaging for Blood Diseases","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Spinologics (Canada)","funders":"Zagazig University","keywords":"Blood smear; Segmentation; Artificial intelligence; Computer science; Digital pathology; Microscope; Pathology; Automated method; Pattern recognition (psychology); Medicine","score_opus":0.0270678525659667,"score_gpt":0.2544679995981731,"score_spread":0.22740014703220643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400229728","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7080373,0.0030950313,0.041741755,0.0006940224,0.00051337195,0.0006709906,0.22091064,0.01338783,0.01094912],"genre_scores_gemma":[0.46582106,0.0009756407,0.08181583,0.00046219968,0.00013244945,0.00046052856,0.44159874,0.0006517998,0.008081685],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994147,0.000041338448,0.000059454742,0.00019597578,0.00019296726,0.000095576754],"domain_scores_gemma":[0.9994816,0.00006465944,0.000043834716,0.00014696427,0.00020053946,0.00006239355],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004728666,0.00076277135,0.00082858943,0.0010819632,0.0003873509,0.000724184,0.0007055787,0.00079521065,0.0041631274],"category_scores_gemma":[0.0010451145,0.00031174615,0.00063625036,0.0011173233,0.0003543188,0.00040201758,0.0008449825,0.00044088103,0.003392984],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029514716,0.0008470285,0.080129996,0.0014649325,0.0004718033,0.0015110208,0.0004514722,0.018204631,0.18071443,0.0016243245,0.19124179,0.52038705],"study_design_scores_gemma":[0.00032840244,0.0012167885,0.29848108,0.00042491377,0.00041757856,0.008565668,0.0009998444,0.15896934,0.2657617,0.004345104,0.26027045,0.00021909147],"about_ca_topic_score_codex":0.004750986,"about_ca_topic_score_gemma":0.006532206,"teacher_disagreement_score":0.004750986,"about_ca_system_score_codex":0.0005448935,"about_ca_system_score_gemma":0.00059469446,"threshold_uncertainty_score":0.013926983},"labels":[],"label_agreement":null},{"id":"W4400229831","doi":"10.1038/s41597-024-03469-9","title":"HVSMR-2.0: A 3D cardiovascular MR dataset for whole-heart segmentation in congenital heart disease","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Congenital Heart Disease Studies","field":"Medicine","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Biomedical Imaging and Bioengineering; National Heart, Lung, and Blood Institute; National Institute on Aging; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Charles H. Hood Foundation; Natural Sciences and Engineering Research Council of Canada; Harvard Catalyst; Philips; National Institutes of Health; U.S. Department of Health and Human Services; National Institute of Neurological Disorders and Stroke; American Heart Association","keywords":"Heart disease; Segmentation; Surgical planning; Magnetic resonance imaging; Medicine; Medical diagnosis; Radiology; Cardiology; Artificial intelligence; Computer science","score_opus":0.058626895430699585,"score_gpt":0.34471851026982797,"score_spread":0.28609161483912837,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400229831","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04204499,0.0060855704,0.03302892,0.0011048702,0.0007277336,0.0010180135,0.8846532,0.023179397,0.00815738],"genre_scores_gemma":[0.03159311,0.0010900906,0.033012062,0.00041007996,0.00016212427,0.0008373433,0.92936045,0.0015842117,0.0019506346],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99915016,0.00014131478,0.00013185006,0.0002613683,0.00022606645,0.00008929219],"domain_scores_gemma":[0.99879944,0.00035895238,0.0001156488,0.00034677156,0.00024844054,0.00013066662],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013238348,0.0021907955,0.0015876202,0.0028676575,0.00076064345,0.0021693553,0.0029511168,0.0032214276,0.011383084],"category_scores_gemma":[0.0041899164,0.00079573144,0.0017498943,0.0017484208,0.00050250365,0.00064549176,0.0022879841,0.0013834863,0.013987526],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001410204,0.0004377692,0.012661154,0.004013407,0.00063660974,0.0016532941,0.00040993866,0.009265392,0.017301276,0.0018273013,0.83899456,0.11138918],"study_design_scores_gemma":[0.001767904,0.000712553,0.076250754,0.0024572567,0.0007664996,0.014550297,0.0006530531,0.052550714,0.025490124,0.011383318,0.8127894,0.0006281557],"about_ca_topic_score_codex":0.008275444,"about_ca_topic_score_gemma":0.02422554,"teacher_disagreement_score":0.011383084,"about_ca_system_score_codex":0.00066216587,"about_ca_system_score_gemma":0.0013994697,"threshold_uncertainty_score":0.038080215},"labels":[],"label_agreement":null},{"id":"W4400246886","doi":"10.1038/s41597-024-03559-8","title":"Motion-BIDS: an extension to the brain imaging data structure to organize motion data for reproducible research","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"","keywords":"Metadata; Interoperability; Motion (physics); Computer science; Motion capture; Data sharing; Modalities; Software; Information retrieval; Computer vision; World Wide Web; Programming language","score_opus":0.35466429661359333,"score_gpt":0.444939037554456,"score_spread":0.09027474094086269,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400246886","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001264776,0.00034352447,0.96483463,0.0002942356,0.00023094947,0.0007861292,0.0082686,0.02233145,0.0016457424],"genre_scores_gemma":[0.018307183,0.0004589144,0.9538079,0.00031400236,0.00023049384,0.002973255,0.016232833,0.005915235,0.0017600746],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99335206,0.0017156733,0.0015093662,0.0013055988,0.001737874,0.00037941066],"domain_scores_gemma":[0.97946435,0.006834851,0.001367551,0.008166714,0.0029968787,0.0011695812],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.014464763,0.0025793407,0.0026282757,0.0068633226,0.0019732276,0.005541967,0.0055309064,0.0018372979,0.024567282],"category_scores_gemma":[0.045216747,0.0023304648,0.003211324,0.008907203,0.0022858907,0.01087574,0.013965537,0.0032544336,0.017676428],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030612943,0.000595603,0.0067461873,0.0024908758,0.0006049068,0.0009965745,0.0018899834,0.0084368605,0.025985379,0.12183433,0.13510443,0.6922536],"study_design_scores_gemma":[0.0008065286,0.0006843011,0.005217359,0.0007185006,0.00025262398,0.0016661764,0.0007738711,0.11457305,0.03023071,0.32591033,0.5187146,0.0004518778],"about_ca_topic_score_codex":0.0023471753,"about_ca_topic_score_gemma":0.003182042,"teacher_disagreement_score":0.98553526,"about_ca_system_score_codex":0.0009202474,"about_ca_system_score_gemma":0.0037516728,"threshold_uncertainty_score":0.082185745},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reproducibility","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"reproducibility","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4400317002","doi":"10.1038/s41597-024-03585-6","title":"Author Correction: PLAS-20k: Extended Dataset of Protein-Ligand Affinities from MD Simulations for Machine Learning Applications","year":2024,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Machine Learning in Bioinformatics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Affinities; Binding affinities; Ligand (biochemistry); Computer science; Computational biology; Artificial intelligence; Chemistry; Biology; Stereochemistry; Biochemistry; Receptor","score_opus":0.035606005382448236,"score_gpt":0.3239491444120766,"score_spread":0.2883431390296284,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400317002","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009940612,0.0012367144,0.021527857,0.04238935,0.8449912,0.00011980779,0.06867972,0.011886819,0.00817441],"genre_scores_gemma":[0.047755618,0.0065228357,0.10388784,0.075329036,0.09230309,0.0018196409,0.22083318,0.071268365,0.38028035],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98722446,0.0017823917,0.0024555193,0.0021973578,0.00581961,0.00052068656],"domain_scores_gemma":[0.90326357,0.025716769,0.004298662,0.011681129,0.05268492,0.002354922],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010546492,0.0026500646,0.0026580745,0.004538199,0.0039446275,0.0057405005,0.0045510735,0.004218261,0.1648825],"category_scores_gemma":[0.14823735,0.0016689681,0.0019824065,0.005395102,0.0022382936,0.003245132,0.0054273424,0.010251668,0.12902004],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040350827,0.000005645135,0.00011792032,0.00016809607,0.000016197639,0.00008863998,0.00003334257,0.00012481015,0.00011451651,0.0007985096,0.9935051,0.004986822],"study_design_scores_gemma":[0.00006219706,0.000013952955,0.0006252991,0.0003328837,0.0000521314,0.0004226211,0.00005978907,0.00063403446,0.0012256372,0.0037029502,0.9928047,0.00006381299],"about_ca_topic_score_codex":0.008690524,"about_ca_topic_score_gemma":0.011386277,"teacher_disagreement_score":0.1648825,"about_ca_system_score_codex":0.0033255115,"about_ca_system_score_gemma":0.007052564,"threshold_uncertainty_score":0.55158687},"labels":[],"label_agreement":null},{"id":"W4400356338","doi":"10.1038/s41597-024-03532-5","title":"Climate data for building simulations with urban heat island effects and nature-based solutions","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Urban Heat Island Mitigation","field":"Environmental Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"National Research Council Canada","funders":"Alliance de recherche numérique du Canada","keywords":"Urban heat island; Environmental science; Ecology; Computer science; Geography; Meteorology; Biology","score_opus":0.02888947267074084,"score_gpt":0.2857401626545027,"score_spread":0.25685068998376187,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400356338","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24383703,0.00027889066,0.032768164,0.0015907367,0.0005121057,0.0005608409,0.68714035,0.008123053,0.025188813],"genre_scores_gemma":[0.4277619,0.00028552412,0.04865095,0.00020958205,0.000057461828,0.00088529906,0.51904523,0.000846022,0.0022580116],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99959344,0.00010368463,0.00004215416,0.00007648252,0.00012801535,0.00005618842],"domain_scores_gemma":[0.99846137,0.0003723603,0.00009811177,0.0005003116,0.00039268858,0.00017528854],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010013946,0.0008360748,0.00049351255,0.0006964845,0.0005288437,0.00076556293,0.001693023,0.0012342087,0.007032064],"category_scores_gemma":[0.0026110339,0.0005032238,0.0011727673,0.0022047868,0.0003674391,0.0009897269,0.0006831644,0.0014351064,0.0023536514],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022632345,0.0004999762,0.019325739,0.00032546534,0.00018396598,0.00016976209,0.00016295872,0.8844093,0.0018794473,0.005165315,0.07532764,0.012324066],"study_design_scores_gemma":[0.00037767863,0.00014447082,0.027546851,0.000096811156,0.00005689015,0.0000782921,0.0003635234,0.8488096,0.005940912,0.0071654757,0.10930232,0.0001172167],"about_ca_topic_score_codex":0.038596515,"about_ca_topic_score_gemma":0.074752115,"teacher_disagreement_score":0.038596515,"about_ca_system_score_codex":0.0010183206,"about_ca_system_score_gemma":0.0014748914,"threshold_uncertainty_score":0.07674372},"labels":[],"label_agreement":null},{"id":"W4400382322","doi":"10.1038/s41597-024-03577-6","title":"40 years of forest dynamics and tree demography in an intact tropical forest at M’Baïki in central Africa","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"African Botany and Ecology Studies","field":"Agricultural and Biological Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"FP7 International Cooperation; Agence Française de Développement; Centre de Coopération Internationale en Recherche Agronomique pour le Développement; Canadian Institute for Advanced Research","keywords":"Diameter at breast height; Geography; Forest dynamics; Forestry; Biomass (ecology); Population; Biodiversity; Tropics; Ecology; Basal area; Forest inventory; Agroforestry; Environmental science; Demography; Forest management; Biology","score_opus":0.032901281414505074,"score_gpt":0.2388499167981641,"score_spread":0.20594863538365904,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400382322","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99983144,0.000017319438,0.0000100784455,0.000004053899,3.4763954e-7,0.0000031712177,0.00007930547,5.6342503e-7,0.000053627056],"genre_scores_gemma":[0.99946433,0.00003791936,0.000052923835,0.0000069313796,0.0000015455088,0.000008188234,0.00025961435,4.743944e-7,0.00016816401],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998679,0.000025509868,0.000006679546,0.000027729044,0.000012780157,0.000059445123],"domain_scores_gemma":[0.9997969,0.000030509063,0.00006626333,0.000015478887,0.000019912828,0.00007095396],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00033547022,0.00021065013,0.0001795557,0.0007671308,0.0006602904,0.00028660396,0.00024104105,0.0001992996,0.000606031],"category_scores_gemma":[0.00037359292,0.00013404244,0.000115019655,0.0007459931,0.00045340153,0.00026393094,0.0003471482,0.00017036925,0.00012282406],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005396864,0.00029363582,0.9789362,0.000030576964,0.00003748934,0.0011996676,0.003718708,0.0003828636,0.006301335,0.00008444366,0.00020642216,0.008268908],"study_design_scores_gemma":[0.00000547977,0.00006611097,0.9988918,0.0000026476184,0.0000039311826,0.000075014636,0.00047831933,0.000115137074,0.00010923427,0.0000091780785,0.00024050809,0.0000026501166],"about_ca_topic_score_codex":0.046325553,"about_ca_topic_score_gemma":0.1030174,"teacher_disagreement_score":0.046325553,"about_ca_system_score_codex":0.0008023855,"about_ca_system_score_gemma":0.0002859361,"threshold_uncertainty_score":0.092111826},"labels":[],"label_agreement":null},{"id":"W4400535776","doi":"10.1038/s41597-024-03436-4","title":"The Canadian Fire Spread Dataset","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Natural Resources Canada; Canadian Forest Service","funders":"Canadian Forest Service; U.S. Forest Service; Government of Alberta; Rocky Mountain Research Station; U.S. Department of Agriculture","keywords":"Visible Infrared Imaging Radiometer Suite; Moderate-resolution imaging spectroradiometer; Environmental science; Remote sensing; Kriging; Radiometer; Meteorology; Fire detection; Fire regime; Range (aeronautics); Suite; Satellite; Computer science; Geography; Machine learning; Ecology","score_opus":0.02323385016940095,"score_gpt":0.2577663515867854,"score_spread":0.23453250141738446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400535776","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018085152,0.00014585191,0.00022987537,0.00007226572,0.000021987651,0.000047338992,0.99569243,0.00019870601,0.0017829209],"genre_scores_gemma":[0.002798455,0.00010675386,0.0006996525,0.000032952496,0.0000062519875,0.00007258698,0.99515533,0.000033810564,0.0010940997],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998955,0.00006773204,0.00008521678,0.00022207275,0.0004678082,0.00020217523],"domain_scores_gemma":[0.99653816,0.0002750984,0.00019663939,0.00039275349,0.0022768003,0.00032057348],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008302895,0.0017231158,0.0009473509,0.005316326,0.0018423883,0.0016871812,0.002824033,0.0010372294,0.017244475],"category_scores_gemma":[0.0041470234,0.00043024958,0.000990038,0.009917087,0.00042830946,0.00072159065,0.001053359,0.001527706,0.010403443],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011332607,0.000049567403,0.010493125,0.00031312718,0.000103931285,0.00009849198,0.00006179751,0.0016650308,0.00026558366,0.001141476,0.97306293,0.012631657],"study_design_scores_gemma":[0.00018551739,0.000024478384,0.06602966,0.00027412936,0.00008754058,0.00017836933,0.00029053935,0.0061023985,0.0008158995,0.0008970612,0.9249984,0.000115866635],"about_ca_topic_score_codex":0.9503651,"about_ca_topic_score_gemma":0.97742724,"teacher_disagreement_score":0.049634874,"about_ca_system_score_codex":0.011645045,"about_ca_system_score_gemma":0.018383417,"threshold_uncertainty_score":0.09985429},"labels":[],"label_agreement":null},{"id":"W4401114855","doi":"10.1038/s41597-024-03639-9","title":"Mapping global land conversion pressure to support conservation planning","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Conservation, Biodiversity, and Resource Management","field":"Environmental Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Land cover; Environmental resource management; Scale (ratio); Climate change; Land use; Environmental science; Land use, land-use change and forestry; Biodiversity; Geography; Ecology; Cartography","score_opus":0.05460364334583896,"score_gpt":0.2708934861808216,"score_spread":0.21628984283498265,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401114855","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.74172604,0.0022683677,0.04979029,0.0049169445,0.00014946853,0.0003817519,0.036253296,0.0019578454,0.16255595],"genre_scores_gemma":[0.9432047,0.0009900061,0.044377685,0.00014382618,0.000041306,0.00018577962,0.00958795,0.00012701431,0.001341754],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996228,0.00012218968,0.000023716142,0.000051201325,0.00012135192,0.00005868951],"domain_scores_gemma":[0.9989159,0.00031074393,0.00026306603,0.00013266716,0.00025322664,0.00012434789],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012847504,0.00042663733,0.00025184912,0.0040371493,0.00031246597,0.001806633,0.00035051646,0.00021715379,0.0040248],"category_scores_gemma":[0.0038183844,0.00012750665,0.00019311186,0.009011984,0.00030637186,0.0015034461,0.0010593695,0.0005855955,0.00046144484],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008902445,0.0001619444,0.48858717,0.0003307338,0.00020641559,0.00028470712,0.0015106174,0.052996054,0.002586982,0.020471983,0.036565784,0.39620858],"study_design_scores_gemma":[0.000034528377,0.00010254705,0.81098044,0.0002932576,0.000093457005,0.00012899059,0.005688456,0.0619446,0.0013205517,0.024093814,0.095231645,0.00008775103],"about_ca_topic_score_codex":0.027395774,"about_ca_topic_score_gemma":0.04254138,"teacher_disagreement_score":0.027395774,"about_ca_system_score_codex":0.0011490111,"about_ca_system_score_gemma":0.0014056113,"threshold_uncertainty_score":0.054472625},"labels":[],"label_agreement":null},{"id":"W4401283762","doi":"10.1038/s41597-024-03685-3","title":"A Global Dataset of Potential Chloride Deposits on Mars as Identified by TGO CaSSIS","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Planetary Science and Exploration","field":"Physics and Astronomy","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Space Agency; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; University of Bern; National Aeronautics and Space Administration","keywords":"Hesperian; Mars Exploration Program; Geology; Noachian; Geochemistry; Astrobiology; Chloride; Martian; Earth science; Chemistry","score_opus":0.025226119467857896,"score_gpt":0.2861489688470883,"score_spread":0.2609228493792304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401283762","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.31732297,0.00092835346,0.0011662586,0.0004862735,0.000107580265,0.000055155473,0.6719003,0.0014732912,0.006559824],"genre_scores_gemma":[0.20544252,0.00041702716,0.0034111326,0.00009652399,0.000044410535,0.00010035892,0.78862005,0.00008569291,0.0017823847],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99982196,0.000013757944,0.000015784004,0.000060409464,0.000051556606,0.00003650211],"domain_scores_gemma":[0.99966514,0.000038590242,0.00005738436,0.000063111176,0.00013000281,0.000045746525],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00016987404,0.0005765878,0.0003557478,0.001662625,0.000388614,0.00049079774,0.0005823164,0.00073732954,0.0026973523],"category_scores_gemma":[0.00061289856,0.00016104856,0.00046417542,0.0020424072,0.0002484766,0.00038881172,0.00084187184,0.00041446844,0.0021306442],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008277843,0.0003836557,0.34910962,0.0016438196,0.00058498414,0.0030146483,0.00083926826,0.029121403,0.024067234,0.0014056888,0.5106724,0.078329414],"study_design_scores_gemma":[0.00017101405,0.00013068932,0.67871904,0.0002497471,0.000160993,0.00080662034,0.0015235908,0.025717376,0.007716806,0.0011087874,0.28357962,0.00011571301],"about_ca_topic_score_codex":0.03433088,"about_ca_topic_score_gemma":0.06071153,"teacher_disagreement_score":0.03433088,"about_ca_system_score_codex":0.0005377277,"about_ca_system_score_gemma":0.00058038096,"threshold_uncertainty_score":0.0682621},"labels":[],"label_agreement":null},{"id":"W4401333304","doi":"10.1038/s41597-024-03499-3","title":"Nine months of daily LiDAR, orthophotos and MetOcean data from the eroding soft cliff coast at Happisburgh, UK","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Coastal and Marine Dynamics","field":"Earth and Planetary Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Tellabs (Canada)","funders":"Engineering and Physical Sciences Research Council; Natural Environment Research Council; Research Councils UK; Sight Research UK","keywords":"Lidar; Shore; Transect; Cliff; Geology; Orthophoto; Subaerial; Submarine pipeline; Oceanography; Environmental science; Remote sensing","score_opus":0.041130662941605535,"score_gpt":0.24798656887872955,"score_spread":0.20685590593712402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401333304","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47340572,0.0002810973,0.00086536485,0.00014191867,0.000049076567,0.00012171149,0.5190831,0.00018371701,0.0058683157],"genre_scores_gemma":[0.38891456,0.00054246315,0.003560291,0.00011168468,0.00003321567,0.00033634636,0.5967833,0.000091508366,0.009626602],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996604,0.000021855072,0.000046263485,0.00009042415,0.00011177473,0.000069299],"domain_scores_gemma":[0.9990201,0.0000808498,0.00020074224,0.00013764403,0.00044096986,0.00011963695],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001818561,0.0003261695,0.00036991533,0.0017144384,0.00028474862,0.00063531956,0.0003930548,0.0004537163,0.005218249],"category_scores_gemma":[0.0008350752,0.0002298799,0.00016464104,0.003459494,0.00025936315,0.00036393394,0.00059084216,0.00023385939,0.0031868143],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000514788,0.00023944292,0.74766946,0.0012336238,0.0001936961,0.002552536,0.0033228574,0.0074748187,0.010695905,0.0004138213,0.14009032,0.085598774],"study_design_scores_gemma":[0.00002051748,0.00005010867,0.9473167,0.00011104727,0.0000152507,0.00018475862,0.002018895,0.0012808163,0.0006930085,0.000052211763,0.048228297,0.000028297038],"about_ca_topic_score_codex":0.1911017,"about_ca_topic_score_gemma":0.4546807,"teacher_disagreement_score":0.1911017,"about_ca_system_score_codex":0.0006178027,"about_ca_system_score_gemma":0.0008557988,"threshold_uncertainty_score":0.37997872},"labels":[],"label_agreement":null},{"id":"W4401419620","doi":"10.1038/s41597-024-03691-5","title":"Where should we go - Estimating travel times for modelling accessibility to 24-hour emergency departments in Canada","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Urban Transport and Accessibility","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Carleton University","funders":"Government of Ontario","keywords":"Census; Travel time; Population; Geography; Service (business); Transport engineering; Subdivision; Health care; Business; Computer science; Medicine; Environmental health; Marketing; Engineering; Economic growth","score_opus":0.15627696841106278,"score_gpt":0.38433990418241,"score_spread":0.2280629357713472,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401419620","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8974384,0.0027411988,0.03999428,0.0025062535,0.00010605533,0.0006977917,0.034057368,0.00057974056,0.021878958],"genre_scores_gemma":[0.97108275,0.00084761344,0.018038444,0.000072529154,0.000008224429,0.00016412316,0.005145536,0.000057828267,0.004582874],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99935895,0.0001819781,0.000039509785,0.00013768744,0.00011548593,0.00016641],"domain_scores_gemma":[0.9982054,0.00057102874,0.00012517028,0.00007848523,0.0008512601,0.00016856614],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012517781,0.00048179732,0.0005489559,0.0020022448,0.0014034676,0.0024692933,0.0016248239,0.00069062115,0.00328178],"category_scores_gemma":[0.007357313,0.00032541348,0.0011243112,0.0036003683,0.00059617904,0.0008872096,0.0009510303,0.0008176149,0.00033457574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029337313,0.00016518922,0.20961832,0.00044233262,0.00037883478,0.00027437345,0.0010577843,0.7001534,0.0004125369,0.01798317,0.011243687,0.057977032],"study_design_scores_gemma":[0.00005489454,0.00005205336,0.10218623,0.00025580497,0.00012795854,0.000059604732,0.0019821282,0.8793751,0.0003296051,0.0034646671,0.011998729,0.000113168404],"about_ca_topic_score_codex":0.99593735,"about_ca_topic_score_gemma":0.99502665,"teacher_disagreement_score":0.033867814,"about_ca_system_score_codex":0.033867814,"about_ca_system_score_gemma":0.03639198,"threshold_uncertainty_score":0.24572933},"labels":[],"label_agreement":null},{"id":"W4401455833","doi":"10.1038/s41597-024-03695-1","title":"MSPB: a longitudinal multi-sensor dataset with phenotypic trait measurements from honey bees","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Insect and Arachnid Ecology and Behavior","field":"Biochemistry, Genetics and Molecular Biology","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval; Institut National de la Recherche Scientifique; Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Varroa destructor; Biology; Apiary; Honey bee; Beehive; Population; Zoology; Ecology; Demography","score_opus":0.10953250777004635,"score_gpt":0.3222916178320017,"score_spread":0.21275911006195536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401455833","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12611695,0.0011731941,0.0057665026,0.00044312293,0.00019974879,0.00018311162,0.8608986,0.0029994985,0.002219209],"genre_scores_gemma":[0.09246961,0.00027836623,0.008759914,0.00013450863,0.00003268152,0.00026736772,0.89675695,0.00009959228,0.0012008992],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9995036,0.000046500645,0.000041461917,0.00016292305,0.00015010018,0.00009533709],"domain_scores_gemma":[0.9990941,0.00011656687,0.000073314906,0.00019152506,0.00041061998,0.00011372184],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046676886,0.0014350824,0.0008891039,0.0016894789,0.00064212375,0.00073269324,0.0014707785,0.0011823598,0.0018142255],"category_scores_gemma":[0.0014926688,0.00030984977,0.00094144826,0.0022153426,0.0004035325,0.000555093,0.00095055875,0.0010033327,0.0019521604],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013712028,0.0012967264,0.29244754,0.0031540715,0.0013379744,0.00200971,0.0009791603,0.032582093,0.03626371,0.0013135256,0.5070165,0.12022792],"study_design_scores_gemma":[0.00023002666,0.0003781588,0.7400973,0.00034644347,0.00021081032,0.00078010355,0.001150441,0.057270784,0.00946772,0.0011770836,0.18861255,0.00027858722],"about_ca_topic_score_codex":0.20649597,"about_ca_topic_score_gemma":0.34966508,"teacher_disagreement_score":0.20649597,"about_ca_system_score_codex":0.00139468,"about_ca_system_score_gemma":0.0015540156,"threshold_uncertainty_score":0.4105881},"labels":[],"label_agreement":null},{"id":"W4401591794","doi":"10.1038/s41597-024-03683-5","title":"The NACOB multi-surface walking dataset","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Gait Recognition and Analysis","field":"Engineering","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institut national de psychiatrie légale Philippe-Pinel; Université de Montréal; Centre Hospitalier Universitaire Sainte-Justine; McGill University","funders":"Fonds de Recherche du Québec - Santé; Canada First Research Excellence Fund; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Computer science; Information retrieval; Environmental science","score_opus":0.06017147036614638,"score_gpt":0.29959759865658775,"score_spread":0.23942612829044135,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401591794","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1560343,0.0012567914,0.0105713485,0.0004357912,0.00044762457,0.0006272437,0.8186168,0.0047412026,0.0072689215],"genre_scores_gemma":[0.07568181,0.00022980619,0.008620412,0.00009064526,0.000037279795,0.00064309704,0.9124059,0.00010727846,0.0021837612],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996464,0.00005374066,0.000041933454,0.00009713281,0.000109003646,0.00005175064],"domain_scores_gemma":[0.9996037,0.00005968216,0.000032089378,0.00009745499,0.00014995354,0.000057294983],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003391582,0.0012877703,0.0008758657,0.0013761367,0.0005107367,0.0005063255,0.0014220049,0.0013018816,0.0053392886],"category_scores_gemma":[0.0013698373,0.00019121294,0.000884638,0.0012525354,0.00027742778,0.00032531237,0.0009686461,0.0006311009,0.0059003695],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018664231,0.0015314755,0.061154645,0.0027423825,0.00056578335,0.0018527295,0.00037526342,0.014688113,0.010186253,0.0015763602,0.70958245,0.19387808],"study_design_scores_gemma":[0.00085942465,0.0013394312,0.34714505,0.0009567062,0.00029165228,0.003278855,0.0016602625,0.09562961,0.009247302,0.0052564195,0.5340106,0.00032463888],"about_ca_topic_score_codex":0.012110489,"about_ca_topic_score_gemma":0.042150762,"teacher_disagreement_score":0.012110489,"about_ca_system_score_codex":0.0003157224,"about_ca_system_score_gemma":0.0006112453,"threshold_uncertainty_score":0.024080038},"labels":[],"label_agreement":null},{"id":"W4401659238","doi":"10.1038/s41597-024-03679-1","title":"EDARA: An ERA5-based Dataset for Atmospheric River Analysis","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"","keywords":"Precipitation; Environmental science; Troposphere; Flooding (psychology); Climatology; Range (aeronautics); Resource (disambiguation); Meteorology; Geography; Computer science; Geology","score_opus":0.0662248988126431,"score_gpt":0.32506288077919315,"score_spread":0.25883798196655006,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401659238","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0010624808,0.00007747835,0.00050177914,0.00006239765,0.0000425688,0.00002097944,0.9968714,0.0008333489,0.0005275843],"genre_scores_gemma":[0.0025263033,0.00005761858,0.0015887817,0.00003485085,0.0000151111335,0.000091068636,0.99529606,0.000121538105,0.00026861433],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992986,0.00010752013,0.00013794171,0.00019056226,0.00017091227,0.00009446056],"domain_scores_gemma":[0.99899596,0.00018840437,0.00014681515,0.00025321412,0.00031464023,0.000100970705],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00070591294,0.0011813152,0.0010404518,0.0019612256,0.00046576327,0.0014453483,0.0022163792,0.0015458858,0.012817961],"category_scores_gemma":[0.002491277,0.00046695952,0.0014656285,0.0044599837,0.00023297587,0.0011772122,0.0012935215,0.0016313773,0.0140452515],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018624771,0.00010020899,0.009340862,0.0011394469,0.00028248792,0.00016196627,0.00011438901,0.008504071,0.0016776344,0.0020917738,0.9641309,0.012270032],"study_design_scores_gemma":[0.00044663463,0.00004000071,0.032388244,0.0002585569,0.00009176149,0.00010778254,0.00018709652,0.011859586,0.0019185507,0.0028838618,0.9496841,0.00013381989],"about_ca_topic_score_codex":0.022782164,"about_ca_topic_score_gemma":0.026683146,"teacher_disagreement_score":0.022782164,"about_ca_system_score_codex":0.00065107737,"about_ca_system_score_gemma":0.0013667399,"threshold_uncertainty_score":0.045299113},"labels":[],"label_agreement":null},{"id":"W4401666731","doi":"10.1038/s41597-024-03722-1","title":"ANPHY-Sleep: an Open Sleep Database from Healthy Adults Using High-Density Scalp Electroencephalogram","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Sleep and Wakefulness Research","field":"Neuroscience","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University Health Centre; McGill University; Montreal Neurological Institute and Hospital","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Montreal Neurological Institute and Hospital","keywords":"Sleep (system call); Scalp; Electroencephalography; Database; Audiology; Medicine; Psychology; Computer science; Neuroscience; Dermatology","score_opus":0.17170178281590792,"score_gpt":0.4042917301922292,"score_spread":0.23258994737632127,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401666731","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07418681,0.0017959286,0.008863676,0.00028042123,0.00017910848,0.0006975552,0.9082579,0.0028882339,0.0028503903],"genre_scores_gemma":[0.039430507,0.000541189,0.0070662005,0.00014643135,0.000063214975,0.0010705749,0.9501503,0.0001510332,0.0013806572],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99958163,0.000059880673,0.000105191415,0.0001332889,0.00008286678,0.000037116737],"domain_scores_gemma":[0.9990957,0.00015343625,0.00012857569,0.00022221694,0.00026508953,0.00013494446],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005829317,0.0009411729,0.0008560511,0.0018383869,0.00036706304,0.000738985,0.001377995,0.0010975839,0.0055427058],"category_scores_gemma":[0.0022977719,0.00024921543,0.00077982363,0.0013411709,0.00021936905,0.0006268608,0.0015567977,0.00045950845,0.004314719],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00458685,0.0007394174,0.108781345,0.0054962914,0.0012641046,0.0026198179,0.0008951654,0.0034754907,0.015239291,0.0019146571,0.6439914,0.2109962],"study_design_scores_gemma":[0.0013895677,0.0010377702,0.45267835,0.001128792,0.00075352547,0.0048309793,0.0010558467,0.010394828,0.008973594,0.0037295492,0.5136802,0.0003470664],"about_ca_topic_score_codex":0.007484901,"about_ca_topic_score_gemma":0.017285144,"teacher_disagreement_score":0.007484901,"about_ca_system_score_codex":0.00036554417,"about_ca_system_score_gemma":0.00086965284,"threshold_uncertainty_score":0.01854217},"labels":[],"label_agreement":null},{"id":"W4401915925","doi":"10.1038/s41597-024-03611-7","title":"A dataset of acoustic measurements from soundscapes collected worldwide during the COVID-19 pandemic","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Noise Effects and Management","field":"Health Professions","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital; University of Toronto; Parks Canada; Université du Québec à Trois-Rivières; Environment and Climate Change Canada; Carleton University; University of Alberta","funders":"Leverhulme Trust","keywords":"Soundscape; Metadata; Pandemic; Data science; Data collection; Usability; Computer science; Coronavirus disease 2019 (COVID-19); Geography; Sound (geography); World Wide Web; Human–computer interaction; Oceanography; Sociology; Geology; Social science","score_opus":0.26853938945134587,"score_gpt":0.4603518546268951,"score_spread":0.19181246517554923,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401915925","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18664682,0.00095991406,0.009758157,0.00076271885,0.0007370235,0.0004436287,0.7849568,0.0040099407,0.011725013],"genre_scores_gemma":[0.13268876,0.00049675634,0.01077995,0.00030772877,0.00024339171,0.00089322386,0.8500706,0.00029835556,0.004221304],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984819,0.00030341194,0.00020020099,0.0003684552,0.0004763265,0.0001697147],"domain_scores_gemma":[0.9969227,0.0007626141,0.00024642886,0.00064456917,0.0010590841,0.00036456087],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00087967707,0.0010023224,0.0006939433,0.0029749973,0.00067884586,0.0011023486,0.0010910762,0.0015568036,0.003499502],"category_scores_gemma":[0.0037113023,0.00026871718,0.0007012779,0.00338632,0.0005592358,0.0010029619,0.0023132327,0.000976937,0.0056495992],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001085654,0.0008828763,0.1443035,0.005046293,0.0006047293,0.0022114648,0.0049141035,0.00925253,0.022776684,0.0020152694,0.62467337,0.18223357],"study_design_scores_gemma":[0.00012895424,0.000387769,0.61733305,0.00079660467,0.00018818228,0.0010288274,0.00713791,0.010348582,0.0070142546,0.0018875396,0.3534398,0.00030854758],"about_ca_topic_score_codex":0.019950705,"about_ca_topic_score_gemma":0.03630907,"teacher_disagreement_score":0.019950705,"about_ca_system_score_codex":0.0005944876,"about_ca_system_score_gemma":0.0010137763,"threshold_uncertainty_score":0.039669156},"labels":[],"label_agreement":null},{"id":"W4402078561","doi":"10.1038/s41597-024-03771-6","title":"A synthetic vulnerable population dataset for fine scale geographical equity analysis and urban planning","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Health disparities and outcomes","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université de Sherbrooke; Transport Canada","funders":"","keywords":"Comparability; Equity (law); Vulnerability (computing); Scale (ratio); Population; Social equality; Environmental resource management; Context (archaeology); Risk analysis (engineering); Public economics; Environmental planning; Data science; Business; Computer science; Geography; Economics; Political science; Sociology","score_opus":0.0990967556283731,"score_gpt":0.4382109512222689,"score_spread":0.3391141955938958,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402078561","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.051210444,0.00024760788,0.018275918,0.00088043435,0.000114835704,0.0009682922,0.9146104,0.00086684735,0.012825133],"genre_scores_gemma":[0.099090725,0.0002662717,0.031217726,0.00020641429,0.000028660777,0.0013304333,0.86446613,0.000109220804,0.0032843924],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992681,0.00012956893,0.000058788875,0.00013524402,0.00026397017,0.00014421316],"domain_scores_gemma":[0.9977731,0.00035367205,0.0001718098,0.00047688553,0.0009906295,0.0002338613],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001212954,0.00045703395,0.00042376335,0.0029759577,0.0012728573,0.0009694241,0.0017518781,0.00066144404,0.007024842],"category_scores_gemma":[0.006781759,0.00022458911,0.0007346258,0.0056774323,0.00040157157,0.0004460425,0.0017880669,0.0009772805,0.0019588666],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038473282,0.00033129155,0.15849183,0.00074224325,0.00034127143,0.0005155534,0.001239535,0.031592116,0.0014675955,0.023754112,0.6625715,0.11856832],"study_design_scores_gemma":[0.00025508425,0.00008118371,0.28068653,0.0006151281,0.0001350868,0.00035930544,0.0035168545,0.0706729,0.0020756696,0.01465936,0.6267466,0.00019631111],"about_ca_topic_score_codex":0.80558646,"about_ca_topic_score_gemma":0.88007385,"teacher_disagreement_score":0.80558646,"about_ca_system_score_codex":0.0072902367,"about_ca_system_score_gemma":0.014144737,"threshold_uncertainty_score":0.39111692},"labels":[],"label_agreement":null},{"id":"W4402118623","doi":"10.1038/s41597-024-03785-0","title":"A comprehensive dataset for digital restoration of Dunhuang murals","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"3D Surveying and Cultural Heritage","field":"Earth and Planetary Sciences","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Natural Science Foundation of China","keywords":"Mural; Computer science; Cultural heritage; Inpainting; Segmentation; Safeguarding; Digital preservation; Digital image; Resource (disambiguation); Artificial intelligence; Visual arts; Image processing; World Wide Web; Archaeology; Image (mathematics); History; Art","score_opus":0.15001813008581763,"score_gpt":0.32049508902344986,"score_spread":0.17047695893763223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402118623","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16833998,0.009826011,0.014150131,0.0011763808,0.0012383276,0.0009365751,0.7516889,0.0136740375,0.038969684],"genre_scores_gemma":[0.10313455,0.0012654952,0.019691655,0.00022368696,0.00011709248,0.00029082663,0.8678322,0.00043953318,0.007004967],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99942017,0.000062049476,0.000049041566,0.00014873261,0.00020959722,0.000110438705],"domain_scores_gemma":[0.99964786,0.00004214443,0.000031068692,0.000121719604,0.00011729663,0.00003987648],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00034080603,0.0011108404,0.0007955366,0.0026295946,0.0008519292,0.0009999188,0.0011983418,0.0011380294,0.0068173623],"category_scores_gemma":[0.0010073411,0.0002509589,0.0006930154,0.0021087774,0.00051667815,0.00056197663,0.0011581166,0.00069497875,0.006183919],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006469613,0.00053351466,0.016008517,0.003168323,0.00028065252,0.0018245843,0.00080101413,0.0066494998,0.027879288,0.0019726485,0.6876329,0.252602],"study_design_scores_gemma":[0.000107010295,0.00014760775,0.07208903,0.00055693835,0.00011873054,0.0021991709,0.0018854013,0.0106511,0.013216158,0.0013189546,0.89759475,0.000115279814],"about_ca_topic_score_codex":0.017007409,"about_ca_topic_score_gemma":0.05895024,"teacher_disagreement_score":0.017007409,"about_ca_system_score_codex":0.0006286869,"about_ca_system_score_gemma":0.00067451445,"threshold_uncertainty_score":0.033816814},"labels":[],"label_agreement":null},{"id":"W4402207096","doi":"10.1038/s41597-024-03805-z","title":"Effects of changing farming practices in African agriculture","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Agricultural Innovations and Practices","field":"Agricultural and Biological Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada; Ouranos","funders":"Foreign Agricultural Service; International Fund for Agricultural Development; European Commission; Consortium of International Agricultural Research Centers; U.S. Department of Agriculture","keywords":"Agriculture; Sustainability; Productivity; Psychological resilience; Business; Resilience (materials science); Natural resource economics; Psychological intervention; Livestock; Environmental resource management; Agricultural productivity; Environmental planning; Geography; Economic growth; Economics; Ecology","score_opus":0.07111008588108822,"score_gpt":0.3088407884436994,"score_spread":0.23773070256261117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402207096","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98136055,0.0047510215,0.00055114756,0.0018674426,0.00004720686,0.00003353443,0.008085583,0.000014898734,0.0032886132],"genre_scores_gemma":[0.99609643,0.0018607522,0.0003391546,0.00010740961,0.000018073155,0.000041874053,0.0012815177,0.000006414182,0.0002485129],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9970176,0.001647697,0.00019010427,0.0004239808,0.00029125132,0.00042939602],"domain_scores_gemma":[0.9939546,0.0027915833,0.0023509115,0.00036250168,0.000284991,0.000255409],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002215483,0.00027461225,0.00031466648,0.0013554372,0.00058983784,0.00088258117,0.00039033603,0.000479717,0.0036318258],"category_scores_gemma":[0.012935838,0.00017216495,0.000489309,0.0048133372,0.0008579428,0.0012557404,0.0021965401,0.00094296585,0.00025464222],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004486783,0.00015767328,0.9214825,0.0009292399,0.0006682935,0.0005027763,0.0049425787,0.0029109758,0.0011467102,0.0028150594,0.004471062,0.05952452],"study_design_scores_gemma":[0.000009524675,0.000059224858,0.9881413,0.00036847222,0.000082497194,0.00008326618,0.0025962421,0.00046048072,0.00023451023,0.0008946026,0.007058048,0.000011816573],"about_ca_topic_score_codex":0.025494067,"about_ca_topic_score_gemma":0.025260245,"teacher_disagreement_score":0.025494067,"about_ca_system_score_codex":0.0014816166,"about_ca_system_score_gemma":0.0009252839,"threshold_uncertainty_score":0.050691366},"labels":[],"label_agreement":null},{"id":"W4402644711","doi":"10.1038/s41597-024-03869-x","title":"High-throughput drug screening identifies novel therapeutics for Low Grade Serous Ovarian Carcinoma","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ovarian cancer diagnosis and treatment","field":"Medicine","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; BC Cancer Agency","funders":"Therapeutic Innovation Australia; BC Cancer Foundation; Medical Research Future Fund; Australian Cancer Research Foundation; Australian Government; Ovarian Cancer Canada; Marshfield Clinic Research Foundation; Cancer Research Society; Peter MacCallum Cancer Centre; Victorian Cancer Agency","keywords":"Serous fluid; Ovarian cancer; Serous carcinoma; Ovarian carcinoma; Oncology; Clear cell; Ovary; Clear cell carcinoma; Internal medicine; Medicine; Cancer research; Drug; Cancer; Carcinoma; Pharmacology","score_opus":0.11066439839271523,"score_gpt":0.34568069755072056,"score_spread":0.23501629915800532,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402644711","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96357155,0.006092047,0.018380538,0.0007895236,0.000045733574,0.000292244,0.0067567094,0.0010157876,0.003055861],"genre_scores_gemma":[0.97593397,0.003296102,0.012992276,0.00025737652,0.000014003216,0.00012234953,0.0056904657,0.000055130633,0.0016384335],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9997125,0.000046259458,0.00002001317,0.000047759673,0.00013587155,0.000037555055],"domain_scores_gemma":[0.9996911,0.00012339615,0.00005509215,0.000037250345,0.00006750887,0.000025683588],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005227824,0.0005480437,0.00093661156,0.00078828016,0.000288094,0.001028568,0.00037557518,0.00050669437,0.001834245],"category_scores_gemma":[0.0006618274,0.0002435468,0.00066156604,0.0006601872,0.0002912079,0.00038651045,0.0003547357,0.000653133,0.0005495632],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010123076,0.000518567,0.0116331475,0.0009886444,0.00023095804,0.0004102488,0.00008014678,0.019123249,0.9042034,0.00063017063,0.0029123947,0.05825662],"study_design_scores_gemma":[0.00016756025,0.0022793869,0.022528362,0.00006730989,0.00030170052,0.0010864073,0.00011874077,0.047112558,0.9146763,0.0007194485,0.010857693,0.00008445378],"about_ca_topic_score_codex":0.0018454163,"about_ca_topic_score_gemma":0.004583257,"teacher_disagreement_score":0.0018454163,"about_ca_system_score_codex":0.00064572284,"about_ca_system_score_gemma":0.0006483176,"threshold_uncertainty_score":0.006136179},"labels":[],"label_agreement":null},{"id":"W4402645290","doi":"10.1038/s41597-024-03871-3","title":"A global dataset of gross nitrogen transformation rates across terrestrial ecosystems","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Carbon and Nitrogen Dynamics","field":"Agricultural and Biological Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Research Foundation of Korea; Yonsei University; National Science Foundation","keywords":"Ecosystem; Environmental science; Transformation (genetics); Terrestrial ecosystem; Nitrogen; Nitrogen cycle; Ecology; Biology; Chemistry","score_opus":0.05211844908467723,"score_gpt":0.31494454221743395,"score_spread":0.26282609313275673,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402645290","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025825258,0.0006354809,0.0017721253,0.000100669145,0.000032344262,0.000030951953,0.9696358,0.00032240865,0.0016449805],"genre_scores_gemma":[0.034399427,0.00037581133,0.0040950924,0.00006010481,0.000020394282,0.00018535386,0.96026355,0.000065502005,0.00053473393],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99895227,0.00014255906,0.00018966598,0.00038142287,0.00024685144,0.000087206456],"domain_scores_gemma":[0.99594647,0.00096116593,0.0011105563,0.00072522316,0.0010140676,0.00024250071],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010919414,0.000805388,0.00064571697,0.0043417388,0.0003737632,0.00087301777,0.00089631655,0.00085650675,0.0031660693],"category_scores_gemma":[0.0038623873,0.00035820898,0.0009920293,0.011597598,0.00034672118,0.0007713281,0.0011270832,0.00076601026,0.0023005162],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006463763,0.00038779195,0.44758692,0.0081647625,0.0016733706,0.0010977441,0.0008219046,0.034916326,0.014611156,0.0050220797,0.37261054,0.11246105],"study_design_scores_gemma":[0.00014207202,0.00010495784,0.5170049,0.0005331516,0.0002983997,0.00041494097,0.00037322025,0.0051430017,0.0038957412,0.0023102767,0.46966952,0.00010982738],"about_ca_topic_score_codex":0.017579222,"about_ca_topic_score_gemma":0.021551445,"teacher_disagreement_score":0.017579222,"about_ca_system_score_codex":0.0007942917,"about_ca_system_score_gemma":0.0017288489,"threshold_uncertainty_score":0.034953833},"labels":[],"label_agreement":null},{"id":"W4403036292","doi":"10.1038/s41597-024-03865-1","title":"The International Climate Psychology Collaboration: Climate change-related data collected from 63 countries","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate Change Communication and Perception","field":"Social Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kellogg's (Canada); Simon Fraser University; University of British Columbia","funders":"Fondo de Financiamiento de Centros de Investigación en Áreas Prioritarias; Japan Society for the Promotion of Science; Biotechnology and Biological Sciences Research Council; Fundação para a Ciência e a Tecnologia; Uniwersytet Śląski w Katowicach; Shell Brasil; NOMIS Stiftung; Université de Lausanne; Centro de Estudios de Conflicto y Cohesión Social; Agencia Nacional de Investigación y Desarrollo; Erasmus Universiteit Rotterdam; Université de Genève; Austrian Science Fund; Stanford Center on Philanthropy and Civil Society; Thammasat University; Universität Hamburg; Norges Forskningsråd; Agentúra na Podporu Výskumu a Vývoja; Morsani College of Medicine; Simon Fraser University; Medical Research Council; National Research Foundation of Korea; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Riksbankens Jubileumsfond; Fonds Wetenschappelijk Onderzoek; Universitat Ramon Llull; National Research University Higher School of Economics; National Geographic Society; American University of Sharjah; Universität Wien; James S. McDonnell Foundation; Aarhus Universitets Forskningsfond; Agence Nationale de la Recherche; Fonds De La Recherche Scientifique - FNRS; Canada Research Chairs; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Research Foundation; Fundação de Amparo à Pesquisa do Estado de São Paulo; Jacobs Foundation; Pomona College; Wellcome Trust; Social Sciences and Humanities Research Council of Canada; University of Colorado Boulder; University of St Andrews; Aarhus Universitet; National Science Foundation; UK Research and Innovation; Clemson University; Universitetet i Stavanger; Vetenskapsrådet; John Templeton Foundation","keywords":"Climate change; Psychological intervention; Biology and political orientation; Psychology; Social psychology; Applied psychology; Environmental resource management; Geography; Political science; Politics; Ecology; Environmental science","score_opus":0.5800491103888415,"score_gpt":0.5281463008214803,"score_spread":0.05190280956736126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403036292","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8019694,0.0005593735,0.005267409,0.0012902666,0.00013173654,0.0030097964,0.17134437,0.00023800731,0.016189672],"genre_scores_gemma":[0.75129795,0.00051708805,0.023012867,0.00072042405,0.000119177064,0.018359937,0.20093752,0.0002251132,0.0048099505],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99419147,0.002975124,0.0006117968,0.0006868145,0.0010136158,0.0005210785],"domain_scores_gemma":[0.9809234,0.004495907,0.0036319925,0.0035782144,0.0046531796,0.0027172058],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0071767187,0.0005333572,0.00061235524,0.003792949,0.0015798061,0.0012470102,0.0005417637,0.0011123552,0.0055886954],"category_scores_gemma":[0.015645286,0.00036160782,0.00037980775,0.009583334,0.00063122,0.0011203202,0.004354066,0.0009598041,0.0024620958],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009328001,0.0018497241,0.71056855,0.0016978985,0.00034372887,0.0006920344,0.03690534,0.0016053073,0.004910608,0.0025343178,0.119438305,0.11852135],"study_design_scores_gemma":[0.00017237855,0.00042344304,0.86502993,0.00027627836,0.00006387442,0.0002895734,0.02042522,0.00048125567,0.0015559864,0.0007119685,0.110477045,0.00009303072],"about_ca_topic_score_codex":0.010673764,"about_ca_topic_score_gemma":0.017967548,"teacher_disagreement_score":0.010673764,"about_ca_system_score_codex":0.00078642694,"about_ca_system_score_gemma":0.0038409017,"threshold_uncertainty_score":0.03795451},"labels":[],"label_agreement":null},{"id":"W4403133634","doi":"10.1038/s41597-024-03931-8","title":"Course-Skill Atlas: A national longitudinal dataset of skills taught in U.S. higher education curricula","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Labor market dynamics and wage inequality","field":"Economics, Econometrics and Finance","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Russell Sage Foundation","keywords":"Curriculum; Atlas (anatomy); Longitudinal data; Psychology; Mathematics education; Pedagogy; Biology; Sociology; Demography; Anatomy","score_opus":0.07755958872621702,"score_gpt":0.3232781377933296,"score_spread":0.24571854906711257,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403133634","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04025485,0.00017059645,0.0010633545,0.0003568168,0.00005205384,0.0000636285,0.9552936,0.0003473231,0.00239775],"genre_scores_gemma":[0.03373188,0.00012898554,0.0025573296,0.00014550153,0.000030384776,0.0002306299,0.96135545,0.00003724496,0.0017825105],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99949753,0.00008808176,0.000059301103,0.00015583166,0.00011938176,0.00007987282],"domain_scores_gemma":[0.9973346,0.00051073864,0.00060226716,0.00037927122,0.00066384824,0.0005092598],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007642381,0.00048058984,0.00036690125,0.0024736254,0.00047926232,0.0007607315,0.0009196557,0.0008192187,0.0058109006],"category_scores_gemma":[0.003841925,0.00024484133,0.00045255653,0.004351741,0.00018638761,0.000601528,0.0013134411,0.0011686216,0.005323485],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030465965,0.0006082609,0.25458238,0.00046213256,0.0001798749,0.00018818701,0.0007042494,0.0018977291,0.0011553215,0.0029336535,0.7037793,0.033204313],"study_design_scores_gemma":[0.0001692412,0.00013165544,0.62072086,0.00020975675,0.00005350248,0.00013846355,0.0010349496,0.004461611,0.0012025959,0.002277569,0.36953756,0.00006222143],"about_ca_topic_score_codex":0.04315401,"about_ca_topic_score_gemma":0.091344,"teacher_disagreement_score":0.04315401,"about_ca_system_score_codex":0.0010142226,"about_ca_system_score_gemma":0.0018490015,"threshold_uncertainty_score":0.085805655},"labels":[],"label_agreement":null},{"id":"W4403228728","doi":"10.1038/s41597-024-03752-9","title":"The Global Dam Watch database of river barrier and reservoir information for large-scale applications","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Flood Risk Assessment and Management","field":"Environmental Science","cited_by":85,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Wildlife Federation; McGill University","funders":"Agence Nationale de la Recherche; European Commission; Oak Ridge National Laboratory; National Socio-Environmental Synthesis Center; McGill University; World Wildlife Fund; European Regional Development Fund; World Bank Group; National Science Foundation","keywords":"Hydropower; Scale (ratio); Environmental science; Database; Computer science; Environmental resource management; Hydrology (agriculture); Geography; Cartography; Ecology; Geology","score_opus":0.016121120110965486,"score_gpt":0.29012361459557046,"score_spread":0.27400249448460495,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403228728","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010341874,0.0002723353,0.010448161,0.00023408048,0.000044624325,0.00023614892,0.967303,0.007785056,0.0033347656],"genre_scores_gemma":[0.02066314,0.00021850209,0.013157058,0.00006206877,0.00001539196,0.0003993858,0.9642343,0.0004826965,0.00076741434],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99891055,0.00014564196,0.00027640018,0.000256018,0.00033072758,0.000080679725],"domain_scores_gemma":[0.99569064,0.0009491209,0.0007450713,0.0014421381,0.0007789032,0.00039422896],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019653891,0.0009992803,0.00071825524,0.0071453257,0.00041314025,0.0019607618,0.001827407,0.0010236718,0.00848757],"category_scores_gemma":[0.008628264,0.00048702912,0.00070816284,0.009212029,0.00034925493,0.0022120252,0.0020806855,0.00086682756,0.0063290233],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030757036,0.00013300864,0.036922995,0.0024769045,0.00024372888,0.00060204964,0.00048907485,0.014556506,0.0062686633,0.016074125,0.8439964,0.077928886],"study_design_scores_gemma":[0.00020442196,0.000082289815,0.06802119,0.00057990075,0.00017893384,0.00046186364,0.0005675664,0.02831419,0.007864988,0.008976764,0.8845975,0.00015036289],"about_ca_topic_score_codex":0.010856592,"about_ca_topic_score_gemma":0.011338685,"teacher_disagreement_score":0.010856592,"about_ca_system_score_codex":0.00079233333,"about_ca_system_score_gemma":0.0033402375,"threshold_uncertainty_score":0.028393805},"labels":[],"label_agreement":null},{"id":"W4403282618","doi":"10.1038/s41597-024-03956-z","title":"Author Correction: Mechanical Unloading of Engineered Human Meniscus Models Under Simulated Microgravity: A Transcriptomic Study","year":2024,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Spaceflight effects on biology","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Meniscus; Transcriptome; Biology; Physics; Biochemistry; Gene expression; Optics","score_opus":0.1023104630241571,"score_gpt":0.37278572169367397,"score_spread":0.2704752586695169,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403282618","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00057440496,0.00096102443,0.0025561634,0.023007067,0.9670419,0.000036182882,0.0023544203,0.00058841595,0.0028804885],"genre_scores_gemma":[0.04608814,0.015069651,0.029600693,0.07402919,0.22377232,0.0005116093,0.01954598,0.0071807974,0.58420163],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9959927,0.00042312386,0.0006877252,0.00057209143,0.0020832021,0.00024121611],"domain_scores_gemma":[0.97057724,0.00536689,0.0010410176,0.0017595886,0.020315988,0.0009393685],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0036124624,0.00139523,0.001390407,0.0023849776,0.0024670335,0.002513995,0.0025339709,0.0034470882,0.058411654],"category_scores_gemma":[0.041174177,0.0007541393,0.0012517327,0.0019966853,0.0014723262,0.0018884104,0.0019310594,0.006548233,0.033195358],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007278691,0.000011744267,0.00012821026,0.00017064837,0.0000139451195,0.0004527894,0.00005153412,0.000060870403,0.0004853024,0.00071600906,0.9879302,0.009905962],"study_design_scores_gemma":[0.00003270466,0.000027179181,0.00078917545,0.0002553198,0.000049929036,0.0011509794,0.00010515934,0.00026021575,0.002111191,0.00092599815,0.9942492,0.000042914788],"about_ca_topic_score_codex":0.009157634,"about_ca_topic_score_gemma":0.012281104,"teacher_disagreement_score":0.058411654,"about_ca_system_score_codex":0.0022168676,"about_ca_system_score_gemma":0.0045069614,"threshold_uncertainty_score":0.19540644},"labels":[],"label_agreement":null},{"id":"W4403303688","doi":"10.1038/s41597-024-03965-y","title":"A temporal (phospho-)proteomic dataset of neurotrophic receptor tyrosine kinase signalling in neuroblastoma","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neuroblastoma Research and Treatments","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Children's Health Foundation; Irish Research Council; Science Foundation Ireland","keywords":"Neurotrophin; Neuroblastoma; Signalling; Receptor tyrosine kinase; Signalling pathways; Tyrosine kinase; Receptor; Biology; Tropomyosin receptor kinase B; Neuroscience; Computational biology; Signal transduction; Neurotrophic factors; Bioinformatics; Cell biology; Biochemistry; Genetics","score_opus":0.058551411565938495,"score_gpt":0.3358667250819108,"score_spread":0.2773153135159723,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403303688","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08592323,0.00396875,0.0025213752,0.00028217467,0.00012814465,0.000059557053,0.90325326,0.0012763159,0.0025871727],"genre_scores_gemma":[0.046738565,0.0022330142,0.007926917,0.00013638157,0.000030349573,0.00014377301,0.9415074,0.00014007361,0.0011434535],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99950707,0.000036417954,0.00006459705,0.0001866575,0.00012767414,0.000077489116],"domain_scores_gemma":[0.99943215,0.000104603314,0.000095267744,0.0000875424,0.00019075228,0.00008967825],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000502933,0.00095398835,0.0008392652,0.00226117,0.0008094473,0.0009314189,0.0005357392,0.00088927004,0.003460561],"category_scores_gemma":[0.0009847821,0.00023308805,0.0010670247,0.002857129,0.00032278715,0.0005493718,0.000950269,0.0008419152,0.0031294818],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0046428125,0.00041972584,0.081358366,0.014228076,0.0016608144,0.00378056,0.00087683595,0.009452712,0.52665436,0.0028985178,0.2730113,0.081015885],"study_design_scores_gemma":[0.00023746936,0.00036999292,0.43351677,0.0009452661,0.0007965196,0.00474766,0.0008351611,0.010169999,0.06741033,0.0046259924,0.47602507,0.00031978256],"about_ca_topic_score_codex":0.007913661,"about_ca_topic_score_gemma":0.018270902,"teacher_disagreement_score":0.007913661,"about_ca_system_score_codex":0.0008706266,"about_ca_system_score_gemma":0.0013866435,"threshold_uncertainty_score":0.01573515},"labels":[],"label_agreement":null},{"id":"W4403377357","doi":"10.1038/s41597-024-03962-1","title":"A practitioner-oriented regional hydrology data product for use in site-specific hydraulic applications","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Global Institute for Water Security; Environment and Climate Change Canada; University of Saskatchewan","funders":"Canada First Research Excellence Fund","keywords":"Hydrology (agriculture); Environmental science; Product (mathematics); Geology; Mathematics; Geotechnical engineering","score_opus":0.08879299991659176,"score_gpt":0.30345120374451673,"score_spread":0.21465820382792497,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403377357","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0072237793,0.000049324648,0.02292254,0.00021036544,0.000034403693,0.00037252993,0.95189357,0.007031213,0.010262269],"genre_scores_gemma":[0.02565202,0.00012507982,0.049063414,0.00006499033,0.000015359812,0.00049593317,0.9204471,0.001046374,0.0030897784],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99909365,0.00008486209,0.00011519603,0.00020238599,0.00041220782,0.00009176157],"domain_scores_gemma":[0.9949917,0.0005048343,0.00050387136,0.0013021511,0.0023730325,0.00032434688],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018857168,0.00067457225,0.0005373903,0.003434637,0.0005050576,0.0015491821,0.0011734718,0.0005782281,0.013300808],"category_scores_gemma":[0.0070169433,0.0004860881,0.0005045131,0.0070883394,0.0002812307,0.0011731667,0.0011716974,0.00088621094,0.0081596635],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038701977,0.00028378068,0.050382193,0.00095626124,0.00015969713,0.00028518494,0.00060178613,0.02023079,0.010889381,0.011363071,0.7223835,0.18207735],"study_design_scores_gemma":[0.0001898319,0.000061140985,0.117368974,0.00022194319,0.000051309904,0.00015067248,0.00047162652,0.02486557,0.0065561435,0.0041794777,0.8457461,0.00013719409],"about_ca_topic_score_codex":0.20854834,"about_ca_topic_score_gemma":0.24680065,"teacher_disagreement_score":0.20854834,"about_ca_system_score_codex":0.0019336728,"about_ca_system_score_gemma":0.0060440404,"threshold_uncertainty_score":0.41466892},"labels":[],"label_agreement":null},{"id":"W4403442553","doi":"10.1038/s41597-024-03973-y","title":"Age-specific ASPECTS atlas of Chinese subjects across different age groups for assessing acute ischemic stroke","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Acute Ischemic Stroke Management","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Higher Education Discipline Innovation Project; National Natural Science Foundation of China","keywords":"Ischemic stroke; Atlas (anatomy); Stroke (engine); Medicine; Gerontology; Internal medicine; Ischemia; Anatomy; Engineering","score_opus":0.04633025133040226,"score_gpt":0.34406892098895075,"score_spread":0.2977386696585485,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403442553","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7465395,0.003982561,0.14141859,0.001184223,0.00031103642,0.0017479981,0.066027045,0.005811643,0.032977514],"genre_scores_gemma":[0.8533864,0.0023154642,0.09271237,0.00034867067,0.00016858563,0.0013940657,0.039651938,0.0007392408,0.00928331],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9996246,0.00006310784,0.000051336658,0.00012749192,0.00008051634,0.00005290951],"domain_scores_gemma":[0.99929297,0.00010045327,0.00011624235,0.00018898971,0.00024194962,0.000059298694],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009483172,0.0006570986,0.00037347255,0.0030157764,0.00052182504,0.0010113785,0.00049769715,0.00047285057,0.0057350737],"category_scores_gemma":[0.0024077925,0.00030586755,0.0005644484,0.0020875758,0.0004717103,0.0006407953,0.0009846255,0.00031942764,0.0015608914],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018075926,0.00017520515,0.26209491,0.0017216098,0.00057449745,0.0024543528,0.0039881566,0.007907376,0.053775717,0.00763809,0.08450752,0.573355],"study_design_scores_gemma":[0.00013066584,0.00036800338,0.82749355,0.00036225692,0.0007413392,0.0061934856,0.0016995172,0.02402655,0.025303796,0.008419156,0.10502169,0.0002399085],"about_ca_topic_score_codex":0.01583005,"about_ca_topic_score_gemma":0.030937556,"teacher_disagreement_score":0.01583005,"about_ca_system_score_codex":0.00065511855,"about_ca_system_score_gemma":0.0018453168,"threshold_uncertainty_score":0.031475782},"labels":[],"label_agreement":null},{"id":"W4403545417","doi":"10.1038/s41597-024-03940-7","title":"Lake Superior evaporation: A long-term eddy covariance dataset at Stannard Rock Lighthouse (2008–2022)","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Water Relations and Carbon Dynamics","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"Environment and Climate Change Canada; Michigan Technological University; National Oceanic and Atmospheric Administration; NOAA Great Lakes Environmental Research Laboratory; U.S. Department of Commerce; National Science Foundation","keywords":"Eddy covariance; Lake ecosystem; Hydrology (agriculture); Environmental science; Term (time); Snow; Range (aeronautics); Evaporation; Ecosystem; Physical geography; Geology; Geography; Meteorology; Ecology","score_opus":0.025844394886345833,"score_gpt":0.26233651235338723,"score_spread":0.2364921174670414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403545417","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45762038,0.00022115823,0.00066713407,0.000509368,0.0000321995,0.000048560523,0.53459907,0.0009186318,0.00538354],"genre_scores_gemma":[0.27415302,0.00014041543,0.0037382157,0.00013653301,0.000052156007,0.00014819612,0.7179268,0.00013404197,0.003570606],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998573,0.000013091578,0.00001043047,0.00003859891,0.000054569053,0.000025944697],"domain_scores_gemma":[0.99970347,0.0000147998635,0.000047045443,0.000044364693,0.00014070138,0.0000495091],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00029430544,0.0003188915,0.00030123175,0.0005695427,0.00041246632,0.0005083786,0.00050844986,0.0004812482,0.0032995122],"category_scores_gemma":[0.0005195955,0.00017359546,0.00027628904,0.0011660146,0.00016473635,0.00038199918,0.0006412972,0.00032956907,0.0015631298],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048747525,0.00028403636,0.52932537,0.00035848626,0.00046748176,0.0005226261,0.000732523,0.009309046,0.010554822,0.0010512909,0.41216156,0.034745272],"study_design_scores_gemma":[0.00023769017,0.000044440152,0.87753105,0.00006929069,0.00005418136,0.00006444838,0.00032086045,0.011285336,0.0015290135,0.00027002103,0.108543515,0.000050254526],"about_ca_topic_score_codex":0.17972946,"about_ca_topic_score_gemma":0.39299473,"teacher_disagreement_score":0.82027054,"about_ca_system_score_codex":0.00072747265,"about_ca_system_score_gemma":0.0011864993,"threshold_uncertainty_score":0.35736662},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4403547097","doi":"10.1038/s41597-024-03972-z","title":"A dataset of Solicited Cough Sound for Tuberculosis Triage Testing","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Respiratory and Cough-Related Research","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Centre Hospitalier de l’Université de Montréal","funders":"National Institute of Allergy and Infectious Diseases; National Institutes of Health; Fonds de Recherche du Québec - Santé; Bill and Melinda Gates Foundation; U.S. Department of Health and Human Services","keywords":"Medicine; Tuberculosis; Triage; Intensive care medicine; Disease; Pandemic; Limiting; Infectious disease (medical specialty); Pathology; Medical emergency; Coronavirus disease 2019 (COVID-19)","score_opus":0.31017244749585604,"score_gpt":0.44934048993525966,"score_spread":0.13916804243940362,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403547097","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.060287517,0.0019749652,0.0026373747,0.00093674904,0.00045392415,0.00033571443,0.9276909,0.0015983661,0.0040844795],"genre_scores_gemma":[0.04427978,0.0004275158,0.0043577347,0.00039159888,0.000107288666,0.00040722574,0.9484099,0.000056138604,0.0015627703],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988312,0.00027038716,0.00020685092,0.00026914227,0.00029456496,0.000127877],"domain_scores_gemma":[0.99800426,0.00072943815,0.00019402614,0.0003033848,0.0005146506,0.0002542167],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00082731055,0.0011247665,0.0008447338,0.0022226232,0.00060445326,0.0011017849,0.0015715625,0.0025092294,0.004782629],"category_scores_gemma":[0.004206854,0.00025565567,0.00095868995,0.002056702,0.00041180567,0.0005688987,0.0012935965,0.001031939,0.007867259],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015497835,0.0011497425,0.12758732,0.004093173,0.0005166302,0.0025002067,0.0007113004,0.0070619797,0.0074876347,0.0011558473,0.7560549,0.09013148],"study_design_scores_gemma":[0.0005726966,0.000856059,0.33999187,0.001433599,0.00035169887,0.0038865455,0.0024363769,0.033883095,0.0061150678,0.0029700838,0.60717034,0.00033251994],"about_ca_topic_score_codex":0.012538093,"about_ca_topic_score_gemma":0.030199582,"teacher_disagreement_score":0.012538093,"about_ca_system_score_codex":0.0007440424,"about_ca_system_score_gemma":0.0009808948,"threshold_uncertainty_score":0.024930239},"labels":[],"label_agreement":null},{"id":"W4403872579","doi":"10.1038/s41597-024-04026-0","title":"A global dataset of salmonid biomass in streams","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fish Ecology and Management Studies","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Biomass (ecology); STREAMS; Range (aeronautics); Sampling (signal processing); Scope (computer science); Environmental science; Taxon; Ecology; Fishery; Biology; Computer science","score_opus":0.027897313757870572,"score_gpt":0.29157818573838074,"score_spread":0.26368087198051016,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403872579","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019791842,0.00021603127,0.00067064376,0.00007490953,0.000028175802,0.000031565905,0.9773932,0.00037545207,0.0014182095],"genre_scores_gemma":[0.013366872,0.00009381442,0.0016415572,0.00003031029,0.000015457437,0.00011554876,0.9841613,0.000044465163,0.0005306349],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99911505,0.00013512892,0.00016399763,0.0002962129,0.00019548161,0.00009415175],"domain_scores_gemma":[0.9983392,0.0003156788,0.00032361425,0.00024015622,0.0006086154,0.0001727685],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.00078528683,0.0006858439,0.00074354734,0.002730276,0.0003472541,0.0008992313,0.0008393928,0.0007843333,0.007514737],"category_scores_gemma":[0.0029797137,0.0003177068,0.0007753537,0.0059482264,0.00020981107,0.00063820864,0.0009646882,0.0007900628,0.006197896],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006017568,0.00030420002,0.26986566,0.0021484771,0.0006121875,0.00040987323,0.0004816865,0.015572559,0.0031910322,0.0023506912,0.6329293,0.07153256],"study_design_scores_gemma":[0.00031134504,0.00011682177,0.46730575,0.00047503482,0.00014622859,0.0002947334,0.00045934808,0.008535599,0.001540001,0.0021083625,0.51859534,0.000111445006],"about_ca_topic_score_codex":0.012494658,"about_ca_topic_score_gemma":0.019392058,"teacher_disagreement_score":0.9991606,"about_ca_system_score_codex":0.000550032,"about_ca_system_score_gemma":0.0011502391,"threshold_uncertainty_score":0.025139332},"labels":[],"label_agreement":null},{"id":"W4403891519","doi":"10.1038/s41597-024-03897-7","title":"FARFUM-RoP, A dataset for computer-aided detection of Retinopathy of Prematurity","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Retinopathy of Prematurity Studies","field":"Medicine","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Retinopathy of prematurity; Computer science; Medicine; Gestational age; Biology; Pregnancy","score_opus":0.07432057729253969,"score_gpt":0.3530332026409905,"score_spread":0.27871262534845076,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403891519","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030373368,0.0051766513,0.0022519021,0.0006822498,0.00021877898,0.0003723874,0.9556453,0.0011941481,0.0040852176],"genre_scores_gemma":[0.026114453,0.0010215306,0.0054174983,0.00015621935,0.000064481676,0.0004548988,0.9657976,0.000058356873,0.00091484044],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992291,0.00014391393,0.00013959596,0.00022343949,0.00017917877,0.00008480992],"domain_scores_gemma":[0.9986473,0.0004283741,0.00019962763,0.00025034926,0.00032220187,0.00015223629],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008717024,0.0009915574,0.0007443378,0.0038205343,0.00065048784,0.0008802602,0.0015918647,0.0018191619,0.004994226],"category_scores_gemma":[0.0041600643,0.0002472977,0.0010268522,0.0023999298,0.00030396433,0.00040444004,0.0010115115,0.00082321663,0.0029964165],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013673677,0.00039727203,0.054827686,0.005497102,0.00050968555,0.002292998,0.00037353084,0.0022290046,0.0063538253,0.0016216259,0.83757126,0.08695859],"study_design_scores_gemma":[0.0005782559,0.00038209124,0.2217785,0.001841591,0.00045740601,0.0060250736,0.000794978,0.0065046377,0.006525365,0.002349777,0.75260234,0.00016001644],"about_ca_topic_score_codex":0.014406319,"about_ca_topic_score_gemma":0.031151962,"teacher_disagreement_score":0.014406319,"about_ca_system_score_codex":0.0009727213,"about_ca_system_score_gemma":0.0017182088,"threshold_uncertainty_score":0.02864492},"labels":[],"label_agreement":null},{"id":"W4404199554","doi":"10.1038/s41597-024-04010-8","title":"Neural Dynamics of Creative Movements During the Rehearsal and Performance of “LiveWire\"","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neural dynamics and brain function","field":"Neuroscience","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"McGill University; Rice University; National Science Foundation","keywords":"Electrooculography; Forehead; Electroencephalography; Computer science; Dynamics (music); Brain activity and meditation; Inertial measurement unit; Motion (physics); Artificial intelligence; Brain–computer interface; Computer vision; Psychology; Eye movement; Neuroscience; Medicine","score_opus":0.03630925699347074,"score_gpt":0.27132811150185093,"score_spread":0.2350188545083802,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404199554","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9905193,0.00020038443,0.006100468,0.00003876447,0.000018184126,0.00010245407,0.0003633965,0.00009417409,0.0025627958],"genre_scores_gemma":[0.9909451,0.00014145009,0.0061577572,0.000030065841,0.000014735585,0.000112962865,0.00032823585,0.000034430705,0.0022351167],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99987054,0.000023812867,0.000006968464,0.000036652036,0.00004257127,0.000019477393],"domain_scores_gemma":[0.99979895,0.000074519994,0.000043854317,0.000021065363,0.00003530292,0.000026321086],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00014523928,0.00021800955,0.00016509536,0.00029726472,0.00015958231,0.0002626707,0.0002205399,0.00020108755,0.0018538021],"category_scores_gemma":[0.0013681627,0.00009825528,0.00012244833,0.00023693078,0.00030374737,0.00016074862,0.00034837684,0.00023948631,0.00027976837],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012505475,0.00017359416,0.018336993,0.00039406124,0.00009685329,0.0006974462,0.0050396947,0.0007623062,0.8184603,0.0005898206,0.0008542498,0.15334414],"study_design_scores_gemma":[0.000048840528,0.001000181,0.9287187,0.00006257221,0.00005594299,0.0013181744,0.0017451893,0.003756129,0.058992606,0.00054786995,0.003702435,0.000051472063],"about_ca_topic_score_codex":0.0008945705,"about_ca_topic_score_gemma":0.0020656097,"teacher_disagreement_score":0.0018538021,"about_ca_system_score_codex":0.000102378304,"about_ca_system_score_gemma":0.00009293939,"threshold_uncertainty_score":0.0062015653},"labels":[],"label_agreement":null},{"id":"W4404347553","doi":"10.1038/s41597-024-04011-7","title":"A Biomechanical Dataset of 1,798 Healthy and Injured Subjects During Treadmill Walking and Running","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Lower Extremity Biomechanics and Pathologies","field":"Engineering","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Running Injury Clinic; University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Treadmill; Physical medicine and rehabilitation; Medicine; Physical therapy; Computer science","score_opus":0.037846277827889616,"score_gpt":0.2801639564275309,"score_spread":0.24231767859964126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404347553","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.062817514,0.00083907956,0.0060676215,0.00044114923,0.00023118127,0.00035861586,0.923557,0.0015348756,0.004152872],"genre_scores_gemma":[0.041834056,0.00040247056,0.0055082464,0.00019115042,0.00010934666,0.0010877742,0.9469992,0.00012592087,0.0037418567],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996201,0.00006357947,0.000066591834,0.00012087053,0.00008778728,0.000041011288],"domain_scores_gemma":[0.99891436,0.00026000757,0.000102899874,0.00021896293,0.00039893552,0.000104848164],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00055927684,0.00087902095,0.00087509997,0.0013578442,0.0004688098,0.0005011514,0.0009465944,0.001203138,0.015462703],"category_scores_gemma":[0.002778777,0.0002580679,0.00068495487,0.0013353353,0.00027440363,0.00039786586,0.0010034154,0.00053431734,0.013049477],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025233668,0.0012028044,0.07426539,0.003524322,0.0003765606,0.0011115406,0.0005411877,0.0029423514,0.0072810072,0.0010811182,0.7523611,0.15278928],"study_design_scores_gemma":[0.00073175033,0.0013158374,0.5999452,0.0013843233,0.00039868208,0.0027611123,0.0011232552,0.006170818,0.0042375894,0.0047986596,0.37686586,0.00026683594],"about_ca_topic_score_codex":0.0054295897,"about_ca_topic_score_gemma":0.0153231295,"teacher_disagreement_score":0.015462703,"about_ca_system_score_codex":0.0002755614,"about_ca_system_score_gemma":0.0007077067,"threshold_uncertainty_score":0.05172789},"labels":[],"label_agreement":null},{"id":"W4404484879","doi":"10.1038/s41597-024-03994-7","title":"Historical datasets (1950–2022) of monthly water balance components for the Laurentian Great Lakes","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Environment and Climate Change Canada","funders":"National Oceanic and Atmospheric Administration; Environment and Climate Change Canada","keywords":"Water balance; Balance (ability); Bayesian probability; Environmental science; Psychological resilience; Reliability (semiconductor); Water resources; Ecosystem; Environmental resource management; Computer science; Ecology; Geology","score_opus":0.0376020613180141,"score_gpt":0.25516026664432484,"score_spread":0.21755820532631073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404484879","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.27914694,0.0007910089,0.0045865863,0.0006691945,0.00006991895,0.00008314396,0.7058276,0.0007519533,0.00807359],"genre_scores_gemma":[0.3020704,0.00045103408,0.008492386,0.00019078334,0.000047246962,0.0002643858,0.6852153,0.00008458566,0.0031838766],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9997023,0.00003688615,0.00003705737,0.00008208257,0.000108885266,0.00003279174],"domain_scores_gemma":[0.9986957,0.00012222114,0.00027933667,0.00018172122,0.00062196056,0.00009912449],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006753223,0.0003691069,0.00023536281,0.002426127,0.00032673657,0.0004615018,0.0005307781,0.00039897492,0.0025910458],"category_scores_gemma":[0.002113714,0.00021137405,0.00034140114,0.0043215416,0.00023256817,0.00059736066,0.00064316037,0.00040438626,0.00096972566],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020599995,0.00018831393,0.64078546,0.0005733523,0.0005793119,0.0004625748,0.00092668564,0.039617747,0.0027354637,0.003663991,0.22894,0.08132117],"study_design_scores_gemma":[0.000039551927,0.000028496715,0.81159014,0.00008143423,0.000060638213,0.000096528165,0.00030474734,0.01508966,0.0014103265,0.0010402836,0.17020647,0.000051629933],"about_ca_topic_score_codex":0.26767752,"about_ca_topic_score_gemma":0.39236388,"teacher_disagreement_score":0.73232245,"about_ca_system_score_codex":0.0017045293,"about_ca_system_score_gemma":0.002454237,"threshold_uncertainty_score":0.53223896},"labels":[],"label_agreement":null},{"id":"W4404719260","doi":"10.1038/s41597-024-04101-6","title":"A high-resolution satellite-based solar-induced chlorophyll fluorescence dataset for China from 2000 to 2022","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Remote sensing; Environmental science; Satellite; Primary production; Image resolution; Atmospheric sciences; Meteorology; Ecosystem; Computer science; Geography; Geology; Physics; Ecology","score_opus":0.024622025579839064,"score_gpt":0.2577707167241957,"score_spread":0.23314869114435666,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404719260","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.522972,0.0003421124,0.0026533224,0.00018533556,0.000037357226,0.00013682114,0.47027165,0.0004948455,0.0029066214],"genre_scores_gemma":[0.33159178,0.00012079397,0.005215653,0.000057196707,0.0000131905435,0.00015156827,0.6616466,0.000029588798,0.0011736328],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99981374,0.0000145416725,0.000018868515,0.00006156231,0.00005649415,0.00003480995],"domain_scores_gemma":[0.99955875,0.000030671214,0.00005902904,0.000084940475,0.00021328185,0.000053331958],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00047255817,0.00065789564,0.00035604046,0.0019107815,0.00032986925,0.00028900156,0.0005938881,0.00043219107,0.001428688],"category_scores_gemma":[0.0006033496,0.00020376862,0.0004319948,0.0029485729,0.00017129218,0.0003539464,0.00041045927,0.00025325583,0.0004973571],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00055599917,0.000619406,0.6224347,0.0007775989,0.0007916066,0.0011236046,0.00036149172,0.11059422,0.024158403,0.001811718,0.12559593,0.1111753],"study_design_scores_gemma":[0.000088074594,0.000066735905,0.8964093,0.000042401476,0.00008659161,0.00011849121,0.00014296285,0.053146206,0.0047194576,0.00038978696,0.044725537,0.00006444141],"about_ca_topic_score_codex":0.12880974,"about_ca_topic_score_gemma":0.15784983,"teacher_disagreement_score":0.12880974,"about_ca_system_score_codex":0.0013309839,"about_ca_system_score_gemma":0.0020545577,"threshold_uncertainty_score":0.25611997},"labels":[],"label_agreement":null},{"id":"W4405075600","doi":"10.1038/s41597-024-04157-4","title":"Generative models of MRI-derived neuroimaging features and associated dataset of 18,000 samples","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute on Aging; Canadian Institutes of Health Research; Genentech; IXICO; H. Lundbeck A/S; Servier; Eisai; Northern California Institute for Research and Education; Pfizer; Novartis Pharmaceuticals Corporation; Alzheimer's Disease Neuroimaging Initiative; Meso Scale Diagnostics; University of Southern California; BioClinica; Biogen; Bristol-Myers Squibb; Eli Lilly and Company; National Institutes of Health; U.S. Department of Health and Human Services","keywords":"Neuroimaging; Computer science; Machine learning; Context (archaeology); Artificial intelligence; Generative model; Normative; Generative grammar; Data science; Psychology; Neuroscience; Geography","score_opus":0.10633155753044737,"score_gpt":0.352174247784702,"score_spread":0.24584269025425465,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405075600","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47132573,0.002258121,0.33091635,0.00430224,0.00054252817,0.0013840512,0.17619048,0.0069124964,0.0061680158],"genre_scores_gemma":[0.6566177,0.0008691199,0.11339781,0.000973973,0.00018339617,0.0020086681,0.22081985,0.00043375566,0.0046956427],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990809,0.00042755218,0.000049406128,0.00022498103,0.00014965488,0.00006746329],"domain_scores_gemma":[0.9937983,0.0041425377,0.00031232872,0.0010589096,0.0005327201,0.000155184],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003394147,0.00084543956,0.0006374813,0.0011152262,0.00044985628,0.0007895503,0.0020611903,0.0016439761,0.004733178],"category_scores_gemma":[0.010998301,0.0005874985,0.0015108616,0.001177095,0.0011031674,0.00057286373,0.0010961791,0.0021818038,0.0016616099],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016785164,0.0010036696,0.091993645,0.0006275185,0.00063997606,0.001641277,0.0005630194,0.64767563,0.0074860966,0.0338258,0.113126814,0.09973807],"study_design_scores_gemma":[0.0005605094,0.00032642632,0.030979536,0.00015437209,0.00017322555,0.0015133704,0.00018765853,0.8837723,0.0053772586,0.04894371,0.027880251,0.00013137897],"about_ca_topic_score_codex":0.007757258,"about_ca_topic_score_gemma":0.012745554,"teacher_disagreement_score":0.007757258,"about_ca_system_score_codex":0.001190275,"about_ca_system_score_gemma":0.0010941426,"threshold_uncertainty_score":0.017950177},"labels":[],"label_agreement":null},{"id":"W4405516627","doi":"10.1038/s41597-024-04239-3","title":"Chromosome-level genome assembly of Phortica okadai, a vector of Thelazia callipaeda","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Parasitic Diseases Research and Treatment","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Petroleum Technology Research Centre; Guizhou Science and Technology Department","keywords":"Biology; Genome; Chromosome; Genetics; Zoology; Computational biology; Gene","score_opus":0.11901215003872492,"score_gpt":0.3880388147661537,"score_spread":0.26902666472742875,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405516627","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7735832,0.0042240615,0.11929833,0.00052964466,0.0003558536,0.0009123664,0.084462315,0.0025561848,0.014078142],"genre_scores_gemma":[0.5020697,0.0037335884,0.19216956,0.00022087398,0.00004827509,0.0009476173,0.28581518,0.0012088517,0.013786233],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99981636,0.0000144789965,0.00001745395,0.00007612223,0.00004569253,0.000029991796],"domain_scores_gemma":[0.99983406,0.00003114948,0.00002538795,0.000019505642,0.00006161108,0.000028250384],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00025753235,0.0006443218,0.00057346467,0.0012605968,0.0005966005,0.00060079515,0.00037770922,0.00038745007,0.0021560837],"category_scores_gemma":[0.00046885406,0.00034507585,0.0006811932,0.0010791282,0.00017915212,0.0003380906,0.00051233947,0.0009775828,0.0015293864],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038773805,0.000041098476,0.0027980125,0.0004286821,0.0000451471,0.00028209083,0.00031269426,0.0010829229,0.97455466,0.000591557,0.0012107268,0.01826463],"study_design_scores_gemma":[0.00028362114,0.0010755901,0.21989033,0.0003963104,0.00094184506,0.0016144465,0.0011607712,0.026715564,0.55745625,0.0013249259,0.18885677,0.00028363033],"about_ca_topic_score_codex":0.008263481,"about_ca_topic_score_gemma":0.011646341,"teacher_disagreement_score":0.008263481,"about_ca_system_score_codex":0.000503362,"about_ca_system_score_gemma":0.00079296704,"threshold_uncertainty_score":0.016430736},"labels":[],"label_agreement":null},{"id":"W4405523294","doi":"10.1038/s41597-024-04261-5","title":"Primate Phenotypes: A Multi-Institution Collection of 3D Morphological Data Housed in MorphoSource","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Primate Behavior and Ecology","field":"Psychology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lakehead University; University of Toronto","funders":"Ministerio de Ciencia e Innovación; American Museum of Natural History; Agencia Estatal de Investigación; National Science Foundation","keywords":"Digitization; Phenomics; Metadata; Natural history; National Museum of Natural History; Computer science; Data science; World Wide Web; Biology; Ecology; Genomics","score_opus":0.16115306330709683,"score_gpt":0.3947069348684969,"score_spread":0.23355387156140006,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405523294","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.059159353,0.000915897,0.018422835,0.00051591115,0.00014668111,0.0005560881,0.8873147,0.010444567,0.022524018],"genre_scores_gemma":[0.052178953,0.00047345652,0.03787297,0.00024535772,0.00005253849,0.0016652276,0.90051335,0.003131588,0.0038665524],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984225,0.00019347687,0.00013169803,0.00037118094,0.00075394165,0.00012714359],"domain_scores_gemma":[0.99735236,0.000291762,0.00034752264,0.0008822708,0.00069765217,0.00042837716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015215918,0.0008225874,0.0008687,0.004691881,0.0012591615,0.0015732305,0.0019582598,0.0006716541,0.01958102],"category_scores_gemma":[0.0051806956,0.0005109541,0.00063230016,0.0046324744,0.0009902027,0.0012358027,0.0045587295,0.0010516604,0.013559898],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006986173,0.00022989044,0.0700899,0.0013765141,0.00034418725,0.0008874637,0.002590517,0.0015218048,0.022720141,0.0068651554,0.6721348,0.22054097],"study_design_scores_gemma":[0.000072386836,0.00007517062,0.22608224,0.00034401767,0.0000770728,0.0012118566,0.00089063734,0.0016584744,0.0050817556,0.0041052783,0.76026547,0.00013559763],"about_ca_topic_score_codex":0.015139648,"about_ca_topic_score_gemma":0.04187343,"teacher_disagreement_score":0.01958102,"about_ca_system_score_codex":0.00086834247,"about_ca_system_score_gemma":0.0024266497,"threshold_uncertainty_score":0.06550503},"labels":[],"label_agreement":null},{"id":"W4405858697","doi":"10.1038/s41597-024-04324-7","title":"Transcriptome and translatome profiling of Col-0 and grp7grp8 under ABA treatment in Arabidopsis","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"RNA Research and Splicing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Key Research and Development Program of China; Institute of Genetics; Universität Bielefeld; Chinese Academy of Sciences","keywords":"Biology; Ribosome profiling; Translational regulation; Arabidopsis; Transcriptome; Translation (biology); Ribosome; Abscisic acid; Computational biology; RNA; RNA-binding protein; Messenger RNA; Arabidopsis thaliana; Immunoprecipitation; Eukaryotic translation; RNA-Seq; Cell biology; Gene expression; Gene; Genetics","score_opus":0.04613357419638759,"score_gpt":0.32769013025125393,"score_spread":0.28155655605486635,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405858697","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9760685,0.0015953364,0.00568676,0.00012494513,0.00006284458,0.000043128854,0.0135887945,0.00032908487,0.0025005762],"genre_scores_gemma":[0.9571698,0.0011002265,0.009213337,0.00038053447,0.000028160064,0.00021719349,0.024376806,0.0003964867,0.0071174437],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998871,0.0000068252293,0.0000050313706,0.000054954067,0.00002559936,0.000020489762],"domain_scores_gemma":[0.99992156,0.000014697029,0.000015537065,0.000008791595,0.000017506365,0.000021950194],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00010052642,0.00029619338,0.00044699575,0.0003341763,0.0002719951,0.0002522264,0.00019080336,0.00024122474,0.0008449306],"category_scores_gemma":[0.00008140537,0.00018624299,0.0003816381,0.00035390642,0.00019867782,0.00020774419,0.00022231162,0.0005829911,0.00040414327],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006508061,0.0000056508648,0.00024845943,0.000031508225,0.0000046481327,0.00003464666,0.000030281204,0.0000568913,0.9986431,0.00002689852,0.000068319445,0.000784478],"study_design_scores_gemma":[0.000040051033,0.00038856777,0.27342227,0.00004155547,0.0001710919,0.000622601,0.0005269632,0.006171927,0.7063367,0.00035417627,0.0118401535,0.00008397574],"about_ca_topic_score_codex":0.0020628348,"about_ca_topic_score_gemma":0.0035061266,"teacher_disagreement_score":0.0020628348,"about_ca_system_score_codex":0.0003624603,"about_ca_system_score_gemma":0.00022248125,"threshold_uncertainty_score":0.0041016936},"labels":[],"label_agreement":null},{"id":"W4406087184","doi":"10.1038/s41597-025-04372-7","title":"A Collaborative and Scalable Geospatial Data Set for Arctic Retrogressive Thaw Slumps with Data Standards","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate change and permafrost","field":"Earth and Planetary Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Permafrost; Geospatial analysis; Arctic; Computer science; Database; Environmental science; Physical geography; Data mining; Data science; Remote sensing; Geography; Geology; Oceanography","score_opus":0.06993715916196934,"score_gpt":0.32993986702557104,"score_spread":0.2600027078636017,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406087184","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023531158,0.00054754084,0.014460941,0.00087382464,0.00031824433,0.00045050526,0.9471073,0.0075882897,0.005122191],"genre_scores_gemma":[0.013661601,0.00014201754,0.02027613,0.000113873815,0.000028440341,0.0004476712,0.9643283,0.0002931448,0.00070886296],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99742764,0.00030539298,0.000394729,0.00086329906,0.00076442276,0.0002444785],"domain_scores_gemma":[0.9958139,0.0006313725,0.0003658949,0.0015439071,0.0012512245,0.0003937667],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002360187,0.001041377,0.00078638503,0.0045704134,0.0015987394,0.0016371803,0.0029213086,0.0015625473,0.0047690487],"category_scores_gemma":[0.008575791,0.00041780857,0.0015115854,0.006077012,0.0008751624,0.0022661036,0.0046189083,0.0017845826,0.0051659225],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004967303,0.00041821785,0.03702458,0.0021153986,0.00036061092,0.000879601,0.0014065242,0.017187664,0.0122248335,0.010920894,0.82165605,0.09530889],"study_design_scores_gemma":[0.00023784718,0.00008653249,0.060368504,0.0005090382,0.000111395646,0.00040196875,0.0016785908,0.023577783,0.010607376,0.00776456,0.89444166,0.00021463208],"about_ca_topic_score_codex":0.04271072,"about_ca_topic_score_gemma":0.08108995,"teacher_disagreement_score":0.04271072,"about_ca_system_score_codex":0.0014274932,"about_ca_system_score_gemma":0.0037313534,"threshold_uncertainty_score":0.08492422},"labels":[],"label_agreement":null},{"id":"W4406087194","doi":"10.1038/s41597-025-04378-1","title":"Reconstruction of long-term hourly sea level data for Prince Edward Island, Canada","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geophysics and Gravity Measurements","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Prince Edward Island","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Natural Resources Canada; Canada Foundation for Innovation; Atlantic Canada Opportunities Agency; Government of Canada","keywords":"Shore; Context (archaeology); Climate change; Storm surge; Term (time); Sea level; Storm; Geography; Oceanography; Historical record; Hazard; Physical geography; Environmental science; Climatology; Meteorology; Geology; Archaeology; Ecology; History","score_opus":0.08145897699803296,"score_gpt":0.2695946761605909,"score_spread":0.18813569916255796,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406087194","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6614249,0.00082985306,0.0046927407,0.0008140545,0.000089775414,0.00008945944,0.3226236,0.0006306543,0.008805009],"genre_scores_gemma":[0.67377025,0.000994818,0.010690683,0.000119873795,0.00002008235,0.00009082457,0.306663,0.00012346759,0.007527001],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997867,0.0000076238643,0.0000132498635,0.000039460858,0.0000864297,0.000066644614],"domain_scores_gemma":[0.99901414,0.00003805853,0.000056685258,0.000073862735,0.00072035094,0.000096866446],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022281286,0.0004036518,0.0002649985,0.0017055334,0.00080889364,0.00090569846,0.001084089,0.00040602716,0.0023658809],"category_scores_gemma":[0.0011337742,0.00029261582,0.0004063629,0.0044435062,0.00034362596,0.00029811842,0.00055226596,0.0006077735,0.0008298134],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033852205,0.00022478738,0.6316406,0.00040737598,0.00046692346,0.0010560293,0.0009293368,0.1502955,0.0044535417,0.0035905133,0.09844429,0.10815252],"study_design_scores_gemma":[0.00006547145,0.000025669377,0.79937506,0.00017581762,0.00007123748,0.00012849752,0.0019861215,0.1189074,0.0020976744,0.00045084642,0.07657689,0.00013929105],"about_ca_topic_score_codex":0.99076813,"about_ca_topic_score_gemma":0.99357295,"teacher_disagreement_score":0.009231865,"about_ca_system_score_codex":0.008620347,"about_ca_system_score_gemma":0.020821346,"threshold_uncertainty_score":0.0625453},"labels":[],"label_agreement":null},{"id":"W4406274339","doi":"10.1038/s41597-024-04175-2","title":"The North American Repository for Archaeological Isotopes","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Archaeology and ancient environmental studies","field":"Earth and Planetary Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Max-Planck-Institut für Menschheitsgeschichte; Max-Planck-Gesellschaft","keywords":"Archaeology; Isotope; Geography; Geology; Physics","score_opus":0.023966023876634945,"score_gpt":0.24392142845207074,"score_spread":0.2199554045754358,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406274339","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003923239,0.0025206436,0.005907528,0.0005911544,0.00045256698,0.0002303454,0.9137186,0.0037939344,0.06886207],"genre_scores_gemma":[0.014011452,0.005096708,0.045136336,0.00047828144,0.00023960219,0.001474271,0.88055676,0.0021446322,0.05086208],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99782383,0.00024707773,0.0003172701,0.00033077036,0.0011230593,0.00015804879],"domain_scores_gemma":[0.99018025,0.00087985117,0.0007751933,0.0029622314,0.0045752637,0.0006271609],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.004022786,0.0010852551,0.0012559408,0.015058336,0.0018732507,0.0040000617,0.0028448743,0.00073791575,0.087302156],"category_scores_gemma":[0.008510024,0.00090040243,0.00034672482,0.032667354,0.00058080646,0.0020428405,0.0030918356,0.0014495531,0.06844547],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019867341,0.00010212009,0.00907564,0.0012210656,0.00006184641,0.0001686409,0.00046297765,0.00023010504,0.0015558042,0.005225724,0.77779764,0.20389985],"study_design_scores_gemma":[0.000018196246,0.0000071395543,0.008630969,0.0003085338,0.000018123163,0.00008764345,0.00011504287,0.0001349523,0.00045336568,0.0013465603,0.9888494,0.000030152903],"about_ca_topic_score_codex":0.03154959,"about_ca_topic_score_gemma":0.0643108,"teacher_disagreement_score":0.99715513,"about_ca_system_score_codex":0.001423115,"about_ca_system_score_gemma":0.009924562,"threshold_uncertainty_score":0.29205477},"labels":[],"label_agreement":null},{"id":"W4406384985","doi":"10.1038/s41597-025-04420-2","title":"Author Correction: Typical and extreme weather datasets for studying the resilience of buildings to climate change and heatwaves","year":2025,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Wind and Air Flow Studies","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; National Research Council Canada","funders":"","keywords":"Resilience (materials science); Climate change; Extreme weather; Environmental science; Climate extremes; Climatology; Geography; Environmental resource management; Ecology; Geology; Biology","score_opus":0.09279450622727894,"score_gpt":0.3117577800378403,"score_spread":0.21896327381056138,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406384985","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008639765,0.0008877337,0.004812617,0.037268814,0.8900741,0.000067026755,0.044870578,0.002491925,0.018663242],"genre_scores_gemma":[0.03325659,0.0063860873,0.030747643,0.036119,0.10588448,0.00047169876,0.11896499,0.017483655,0.6506859],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.995031,0.000630352,0.0008380462,0.0006394423,0.002607344,0.000253881],"domain_scores_gemma":[0.95142204,0.007843571,0.0018261969,0.003684656,0.03429869,0.00092490774],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037667896,0.0020720046,0.0011982671,0.004425165,0.0026875252,0.00434318,0.0024541449,0.0029791866,0.10615592],"category_scores_gemma":[0.06078436,0.0008970299,0.001173403,0.0058003375,0.0013128282,0.0028605598,0.0025381697,0.0058657136,0.065017544],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016186596,0.0000027849553,0.0000942758,0.0000584156,0.0000047309977,0.000060182265,0.000026140699,0.00006283495,0.000041799416,0.00063100643,0.9949243,0.0040773703],"study_design_scores_gemma":[0.00002133621,0.000006770427,0.000708055,0.00019335063,0.000015693133,0.00017330002,0.00009440805,0.00021779771,0.00031289706,0.000812738,0.99741805,0.000025577045],"about_ca_topic_score_codex":0.041657746,"about_ca_topic_score_gemma":0.04797779,"teacher_disagreement_score":0.10615592,"about_ca_system_score_codex":0.0028963182,"about_ca_system_score_gemma":0.0064975307,"threshold_uncertainty_score":0.35512692},"labels":[],"label_agreement":null},{"id":"W4406438417","doi":"10.1038/s41597-024-04289-7","title":"Time series data on typhoid fever incidence during outbreaks from 2000 to 2022","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Salmonella and Campylobacter epidemiology","field":"Agricultural and Biological Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; University of Toronto","funders":"Bill and Melinda Gates Foundation","keywords":"Typhoid fever; Outbreak; Incidence (geometry); Paratyphoid fever; Virology; Series (stratigraphy); Medicine; Geography; Biology; Mathematics","score_opus":0.041881609053480345,"score_gpt":0.27675945089769405,"score_spread":0.2348778418442137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406438417","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.046656758,0.0007160914,0.0009095936,0.00022161369,0.000058113863,0.00006793421,0.94976795,0.00015793281,0.0014440048],"genre_scores_gemma":[0.08278586,0.0006906877,0.0021872674,0.00008489056,0.00004334504,0.00030931385,0.9131747,0.00002762208,0.0006962934],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99879897,0.00023657319,0.00033135264,0.00024528877,0.0002698707,0.00011781763],"domain_scores_gemma":[0.99475443,0.0023345067,0.0014720504,0.000426594,0.00082610454,0.00018631376],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012823237,0.0005428396,0.00059111905,0.005217497,0.00022043772,0.0007943306,0.00060450076,0.0006445914,0.0039723455],"category_scores_gemma":[0.009052655,0.00023005839,0.0007839286,0.0092577515,0.00021181951,0.0005056529,0.0007562872,0.00063873135,0.0016407149],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012807961,0.00035376564,0.5606041,0.0068894536,0.0013183018,0.00087140076,0.00074112695,0.032158766,0.002688179,0.003330261,0.3026021,0.087161765],"study_design_scores_gemma":[0.00016682263,0.00028916105,0.731081,0.0010404582,0.0003713973,0.00071572483,0.00088996557,0.009178541,0.0019635404,0.0018583603,0.2523361,0.00010882948],"about_ca_topic_score_codex":0.02051343,"about_ca_topic_score_gemma":0.016826328,"teacher_disagreement_score":0.02051343,"about_ca_system_score_codex":0.0008228344,"about_ca_system_score_gemma":0.0012461339,"threshold_uncertainty_score":0.040788054},"labels":[],"label_agreement":null},{"id":"W4406617993","doi":"10.1038/s41597-024-04100-7","title":"Perceptions of science, science communication, and climate change attitudes in 68 countries – the TISP dataset","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate Change Communication and Perception","field":"Social Sciences","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Trinity Western University; University of Victoria; University of Alberta; Western University","funders":"NOMIS Stiftung; Universität Hamburg; Aston University; Vetenskapsrådet; Aarhus Universitets Forskningsfond; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; Eidgenössisches Departement für Umwelt, Verkehr, Energie und Kommunikation; Agence Nationale de la Recherche; John Templeton Foundation; Deutsche Forschungsgemeinschaft; Government of Canada; Genome Canada; Harvard University; Victoria University of Wellington; Resnick Sustainability Institute for Science, Energy and Sustainability, California Institute of Technology; European Commission; Trinity Western University; Fédération Wallonie-Bruxelles; Bill and Melinda Gates Foundation","keywords":"Climate change; Perception; Science communication; Climate science; Psychology; Open science; Data science; Science education; Ecology; Computer science; Biology; Statistics; Mathematics; Mathematics education","score_opus":0.4252245965880386,"score_gpt":0.5126555831040039,"score_spread":0.08743098651596537,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406617993","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22569075,0.00016196094,0.0002946684,0.00034875466,0.000024629753,0.0001090114,0.77058816,0.000059811475,0.0027223057],"genre_scores_gemma":[0.19310482,0.00017626488,0.00096135173,0.00016108411,0.000024992763,0.001060223,0.8031745,0.000032890803,0.001303892],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9987413,0.00041021942,0.00019527883,0.00019578623,0.00020374356,0.0002537715],"domain_scores_gemma":[0.9940432,0.0018320382,0.001450809,0.00090215344,0.0012699746,0.0005019373],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0016236076,0.0005118028,0.00057817687,0.0025968219,0.00086795294,0.0013063612,0.0008368854,0.0009675509,0.008450644],"category_scores_gemma":[0.009013108,0.00036200703,0.00068216247,0.006848154,0.00035834304,0.0007203055,0.0022526712,0.0010701821,0.0040904544],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039111063,0.0003209992,0.74800843,0.00084982015,0.00042951573,0.00032658,0.0024515386,0.0017453938,0.00048804653,0.0011674123,0.22825713,0.015564002],"study_design_scores_gemma":[0.00012590077,0.000081562066,0.9298192,0.0002235614,0.00008928569,0.00017699401,0.004338623,0.00078378146,0.0002609252,0.00031514373,0.06373593,0.00004907504],"about_ca_topic_score_codex":0.047561325,"about_ca_topic_score_gemma":0.03606712,"teacher_disagreement_score":0.99869365,"about_ca_system_score_codex":0.00083577406,"about_ca_system_score_gemma":0.001015243,"threshold_uncertainty_score":0.09456897},"labels":[{"model":"gemma","categories":["sts"],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406639191","doi":"10.1038/s41597-025-04458-2","title":"‘DeltaCAN - A new data set of Canadian Arctic and subarctic coastal deltas’","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Methane Hydrates and Related Phenomena","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Nutrasource; Université du Québec à Rimouski; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Subarctic climate; Arctic; Oceanography; Geography; Data set; The arctic; Physical geography; Ecology; Environmental science; Geology; Biology; Statistics; Mathematics","score_opus":0.05409016079844076,"score_gpt":0.2735868980091791,"score_spread":0.21949673721073837,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406639191","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19842422,0.0007110603,0.0032784943,0.0003148117,0.000094468414,0.00023447901,0.78220046,0.00091251207,0.013829462],"genre_scores_gemma":[0.192737,0.00038973006,0.009846997,0.00009772268,0.000017818174,0.00019779157,0.7934002,0.00011115414,0.003201543],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992951,0.000022174354,0.000052548872,0.00015027283,0.00030990638,0.00016997397],"domain_scores_gemma":[0.9967073,0.000114268674,0.0002274902,0.00024657117,0.0023916422,0.00031255576],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045036344,0.0007361043,0.00046391197,0.007376719,0.0018843227,0.0015077097,0.0011460902,0.00043700376,0.0025983935],"category_scores_gemma":[0.0021404324,0.00026055606,0.0005398641,0.012295412,0.00045780704,0.00046660536,0.0010703268,0.000441514,0.0009613067],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004989883,0.00016891041,0.49111927,0.00079677784,0.00042996925,0.0009048851,0.002144012,0.010600084,0.0075010424,0.003469896,0.30027184,0.18209434],"study_design_scores_gemma":[0.000048061152,0.000026918038,0.6462022,0.00016917332,0.00009111676,0.00025161263,0.0018042341,0.0063264007,0.0028814208,0.0004944528,0.3415853,0.0001191246],"about_ca_topic_score_codex":0.97710675,"about_ca_topic_score_gemma":0.9892631,"teacher_disagreement_score":0.02289325,"about_ca_system_score_codex":0.0075408323,"about_ca_system_score_gemma":0.02072394,"threshold_uncertainty_score":0.054712832},"labels":[],"label_agreement":null},{"id":"W4406686892","doi":"10.1038/s41597-025-04452-8","title":"Data on insect biodiversity in a Chinese potato agroecosystem from DNA metabarcoding","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Environmental DNA in Biodiversity Studies","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"Agricultural Research Service; U.S. Department of Agriculture","keywords":"Biodiversity; Agroecosystem; Barcode; Biology; Taxon; PEST analysis; DNA barcoding; Insect pest; Insect; Ecology; Agroforestry; Crop; Solanum tuberosum; Geography; Agriculture; Agronomy; Botany","score_opus":0.05616166878685588,"score_gpt":0.2621959826184704,"score_spread":0.2060343138316145,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406686892","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9657133,0.0014943022,0.002562084,0.000073059324,0.000009560122,0.000049145237,0.028245822,0.00006000917,0.0017926487],"genre_scores_gemma":[0.8636187,0.0019404427,0.0139792375,0.00012295009,0.000018366678,0.00020470146,0.11830574,0.000038361595,0.0017716245],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.999418,0.000038758197,0.0000682285,0.00024109283,0.00015704797,0.00007689596],"domain_scores_gemma":[0.999297,0.000060225288,0.00021611115,0.00009422235,0.00024400852,0.00008845414],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005636184,0.000406391,0.00038294884,0.0042656087,0.0008338876,0.00048741902,0.00033383144,0.00023735607,0.00076382456],"category_scores_gemma":[0.00070114294,0.0001932631,0.00046973932,0.0067426837,0.0003087549,0.00029801897,0.0006011603,0.00022575339,0.00023621845],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041035967,0.00007700862,0.71725816,0.0014991965,0.0004482824,0.00074910617,0.0020379291,0.0018581551,0.16033809,0.0007207038,0.0029604055,0.11164269],"study_design_scores_gemma":[0.000009862035,0.000055925437,0.98024625,0.00005446959,0.00021587235,0.00026350867,0.00037256823,0.0010804123,0.005721301,0.00013256873,0.011822779,0.000024482837],"about_ca_topic_score_codex":0.03153537,"about_ca_topic_score_gemma":0.053996515,"teacher_disagreement_score":0.03153537,"about_ca_system_score_codex":0.0007981429,"about_ca_system_score_gemma":0.0015627235,"threshold_uncertainty_score":0.06270361},"labels":[],"label_agreement":null},{"id":"W4406693740","doi":"10.1038/s41597-025-04409-x","title":"Hydrologic outputs generated over the Great Lakes with a calibrated version of the GEM-Hydro model","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Waterloo; Environment and Climate Change Canada","funders":"","keywords":"Surface runoff; Streamflow; Environmental science; Water balance; Precipitation; Hydrology (agriculture); Snow; Drainage basin; Hydrological modelling; Surface water; Potential evaporation; Data assimilation; Climatology; Meteorology; Geography; Geology; Ecology","score_opus":0.02115464285611983,"score_gpt":0.22934834480210126,"score_spread":0.2081937019459814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406693740","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11557815,0.0001294674,0.0023784763,0.00033306298,0.000103951636,0.00008774442,0.8726518,0.0039503467,0.004787002],"genre_scores_gemma":[0.14101878,0.00013133779,0.005275195,0.00007839105,0.000021864074,0.000113449925,0.85131985,0.00038277468,0.001658295],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997764,0.000028719507,0.000015369878,0.0000661386,0.00007329846,0.00004008607],"domain_scores_gemma":[0.99934334,0.00007771796,0.000039414255,0.00014321921,0.00033795254,0.000058331985],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004140808,0.0008247666,0.00046040883,0.0008313856,0.0003950873,0.00063710473,0.001243862,0.0007718741,0.005055415],"category_scores_gemma":[0.0014031937,0.00041747632,0.0007001638,0.0021834176,0.00035477508,0.0005256389,0.00042815192,0.0008575039,0.0023008282],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005212581,0.00030371646,0.041650906,0.00053721434,0.00045701378,0.0004013893,0.00027273252,0.4972028,0.0041253413,0.003114816,0.42882943,0.022583336],"study_design_scores_gemma":[0.0010831782,0.000115407725,0.13318025,0.00017074514,0.00018111394,0.00015420376,0.00038290297,0.60331696,0.010625778,0.004189463,0.24632716,0.00027279195],"about_ca_topic_score_codex":0.33027592,"about_ca_topic_score_gemma":0.36548585,"teacher_disagreement_score":0.6697241,"about_ca_system_score_codex":0.002288046,"about_ca_system_score_gemma":0.002393995,"threshold_uncertainty_score":0.65670705},"labels":[],"label_agreement":null},{"id":"W4406823140","doi":"10.1038/s41597-025-04447-5","title":"DeepExtremeCubes: Earth system spatio-temporal data for assessing compound heatwave and drought impacts","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Sächsisches Staatsministerium für Wissenschaft und Kunst; Bundesministerium für Wirtschaft und Energie; European Commission","keywords":"Biosphere; Ecosystem; Land cover; Computer science; Structuring; Earth observation; Event (particle physics); Earth system science; Climate change; Remote sensing; Environmental science; Satellite; Data mining; Earth science; Land use; Geography; Ecology; Geology","score_opus":0.04743203166249778,"score_gpt":0.29621963667475115,"score_spread":0.2487876050122534,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406823140","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09581681,0.0011795221,0.037105747,0.0010219635,0.00032332467,0.00021402132,0.8447882,0.012026651,0.007523854],"genre_scores_gemma":[0.21579082,0.00071733637,0.049509093,0.00025325752,0.00010309066,0.00041492717,0.7308112,0.000725268,0.0016749862],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99970824,0.000054384138,0.00003078329,0.00008821719,0.000087268214,0.000031191816],"domain_scores_gemma":[0.999241,0.0001567172,0.0001156526,0.00023092135,0.00015977738,0.000095970296],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00082527526,0.00069443457,0.000436802,0.0012973747,0.0002541467,0.00081121054,0.0009960216,0.00059418846,0.0043894537],"category_scores_gemma":[0.002326973,0.0002807083,0.0005916897,0.0021316702,0.0003255792,0.0014267443,0.0015100145,0.00081892556,0.0019312918],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011665578,0.0005015417,0.16630313,0.0016585909,0.0010504572,0.0006081558,0.0007637518,0.12039686,0.018783761,0.012093916,0.48682523,0.18984804],"study_design_scores_gemma":[0.00039004366,0.0002479545,0.2136016,0.00041374945,0.0001976762,0.00033991056,0.0012995208,0.28693584,0.023532083,0.032976598,0.439811,0.00025410924],"about_ca_topic_score_codex":0.008527132,"about_ca_topic_score_gemma":0.015324807,"teacher_disagreement_score":0.008527132,"about_ca_system_score_codex":0.00037162344,"about_ca_system_score_gemma":0.00080614205,"threshold_uncertainty_score":0.016955018},"labels":[],"label_agreement":null},{"id":"W4406838870","doi":"10.1038/s41597-025-04498-8","title":"Chromosome-level genome assembly and annotation of the gynogenetic large-scale loach (Paramisgurnus dabryanus)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"Chinese Academy of Sciences; National Natural Science Foundation of China","keywords":"Biology; Genome; Cypriniformes; Cobitidae; Genetics; Sequence assembly; Gene; Reference genome; Chromosome; Evolutionary biology; Genome project; Computational biology; Fish <Actinopterygii>; Fishery","score_opus":0.029107682088781988,"score_gpt":0.26910009418465536,"score_spread":0.23999241209587338,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406838870","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6386803,0.0051113297,0.14934933,0.0010078171,0.00027316896,0.00071674056,0.19019495,0.004098666,0.010567681],"genre_scores_gemma":[0.35164753,0.0024602273,0.19407173,0.0003336491,0.00006031326,0.0007035022,0.43880472,0.0018132553,0.01010496],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998084,0.000012374219,0.000015389658,0.000095802425,0.000041807103,0.000026234875],"domain_scores_gemma":[0.99966025,0.000059413458,0.0000618167,0.000044481312,0.000114561655,0.000059489656],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00026267246,0.0006423303,0.0007001751,0.0015161999,0.000761135,0.0008515977,0.00057654444,0.0005615245,0.003235225],"category_scores_gemma":[0.0007241312,0.0003591041,0.00094280863,0.0018207037,0.00023038808,0.00042454936,0.0007634167,0.0010908222,0.0022703854],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008562577,0.00010696645,0.014801144,0.0018964479,0.0001863707,0.0011754248,0.0012591402,0.004702017,0.89751357,0.0017047891,0.007428006,0.06836986],"study_design_scores_gemma":[0.00035305263,0.00055360794,0.5028432,0.0009479027,0.0011638678,0.00250646,0.0015166074,0.03217384,0.1863895,0.0032782312,0.26793888,0.00033486687],"about_ca_topic_score_codex":0.011377123,"about_ca_topic_score_gemma":0.021240858,"teacher_disagreement_score":0.011377123,"about_ca_system_score_codex":0.0005691148,"about_ca_system_score_gemma":0.0015952673,"threshold_uncertainty_score":0.02262181},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"bench_or_experimental","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406872169","doi":"10.1038/s41597-025-04465-3","title":"The UnconTrust Database for Studies of Unconscious Semantic Processing and Attentional Allocation","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neural and Behavioral Psychology Studies","field":"Neuroscience","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research","funders":"Templeton World Charity Foundation; Canadian Institute for Advanced Research","keywords":"Unconscious mind; Computer science; Scope (computer science); Field (mathematics); Information processing; Database; Diversity (politics); Data science; Cognitive psychology; Psychology; Sociology","score_opus":0.31316931525912817,"score_gpt":0.45988314866995456,"score_spread":0.1467138334108264,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406872169","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032771078,0.018525632,0.0067769857,0.00064466294,0.00034650165,0.00061322545,0.9447965,0.0029387549,0.022080708],"genre_scores_gemma":[0.011172181,0.013455374,0.014246413,0.0008339188,0.00033744593,0.0029053104,0.94799995,0.0018674417,0.0071819024],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9919641,0.001343753,0.002821309,0.00095797924,0.002472627,0.00044017],"domain_scores_gemma":[0.8819541,0.06791543,0.012323983,0.02218246,0.01203956,0.003584485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007544574,0.0017746448,0.0039535533,0.02858317,0.0009869429,0.004518128,0.004381554,0.0025199386,0.091299504],"category_scores_gemma":[0.055833265,0.00097496907,0.0014480733,0.02898569,0.0009661217,0.0038375293,0.004794762,0.0018189723,0.03734531],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002257791,0.00034168264,0.0046304557,0.05216791,0.0007986081,0.00034414217,0.0005749688,0.0005965624,0.0068487762,0.009232332,0.6652553,0.25695148],"study_design_scores_gemma":[0.000210109,0.00016230407,0.01030984,0.0063732057,0.00054031855,0.00043626703,0.00012215514,0.00019193141,0.0024778584,0.0051623685,0.973931,0.00008261018],"about_ca_topic_score_codex":0.0027579921,"about_ca_topic_score_gemma":0.004588551,"teacher_disagreement_score":0.091299504,"about_ca_system_score_codex":0.0013940339,"about_ca_system_score_gemma":0.0060001565,"threshold_uncertainty_score":0.30542725},"labels":[],"label_agreement":null},{"id":"W4407063052","doi":"10.1038/s41597-025-04382-5","title":"Investigating the Quality of DermaMNIST and Fitzpatrick17k Dermatological Image Datasets","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cutaneous Melanoma Detection and Management","field":"Medicine","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Alliance de recherche numérique du Canada; Natural Sciences and Engineering Research Council of Canada; Government of Canada; Simon Fraser University; Nvidia","keywords":"Quality (philosophy); Computer science; Information retrieval; Dermatology; Medicine; Physics","score_opus":0.0761432071486184,"score_gpt":0.3701774839851385,"score_spread":0.2940342768365201,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407063052","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7337761,0.008439236,0.04597293,0.008462709,0.0015234047,0.00077709596,0.17014474,0.015344274,0.015559533],"genre_scores_gemma":[0.5538579,0.0011038217,0.078015745,0.0015546168,0.00018295122,0.0004300923,0.3594285,0.0016087577,0.003817603],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9922766,0.001593929,0.0009195678,0.00171636,0.0030343698,0.0004592716],"domain_scores_gemma":[0.98428637,0.006434253,0.0009642713,0.004036118,0.0038013826,0.00047771115],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.008642409,0.0012448984,0.0008141618,0.0045906943,0.0011655388,0.0026236225,0.0022069477,0.0017421641,0.0018384692],"category_scores_gemma":[0.035194043,0.00035300353,0.0011960525,0.004097757,0.0014574517,0.0019701424,0.0022537368,0.0019557497,0.0013772872],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003160387,0.0015335338,0.14124604,0.004176619,0.0024587873,0.0016508424,0.0013131215,0.06885075,0.035709325,0.0075284382,0.39569774,0.3366744],"study_design_scores_gemma":[0.00082447525,0.001150083,0.3008855,0.0010825845,0.00070475484,0.006379714,0.0032972659,0.2902551,0.10737306,0.019515833,0.26804563,0.00048594983],"about_ca_topic_score_codex":0.015283884,"about_ca_topic_score_gemma":0.025461758,"teacher_disagreement_score":0.99135756,"about_ca_system_score_codex":0.0021752827,"about_ca_system_score_gemma":0.0016190629,"threshold_uncertainty_score":0.045705974},"labels":[],"label_agreement":null},{"id":"W4407065613","doi":"10.1038/s41597-025-04396-z","title":"Bias-corrected high-resolution temperature and precipitation projections for Canada","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Manitoba; University of Saskatchewan; University of Calgary","funders":"Global Water Futures","keywords":"Precipitation; Environmental science; Climatology; Geography; Geology; Meteorology","score_opus":0.036079763683100945,"score_gpt":0.26615225314414787,"score_spread":0.23007248946104691,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407065613","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.714934,0.0009919179,0.045365073,0.0010852385,0.00006351172,0.00013471434,0.22246647,0.0026085747,0.0123505285],"genre_scores_gemma":[0.8755766,0.0007418554,0.031215465,0.000087326815,0.000011027584,0.00008366945,0.09084364,0.0001961603,0.0012441976],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996406,0.000043771124,0.00001921203,0.000059212725,0.00016189256,0.00007526334],"domain_scores_gemma":[0.99912184,0.00007811488,0.00006321443,0.00011381809,0.00055689557,0.000066099274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00085090974,0.000441368,0.00026401447,0.00081406673,0.00064270844,0.0006318216,0.00094541005,0.00032743972,0.0015442359],"category_scores_gemma":[0.0018474619,0.000292538,0.0006598158,0.0029508446,0.00023388903,0.00045445585,0.0004002326,0.00051197235,0.00026529707],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000926695,0.000048050177,0.04906117,0.00010625249,0.00018451057,0.000071656985,0.0000784514,0.9058484,0.0017106746,0.0031852685,0.011261671,0.028351156],"study_design_scores_gemma":[0.00011684158,0.000024792927,0.13685524,0.000062156854,0.00008888516,0.00006432526,0.00017496494,0.8204167,0.0043404875,0.0032951497,0.034436785,0.00012367345],"about_ca_topic_score_codex":0.94916874,"about_ca_topic_score_gemma":0.946452,"teacher_disagreement_score":0.05083126,"about_ca_system_score_codex":0.009548589,"about_ca_system_score_gemma":0.017437303,"threshold_uncertainty_score":0.102261186},"labels":[],"label_agreement":null},{"id":"W4407238378","doi":"10.1038/s41597-025-04547-2","title":"A single-nucleus RNA sequencing atlas of the postnatal retina of the shark Scyliorhinus canicula","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"European Regional Development Fund; Natural Sciences and Engineering Research Council of Canada; Xunta de Galicia; European Commission; Biotechnology and Biological Sciences Research Council; Spanish National Plan for Scientific and Technical Research and Innovation; Agencia Estatal de Investigación; Ministerio de Ciencia, Innovación y Universidades","keywords":"Scyliorhinus canicula; Biology; Retina; Vertebrate; Retinal; Anatomy; Evolutionary biology; Neuroscience; Genetics; Fishery","score_opus":0.029446579591905907,"score_gpt":0.24564587453646644,"score_spread":0.21619929494456053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407238378","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7649419,0.0031269295,0.02927803,0.00012311901,0.00008102141,0.00009508312,0.18997882,0.001619035,0.010755974],"genre_scores_gemma":[0.6399438,0.0019913418,0.056633413,0.00019209491,0.00004126585,0.00044902012,0.29242232,0.000571422,0.00775538],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998512,0.0000069601465,0.000008674053,0.000081547114,0.000031931475,0.000019575264],"domain_scores_gemma":[0.9998375,0.000026034997,0.000045495522,0.000022998911,0.00004505829,0.000022832166],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00014427485,0.00026673792,0.00022159485,0.0010195014,0.00046349276,0.00023425616,0.00015945743,0.00019662066,0.001600546],"category_scores_gemma":[0.00022568602,0.00012115727,0.00028270116,0.0011246101,0.00018948546,0.00013522286,0.00034211972,0.00022005763,0.0007440896],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021319924,0.000018049217,0.038496997,0.0006624434,0.000120003286,0.00036047693,0.00044669755,0.000981606,0.91313213,0.00068533607,0.004048538,0.040834498],"study_design_scores_gemma":[0.00002087118,0.00015054073,0.7805551,0.000111150395,0.00030694527,0.0017021846,0.0004648697,0.0043472345,0.11079703,0.0010921985,0.10039538,0.00005650917],"about_ca_topic_score_codex":0.00616014,"about_ca_topic_score_gemma":0.024357535,"teacher_disagreement_score":0.00616014,"about_ca_system_score_codex":0.0002732273,"about_ca_system_score_gemma":0.0008179558,"threshold_uncertainty_score":0.012248576},"labels":[],"label_agreement":null},{"id":"W4407251531","doi":"10.1038/s41597-025-04548-1","title":"An Integrated Database for Exploring Alternative Promoters in Animals","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"Huazhong Agricultural University; Fundamental Research Funds for the Central Universities; Natural Science Foundation of Hubei Province; National Natural Science Foundation of China","keywords":"Biology; Promoter; Enhancer; Computational biology; Database; Transcription factor; Transcription (linguistics); Genetics; Gene expression; Gene; Computer science","score_opus":0.06597041708514718,"score_gpt":0.3276484842748255,"score_spread":0.26167806718967834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407251531","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006734607,0.0047357497,0.03533151,0.0002234337,0.00012368492,0.00030143134,0.8971324,0.047236703,0.008180378],"genre_scores_gemma":[0.007921058,0.0021622507,0.039274316,0.00016276058,0.0000370629,0.00041165177,0.9458967,0.0028928025,0.001241512],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990416,0.00014400472,0.00026588814,0.00025334646,0.00020736738,0.00008777624],"domain_scores_gemma":[0.99802446,0.0007893603,0.00033858395,0.00034919285,0.0002371525,0.00026130225],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00198731,0.002772472,0.0023969337,0.009383588,0.0007550389,0.002428577,0.002976801,0.0022347206,0.027186593],"category_scores_gemma":[0.004688854,0.0013652205,0.001467131,0.010288733,0.00053439656,0.0030250438,0.0031132053,0.0019529697,0.023310632],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0035975133,0.0008269957,0.016390488,0.037002306,0.0013264705,0.0036119448,0.0019445868,0.007426452,0.1284842,0.01975922,0.5021179,0.2775119],"study_design_scores_gemma":[0.00042551773,0.000249814,0.013636632,0.0014670977,0.00048698828,0.0015756197,0.00034687525,0.0064946017,0.013688059,0.009808585,0.951597,0.00022321264],"about_ca_topic_score_codex":0.002351938,"about_ca_topic_score_gemma":0.00268089,"teacher_disagreement_score":0.027186593,"about_ca_system_score_codex":0.00088487356,"about_ca_system_score_gemma":0.0020383087,"threshold_uncertainty_score":0.090948224},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4407251902","doi":"10.1038/s41597-025-04491-1","title":"BenthicNet: A global compilation of seafloor images for deep learning applications","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Underwater Acoustics Research","field":"Earth and Planetary Sciences","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Tula Foundation; Fisheries and Oceans Canada; Dalhousie University; Memorial University of Newfoundland; Bedford Institute of Oceanography; Vector Institute","funders":"Alliance de recherche numérique du Canada; Hakai Institute; Ocean Frontier Institute; Australian Centre for Field Robotics; Atlantic Canada Opportunities Agency; University of Tasmania; Australian Government","keywords":"Seafloor spreading; Geology; Deep learning; Computer science; Artificial intelligence; Oceanography","score_opus":0.04507460497129716,"score_gpt":0.33209781881484973,"score_spread":0.28702321384355256,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407251902","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.042552017,0.0012703233,0.0800452,0.00022107207,0.00039727445,0.000428694,0.7594782,0.101326376,0.014280829],"genre_scores_gemma":[0.028354485,0.00045186238,0.06622637,0.00009133666,0.000029888204,0.00035783014,0.89707017,0.0046006413,0.0028173598],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996568,0.000022331447,0.00003289859,0.000118117474,0.00012832368,0.0000415076],"domain_scores_gemma":[0.99949014,0.000050012328,0.000051552917,0.0001159548,0.00022501439,0.00006720466],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00038041925,0.0022293525,0.0005786954,0.003249897,0.00045101004,0.0008644126,0.001551375,0.00058011466,0.010378279],"category_scores_gemma":[0.001531482,0.0007726159,0.00073761906,0.0027322294,0.00033007626,0.0013001096,0.0016774918,0.0011264832,0.008332759],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005194383,0.00021089612,0.013095537,0.0018678516,0.00038980044,0.0005990204,0.00036529644,0.01862415,0.025878793,0.002752744,0.7283026,0.207394],"study_design_scores_gemma":[0.0002286314,0.0002658552,0.045832112,0.0005658633,0.00020728633,0.0008175468,0.00042719915,0.08761256,0.04906721,0.0055861967,0.80915505,0.00023448041],"about_ca_topic_score_codex":0.015331269,"about_ca_topic_score_gemma":0.0346648,"teacher_disagreement_score":0.015331269,"about_ca_system_score_codex":0.0006268991,"about_ca_system_score_gemma":0.0013855927,"threshold_uncertainty_score":0.034718752},"labels":[],"label_agreement":null},{"id":"W4407391025","doi":"10.1038/s41597-025-04544-5","title":"A high resolution, gridded product for vapor pressure deficit using Daymet","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Science Foundation","keywords":"Vapour Pressure Deficit; Product (mathematics); Vapor pressure; Environmental science; Resolution (logic); Computer science; Thermodynamics; Mathematics; Botany; Physics; Biology; Transpiration; Artificial intelligence","score_opus":0.07467332530574496,"score_gpt":0.3049825374918117,"score_spread":0.23030921218606676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407391025","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16280577,0.00040759044,0.037378106,0.0002163059,0.00022678198,0.00028228076,0.7648651,0.015344749,0.018473305],"genre_scores_gemma":[0.3037839,0.00027982777,0.07801118,0.00010307054,0.00007772119,0.00045466796,0.61064494,0.0015413449,0.0051033404],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997907,0.000018970763,0.000016872335,0.0000746702,0.000078427394,0.000020310292],"domain_scores_gemma":[0.9995933,0.00003489766,0.00006444116,0.00007052103,0.00021309587,0.000023793305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002684918,0.00057321903,0.00050838594,0.0011666085,0.00025085005,0.000752854,0.0007107296,0.00043380287,0.008390064],"category_scores_gemma":[0.00072104763,0.0002946266,0.00043994375,0.00287772,0.00012837368,0.00072750665,0.00031861558,0.00050341355,0.0030012326],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00061674364,0.0004670244,0.20043737,0.0010865957,0.0008192666,0.00047426068,0.0004960663,0.08424652,0.033974856,0.004585652,0.3944979,0.27829778],"study_design_scores_gemma":[0.00027332563,0.00008098862,0.49711648,0.00012508547,0.000118938995,0.00026846025,0.0002917541,0.15436603,0.019766182,0.0022679195,0.32513377,0.00019113823],"about_ca_topic_score_codex":0.042806055,"about_ca_topic_score_gemma":0.05763975,"teacher_disagreement_score":0.042806055,"about_ca_system_score_codex":0.0004913531,"about_ca_system_score_gemma":0.0006333904,"threshold_uncertainty_score":0.08511382},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4407574495","doi":"10.1038/s41597-025-04532-9","title":"A chromosome-level genome assembly of Mylabris sibirica Fischer von Waldheim, 1823 (Coleoptera, Meloidae)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Beetle Biology and Toxicology Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"","keywords":"Biology; Chromosome; Zoology; Genetics; Gene","score_opus":0.047848173343356344,"score_gpt":0.30340353535684067,"score_spread":0.25555536201348433,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407574495","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.730658,0.002278012,0.08788952,0.00031531532,0.00017903722,0.00048937445,0.16561949,0.0026273173,0.009943872],"genre_scores_gemma":[0.4145081,0.0017677683,0.17262806,0.00013456265,0.000032535416,0.0005663111,0.3992001,0.00086532504,0.010297214],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998996,0.000008348654,0.00000836429,0.000042800646,0.000029743114,0.000011042232],"domain_scores_gemma":[0.9998652,0.000018624056,0.00002270516,0.000016837777,0.000057174097,0.00001952393],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00021435568,0.0006179661,0.00043484723,0.0012154359,0.00062468933,0.00038167028,0.00030128317,0.00029014563,0.0035270532],"category_scores_gemma":[0.00044964172,0.00027346396,0.00063955825,0.0012982306,0.00010317766,0.00026249498,0.00032732828,0.00056372985,0.0015316082],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005241211,0.00005458286,0.009272107,0.00061267056,0.00010309048,0.00049888063,0.00046799888,0.002136721,0.9327477,0.0009126015,0.0034399384,0.04922947],"study_design_scores_gemma":[0.00025793447,0.0006508883,0.5214788,0.0003755685,0.0008084029,0.0015444141,0.0007079078,0.040613636,0.22803396,0.0020619906,0.20330203,0.00016438196],"about_ca_topic_score_codex":0.011522571,"about_ca_topic_score_gemma":0.01981012,"teacher_disagreement_score":0.011522571,"about_ca_system_score_codex":0.00047634693,"about_ca_system_score_gemma":0.0009114414,"threshold_uncertainty_score":0.022911012},"labels":[],"label_agreement":null},{"id":"W4407635770","doi":"10.1038/s41597-025-04580-1","title":"A semantic approach to mapping the Provenance Ontology to Basic Formal Ontology","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Computer science; Ontology; Upper ontology; Ontology-based data integration; Information retrieval; Variety (cybernetics); Open Biomedical Ontologies; Interoperability; OWL-S; Process ontology; SPARQL; Ontology alignment; Semantic Web; Formal ontology; Consistency (knowledge bases); Suggested Upper Merged Ontology; Ontology components; World Wide Web; Semantic Web Stack; RDF; Artificial intelligence","score_opus":0.3061377680787358,"score_gpt":0.40577316655683815,"score_spread":0.09963539847810238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407635770","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009629242,0.00016594166,0.98676586,0.0014636279,0.000186081,0.0001706696,0.00021646525,0.0002903423,0.009778124],"genre_scores_gemma":[0.05095433,0.00081084605,0.93832904,0.00070043234,0.00027613153,0.00044985698,0.0009304902,0.00024645118,0.0073022963],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9938856,0.0023330483,0.0007390435,0.0007381338,0.0019310816,0.00037302246],"domain_scores_gemma":[0.992311,0.0028788086,0.000473456,0.0021023683,0.0019320772,0.00030231764],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010551395,0.00083904405,0.0008221933,0.0067193536,0.0039360146,0.007763915,0.0029604111,0.0024467958,0.0067825015],"category_scores_gemma":[0.016434727,0.0011052939,0.0031700765,0.0060439687,0.0073160087,0.015532153,0.0065153437,0.0054543633,0.001731317],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000006926273,0.000023998376,0.0001039904,0.000048600585,0.000010126425,0.00010273791,0.0004754435,0.0014044597,0.0002789746,0.9854295,0.0016006259,0.010514581],"study_design_scores_gemma":[0.000008835874,0.000021598722,0.00011012423,0.00011984817,0.000023151284,0.00024131461,0.0005495782,0.013585456,0.00085178076,0.90702784,0.07743725,0.000023103266],"about_ca_topic_score_codex":0.010408078,"about_ca_topic_score_gemma":0.0088537345,"teacher_disagreement_score":0.010551395,"about_ca_system_score_codex":0.0037117763,"about_ca_system_score_gemma":0.0069447225,"threshold_uncertainty_score":0.05580175},"labels":[],"label_agreement":null},{"id":"W4407856910","doi":"10.1038/s41597-025-04630-8","title":"A standardized lexicon of body odor words crafted from 17 countries","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Olfactory and Sensory Function Studies","field":"Neuroscience","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary; Western University","funders":"","keywords":"Operationalization; Digitization; Nonverbal communication; Odor; Perception; Cognitive psychology; Lexicon; Computer science; Psychology; Communication; Natural language processing; Computer vision","score_opus":0.15883060814098482,"score_gpt":0.328983230531369,"score_spread":0.1701526223903842,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407856910","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11208233,0.0016634081,0.018063933,0.00036996702,0.00046563716,0.00250794,0.8374124,0.0028814613,0.024552917],"genre_scores_gemma":[0.12268551,0.0012758247,0.05522259,0.0003461291,0.00012247713,0.006737325,0.8046846,0.0009806083,0.007944893],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9988581,0.00019505911,0.00045766716,0.00023501216,0.00018160732,0.00007248622],"domain_scores_gemma":[0.99710387,0.001359893,0.00042348486,0.0003552448,0.0006221002,0.00013535503],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00084595894,0.0014094261,0.00083558063,0.007700958,0.0006750559,0.0015362172,0.0006320339,0.0007882546,0.025061145],"category_scores_gemma":[0.0045510754,0.0003386109,0.0007213613,0.006287587,0.0007608526,0.0015571801,0.0022403756,0.0009533927,0.015311185],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025099497,0.0004827305,0.03887684,0.020962045,0.00031799838,0.0028616115,0.009721758,0.002441892,0.07478698,0.009869868,0.31395742,0.5232109],"study_design_scores_gemma":[0.00022463742,0.0002982964,0.1296747,0.001610425,0.00014249726,0.003127312,0.004484235,0.0023524351,0.009608211,0.0029352852,0.84527314,0.00026881375],"about_ca_topic_score_codex":0.002472635,"about_ca_topic_score_gemma":0.0052101933,"teacher_disagreement_score":0.025061145,"about_ca_system_score_codex":0.00077357766,"about_ca_system_score_gemma":0.0016324066,"threshold_uncertainty_score":0.08383793},"labels":[],"label_agreement":null},{"id":"W4407877295","doi":"10.1038/s41597-025-04451-9","title":"Applying the FAIR Principles to computational workflows","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"Biotechnology and Biological Sciences Research Council; Agence Nationale de la Recherche","keywords":"Workflow; Computer science; Data science; Database","score_opus":0.356785298571699,"score_gpt":0.4538253241408091,"score_spread":0.09704002556911012,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407877295","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030371107,0.0022883068,0.8258602,0.114484705,0.0033427007,0.00045998028,0.00025245274,0.0006460896,0.049628537],"genre_scores_gemma":[0.1839716,0.00459761,0.7527317,0.025278637,0.003392521,0.0022398117,0.00050140766,0.001371591,0.0259151],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9049724,0.052072044,0.008360204,0.011731869,0.018208113,0.0046554045],"domain_scores_gemma":[0.8286099,0.10299542,0.005242224,0.035870314,0.022026798,0.005255349],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.14102612,0.0014886255,0.0019724278,0.0053738807,0.015902122,0.025913222,0.01002037,0.012418239,0.009640397],"category_scores_gemma":[0.20014617,0.001963477,0.0053674136,0.0054047415,0.06772124,0.042864423,0.016710235,0.020095287,0.0031744791],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000544666,0.000006708687,0.00007392951,0.000047090565,0.0000072980624,0.000030339132,0.0006406812,0.0007869413,0.00001988401,0.9909718,0.003625992,0.0037838335],"study_design_scores_gemma":[0.000011104897,0.000005078939,0.000032478016,0.00011198725,0.0000069825355,0.00002716786,0.00018819426,0.0012781585,0.0000970547,0.9530613,0.045163155,0.00001728171],"about_ca_topic_score_codex":0.027859673,"about_ca_topic_score_gemma":0.019989084,"teacher_disagreement_score":0.9899796,"about_ca_system_score_codex":0.017094256,"about_ca_system_score_gemma":0.036531474,"threshold_uncertainty_score":0.74582636},"labels":[],"label_agreement":null},{"id":"W4408067639","doi":"10.1038/s41597-025-04621-9","title":"A dataset on formulation parameters and characteristics of drug-loaded PLGA microparticles","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Drug Delivery Systems","field":"Pharmacology, Toxicology and Pharmaceutics","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium; University of Toronto","funders":"","keywords":"PLGA; Drug; Chemistry; Nanotechnology; Materials science; Pharmacology; Medicine; Nanoparticle","score_opus":0.16381189512461256,"score_gpt":0.4457319390361242,"score_spread":0.2819200439115116,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408067639","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008453427,0.0036065476,0.0013426973,0.00022147069,0.00009385479,0.0000941914,0.98351306,0.00097282947,0.0017018944],"genre_scores_gemma":[0.011402854,0.0013637574,0.0032720703,0.00019286398,0.000032750642,0.0003087739,0.9825546,0.00011368661,0.0007586769],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99806625,0.00025992354,0.00041551515,0.0005615367,0.0005645657,0.00013224913],"domain_scores_gemma":[0.9930641,0.003888249,0.00086180324,0.0008005236,0.0011672013,0.0002180092],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018750204,0.0017303448,0.001668412,0.0041428427,0.00042791342,0.0017436667,0.0014471131,0.0023339533,0.009046309],"category_scores_gemma":[0.008256228,0.00035977093,0.0017533111,0.00467931,0.00035008646,0.00079855224,0.0009342898,0.0011274365,0.009618719],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023721007,0.0011365362,0.043688215,0.016543413,0.0010547638,0.00084182684,0.00015026807,0.017854853,0.012171052,0.0017476003,0.7406524,0.16178696],"study_design_scores_gemma":[0.00046645454,0.0005147566,0.047442093,0.0016151576,0.0005100535,0.0010090832,0.00016117778,0.0109623205,0.009243886,0.0028140186,0.92508626,0.00017466609],"about_ca_topic_score_codex":0.0045909733,"about_ca_topic_score_gemma":0.0066011813,"teacher_disagreement_score":0.009046309,"about_ca_system_score_codex":0.0009782373,"about_ca_system_score_gemma":0.0015104976,"threshold_uncertainty_score":0.030262947},"labels":[],"label_agreement":null},{"id":"W4408170081","doi":"10.1038/s41597-025-04724-3","title":"Multifaceted and extensive behavioral trajectories of genomically diverse Drosophila lines","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neurobiology and Insect Physiology Research","field":"Neuroscience","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics; ACT-X; Japan Science and Technology Agency; Japan Society for the Promotion of Science; Japan Science Society","keywords":"Drosophila (subgenus); Computational biology; Evolutionary biology; Computer science; Biology; Genetics; Gene","score_opus":0.11419918126394273,"score_gpt":0.3710138542053832,"score_spread":0.25681467294144045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408170081","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7672197,0.0028577894,0.029110542,0.0002927509,0.00009577157,0.00010265318,0.1944261,0.0017092328,0.0041854586],"genre_scores_gemma":[0.5734314,0.0018713403,0.038255613,0.00019515981,0.00004433403,0.0002850404,0.38350573,0.00042583366,0.0019855553],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99961996,0.000035351277,0.00002429414,0.00016331804,0.00011140679,0.0000456451],"domain_scores_gemma":[0.99922013,0.00016426909,0.00015580877,0.00018095624,0.00019241153,0.000086488966],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004013934,0.0004407031,0.000605595,0.0016822763,0.00045004438,0.00043274346,0.00033746363,0.00031945383,0.00091664545],"category_scores_gemma":[0.0010306819,0.00024566302,0.0004886171,0.0018245131,0.00022895048,0.0003568303,0.0006488356,0.00060909416,0.00075248786],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046257008,0.00027989244,0.52568954,0.0012799265,0.0010432472,0.0010450241,0.0009688763,0.018750964,0.28173178,0.0022069342,0.04250658,0.12403468],"study_design_scores_gemma":[0.000020519277,0.00013665012,0.9343844,0.000098510005,0.00016752498,0.0008460595,0.00037802884,0.01274508,0.013982815,0.0016999037,0.035461888,0.00007859559],"about_ca_topic_score_codex":0.008664903,"about_ca_topic_score_gemma":0.029623164,"teacher_disagreement_score":0.008664903,"about_ca_system_score_codex":0.00038525218,"about_ca_system_score_gemma":0.000340326,"threshold_uncertainty_score":0.017228961},"labels":[],"label_agreement":null},{"id":"W4408195974","doi":"10.1038/s41597-025-04715-4","title":"A Comprehensive Dataset of Surface Water Quality Spanning 1940-2023 for Empirical and ML Adopted Research","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Water Quality and Pollution Assessment","field":"Environmental Science","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Water quality; Quality (philosophy); Environmental science; Surface water; Data science; Computer science; Biology; Ecology; Environmental engineering; Physics","score_opus":0.4412245770869852,"score_gpt":0.520185085393014,"score_spread":0.0789605083060288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408195974","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018389009,0.00030021233,0.0016637596,0.00029321905,0.000057241203,0.000047709465,0.97728753,0.0002422212,0.0017191442],"genre_scores_gemma":[0.03382837,0.00016089405,0.0024783623,0.00008991953,0.000021978012,0.00016770767,0.96272177,0.000029658704,0.00050132687],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99910337,0.00015769897,0.00015060003,0.00024613366,0.00023876395,0.000103376755],"domain_scores_gemma":[0.9970841,0.0007339667,0.0005630358,0.0005643394,0.0008783988,0.00017613705],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010969568,0.0006254075,0.0005800823,0.0027769224,0.00036575767,0.0006958436,0.0010573253,0.0009791221,0.0048874365],"category_scores_gemma":[0.0043909852,0.0002141514,0.00082503696,0.00856048,0.00032014534,0.00078535377,0.0012049295,0.0008742703,0.00303905],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034016778,0.00032444985,0.23837298,0.0016406082,0.0005705231,0.00041772577,0.0003085438,0.020990865,0.0021631466,0.004132219,0.67028713,0.060451668],"study_design_scores_gemma":[0.00014558894,0.0001355362,0.43650436,0.0004967242,0.0001634054,0.00032764062,0.0008775764,0.014598636,0.0024949093,0.004188466,0.5399461,0.0001211923],"about_ca_topic_score_codex":0.042044483,"about_ca_topic_score_gemma":0.050922498,"teacher_disagreement_score":0.042044483,"about_ca_system_score_codex":0.0011816139,"about_ca_system_score_gemma":0.0016835339,"threshold_uncertainty_score":0.08359951},"labels":[],"label_agreement":null},{"id":"W4408279735","doi":"10.1038/s41597-024-04193-0","title":"Harnessing Smartphone Sensors for Enhanced Road Safety: A Comprehensive Dataset and Review","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Autonomous Vehicle Technology and Safety","field":"Engineering","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Data science","score_opus":0.027734023176698493,"score_gpt":0.29099685300039113,"score_spread":0.26326282982369265,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408279735","genre_codex":"review","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.045760456,0.6011942,0.022077763,0.0083441185,0.002898633,0.0011337481,0.292873,0.0015243464,0.02419386],"genre_scores_gemma":[0.12507293,0.40461195,0.023981998,0.0031487653,0.001113153,0.0014079149,0.43648985,0.00021014884,0.003963189],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99858475,0.00030324387,0.00025596208,0.00027925317,0.0004942082,0.000082572275],"domain_scores_gemma":[0.9948625,0.002415355,0.0004404185,0.000375979,0.0018010446,0.000104720355],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018069751,0.0010814056,0.0009296374,0.00399452,0.00041609883,0.001231725,0.0013304772,0.0010758231,0.002105063],"category_scores_gemma":[0.0070377993,0.0002950128,0.0012393349,0.004091449,0.00035516662,0.0016734335,0.0011383877,0.00071085186,0.0014623676],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031947513,0.00023259429,0.03321749,0.049762152,0.00089481135,0.00032153892,0.00039344258,0.0044554817,0.0029489445,0.004272891,0.25528073,0.64790046],"study_design_scores_gemma":[0.00005974523,0.00034856863,0.0857531,0.016088419,0.0014287675,0.0007089771,0.0011339849,0.0067609185,0.0035466999,0.0036539333,0.8802458,0.00027105922],"about_ca_topic_score_codex":0.008444105,"about_ca_topic_score_gemma":0.017119728,"teacher_disagreement_score":0.008444105,"about_ca_system_score_codex":0.00056655565,"about_ca_system_score_gemma":0.0020082577,"threshold_uncertainty_score":0.016789913},"labels":[],"label_agreement":null},{"id":"W4408429731","doi":"10.1038/s41597-025-04763-w","title":"Electromyographic typing gesture classification dataset for neurotechnological human-machine interfaces","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Muscle activation and electromyography studies","field":"Engineering","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Rehabilitation Institute; University of Toronto; University Health Network","funders":"","keywords":"Session (web analytics); Computer science; Task (project management); Electromyography; Typing; Interface (matter); Gesture; Artificial intelligence; Key (lock); Machine learning; Speech recognition; Human–computer interaction; Pattern recognition (psychology); Physical medicine and rehabilitation; Medicine; Engineering; World Wide Web","score_opus":0.05592928662354131,"score_gpt":0.31274121495168883,"score_spread":0.25681192832814753,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408429731","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07673438,0.0031775138,0.010263426,0.00059282465,0.0007889651,0.0016459103,0.8910951,0.008189646,0.0075122425],"genre_scores_gemma":[0.029704383,0.00034215703,0.0076185768,0.0001310846,0.00008318881,0.0012805621,0.9569526,0.00012474503,0.0037627604],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985135,0.00014110465,0.00023322158,0.0004186588,0.00049595523,0.00019747255],"domain_scores_gemma":[0.9987482,0.00019074706,0.00012848881,0.0003752171,0.0004255434,0.00013187421],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007603553,0.0035344418,0.0019868691,0.002708651,0.0007715795,0.0010756612,0.0022601073,0.0023901842,0.009913941],"category_scores_gemma":[0.0025301967,0.00032239562,0.0017658209,0.0020980402,0.000474747,0.0005961606,0.0017697033,0.0015731391,0.016706271],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023071223,0.0023063843,0.024756137,0.0044224695,0.0008087466,0.0019388797,0.00025076748,0.006877199,0.027325958,0.0008857102,0.7124662,0.21565446],"study_design_scores_gemma":[0.0014656573,0.0020809565,0.22144647,0.0013286876,0.00063568994,0.006929365,0.0010050803,0.056456797,0.04186835,0.0032192797,0.66307545,0.00048827782],"about_ca_topic_score_codex":0.0074535245,"about_ca_topic_score_gemma":0.021849245,"teacher_disagreement_score":0.009913941,"about_ca_system_score_codex":0.0006651081,"about_ca_system_score_gemma":0.0009636126,"threshold_uncertainty_score":0.033165455},"labels":[],"label_agreement":null},{"id":"W4408663620","doi":"10.1038/s41597-025-04809-z","title":"A spatially constrained remote sensing-based inventory of glacial lakes worldwide","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cryospheric studies and observations","field":"Earth and Planetary Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Glacial period; Paleolimnology; Environmental science; Remote sensing; Physical geography; Geography; Geology; Oceanography; Climate change; Geomorphology","score_opus":0.04030912102009116,"score_gpt":0.2574513166487394,"score_spread":0.21714219562864823,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408663620","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9855664,0.00013998893,0.0037850742,0.000049790073,0.0000035556577,0.000029012948,0.008045825,0.00013811818,0.0022422408],"genre_scores_gemma":[0.9772321,0.00012769268,0.009949763,0.000019985737,0.00000654204,0.0000423021,0.01205819,0.000011940459,0.0005516063],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998752,0.000025647292,0.000014487667,0.000035712757,0.000030854837,0.000018060344],"domain_scores_gemma":[0.999721,0.00004525336,0.00008134209,0.00005580567,0.00007412366,0.000022355853],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003335127,0.0002906127,0.00016015577,0.0017000522,0.00016302621,0.0005028016,0.00022990815,0.00015395958,0.0010627932],"category_scores_gemma":[0.0005807088,0.00015850936,0.0003000972,0.0015455871,0.00010775139,0.00072948105,0.0003774279,0.00009327357,0.00022012331],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013592331,0.00017155088,0.7718253,0.0001354279,0.00028693388,0.00020027108,0.0003655744,0.08155087,0.01596422,0.0008282572,0.0034723342,0.12506345],"study_design_scores_gemma":[0.000019686711,0.000043951695,0.88100594,0.000037973485,0.00007745535,0.000082557,0.0004336272,0.11331561,0.0013993527,0.00029154113,0.0032653238,0.000026987933],"about_ca_topic_score_codex":0.03619023,"about_ca_topic_score_gemma":0.07524295,"teacher_disagreement_score":0.03619023,"about_ca_system_score_codex":0.00038959697,"about_ca_system_score_gemma":0.0005389358,"threshold_uncertainty_score":0.0719592},"labels":[],"label_agreement":null},{"id":"W4408712625","doi":"10.1038/s41597-025-04456-4","title":"Banana Leaves Imagery Dataset","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Smart Agriculture and AI","field":"Agricultural and Biological Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"International Development Research Centre; Styrelsen för Internationellt Utvecklingssamarbete","keywords":"Tanzania; Fusarium wilt; Agriculture; Data collection; Geography; Race (biology); Computer science; Remote sensing; Cartography; Horticulture; Biology; Botany; Environmental planning; Statistics; Mathematics","score_opus":0.041991631899976914,"score_gpt":0.26667905758240595,"score_spread":0.22468742568242905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408712625","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04478434,0.0039049918,0.0033658398,0.0008130293,0.00056871126,0.00057052134,0.9197051,0.008750275,0.017537128],"genre_scores_gemma":[0.014668341,0.00035080916,0.005552535,0.00017392477,0.000034696324,0.00017859637,0.9766733,0.00012418242,0.0022437212],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994247,0.000064815344,0.000041675525,0.00019791645,0.00016821637,0.00010262327],"domain_scores_gemma":[0.99960965,0.0000629344,0.000029946586,0.00010494396,0.00014447146,0.000048121547],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003870219,0.0019684695,0.0008469631,0.0026645863,0.0010856449,0.0011027424,0.0019778297,0.0015058116,0.009874954],"category_scores_gemma":[0.0008985278,0.000416724,0.0010863154,0.0024401415,0.00041743394,0.0009591628,0.0011384826,0.0013875583,0.013024964],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00084054697,0.00089325826,0.007392332,0.0026732085,0.00025249843,0.00076295465,0.00023143462,0.0031862909,0.022421762,0.0011134152,0.87952924,0.08070311],"study_design_scores_gemma":[0.00035545602,0.00028077752,0.060169242,0.00051004405,0.00017697545,0.0020379971,0.0010405964,0.024471352,0.0149522405,0.0016757513,0.89413697,0.00019264258],"about_ca_topic_score_codex":0.024441639,"about_ca_topic_score_gemma":0.07230342,"teacher_disagreement_score":0.024441639,"about_ca_system_score_codex":0.0009361339,"about_ca_system_score_gemma":0.0007684789,"threshold_uncertainty_score":0.048598707},"labels":[],"label_agreement":null},{"id":"W4408820614","doi":"10.1038/s41597-025-04771-w","title":"The Eye Movement Database of Passage Reading in Vertically Written Traditional Mongolian","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Social Sciences and Humanities Research Council of Canada; Canada Research Chairs; Government of Canada","keywords":"Reading (process); Movement (music); Computer science; Database; Information retrieval; Linguistics; Art","score_opus":0.03549533225539785,"score_gpt":0.3043132777584626,"score_spread":0.26881794550306476,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408820614","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9062384,0.0007663311,0.0034119708,0.00008133133,0.00003136993,0.00031480592,0.0805408,0.0004931781,0.008121852],"genre_scores_gemma":[0.81433976,0.00035367245,0.009814444,0.00005240411,0.0000276253,0.00091945723,0.16802739,0.000115483206,0.0063497513],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996767,0.000063247295,0.000055207533,0.00011023132,0.000060467948,0.000034263972],"domain_scores_gemma":[0.99822384,0.00050962373,0.00024960912,0.00033152205,0.0005765631,0.00010883441],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041743633,0.0002834982,0.0002819061,0.0022871972,0.00041806753,0.00044544917,0.00030341538,0.0003677359,0.0038027633],"category_scores_gemma":[0.0021490904,0.00010826181,0.00013537664,0.002062087,0.00025880884,0.00037881243,0.000602921,0.00016899435,0.0012410105],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020491926,0.0005919591,0.25592044,0.0051713493,0.00025777204,0.003108496,0.023810823,0.0019393808,0.11948972,0.0022732897,0.049244054,0.53614354],"study_design_scores_gemma":[0.000041345585,0.00019029001,0.941327,0.000093944975,0.00004481695,0.0010257185,0.0035977038,0.0013627307,0.0074715298,0.00033276033,0.044466108,0.000046112415],"about_ca_topic_score_codex":0.01562839,"about_ca_topic_score_gemma":0.025335232,"teacher_disagreement_score":0.01562839,"about_ca_system_score_codex":0.0004223812,"about_ca_system_score_gemma":0.0006535131,"threshold_uncertainty_score":0.031074882},"labels":[],"label_agreement":null},{"id":"W4408963841","doi":"10.1038/s41597-025-04863-7","title":"Multimodal precision MRI of the individual human brain at ultra-high fields","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced MRI Techniques and Applications","field":"Medicine","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; McGill University; Montreal Neurological Institute and Hospital","funders":"Wellcome Trust","keywords":"Human brain; Computer science; Artificial intelligence; Computational biology; Neuroscience; Psychology; Biology","score_opus":0.05193038708683744,"score_gpt":0.3736884220903615,"score_spread":0.32175803500352407,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408963841","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.40845364,0.005227854,0.32331726,0.0023483841,0.0003230585,0.0004923253,0.23427102,0.007723943,0.017842535],"genre_scores_gemma":[0.62280756,0.004107346,0.15495992,0.0009820237,0.0002955546,0.0009828416,0.20840523,0.0012934756,0.006165991],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9996213,0.00007466438,0.000033293654,0.00014478664,0.000091498936,0.000034462264],"domain_scores_gemma":[0.99921775,0.00014848467,0.0001440305,0.00031419343,0.00012902824,0.000046545905],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006999923,0.0005433257,0.0005356613,0.0010558073,0.0004120812,0.00086216157,0.0006683159,0.00084611657,0.0036347033],"category_scores_gemma":[0.0024190007,0.00038328502,0.000511774,0.0014313292,0.0006943174,0.00073196675,0.001292703,0.00061997754,0.0020419047],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016651897,0.0003030681,0.05865521,0.003589927,0.0013633118,0.0024988735,0.0014947982,0.04748189,0.25847802,0.025004983,0.23705444,0.36241028],"study_design_scores_gemma":[0.00027598688,0.0006942763,0.28450078,0.00080386444,0.00095360185,0.0133642,0.00079557317,0.04516998,0.09948798,0.113310196,0.44021544,0.00042814566],"about_ca_topic_score_codex":0.0030742236,"about_ca_topic_score_gemma":0.0075491415,"teacher_disagreement_score":0.0036347033,"about_ca_system_score_codex":0.00033833308,"about_ca_system_score_gemma":0.00056934863,"threshold_uncertainty_score":0.012159228},"labels":[],"label_agreement":null},{"id":"W4409000241","doi":"10.1038/s41597-025-04869-1","title":"UrBAN: Urban Beehive Acoustics and PheNotyping Dataset","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Insect and Pesticide Research","field":"Agricultural and Biological Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval; Institut National de la Recherche Scientifique; Université du Québec à Montréal","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Beehive; Apiary; Varroa; Population; Proxy (statistics); Geography; Drone; Swarming (honey bee); Mite; Beekeeping; Biology; Ecology; Computer science; Environmental health; Machine learning; Medicine","score_opus":0.061818336176629074,"score_gpt":0.3063153138626512,"score_spread":0.24449697768602213,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409000241","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.050431516,0.00054852583,0.003242495,0.00033900622,0.0001267475,0.00021554844,0.93984395,0.002053668,0.003198554],"genre_scores_gemma":[0.042482127,0.00017307003,0.0039723474,0.00010514442,0.000029282877,0.00028001686,0.9511627,0.00006237828,0.0017330036],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99953294,0.000061467705,0.000029690242,0.0001227127,0.00012840559,0.00012473753],"domain_scores_gemma":[0.9994018,0.000076062526,0.00004053914,0.00013304675,0.00024563962,0.000102863036],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00037186142,0.0012441203,0.0007520918,0.0015135277,0.00070214196,0.0006531171,0.0015390979,0.0009357731,0.0042875726],"category_scores_gemma":[0.0011253807,0.00023700148,0.0006373768,0.0025992722,0.0004029532,0.00032160085,0.001065901,0.000712882,0.003934239],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007366106,0.0006140823,0.09036613,0.0015366259,0.00048826163,0.0010982928,0.0006178492,0.012566766,0.011210366,0.0011021986,0.78568745,0.09397538],"study_design_scores_gemma":[0.00032225662,0.00028566807,0.4893072,0.0002709163,0.00018076769,0.0008532695,0.001989764,0.034055866,0.0058080233,0.0011486815,0.46553627,0.00024135814],"about_ca_topic_score_codex":0.2969509,"about_ca_topic_score_gemma":0.5273774,"teacher_disagreement_score":0.2969509,"about_ca_system_score_codex":0.0018436287,"about_ca_system_score_gemma":0.001843026,"threshold_uncertainty_score":0.5904449},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4409031891","doi":"10.1038/s41597-024-04207-x","title":"Global Roadkill Data: a dataset on terrestrial vertebrate mortality caused by collision with vehicles","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Wildlife-Road Interactions and Conservation","field":"Environmental Science","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Laurentian University; Université Laval; Centre de Géomatique du Québec; Parks Canada; Lake Simcoe Region Conservation Authority; Concordia University; Memorial University of Newfoundland; Canadian Wildlife Federation; Environment and Climate Change Canada; Mount Allison University","funders":"H2020 European Research Council; Agencia Estatal de Investigación; Instituto Chico Mendes de Conservação da Biodiversidade; Tanzania Wildlife Research Institute; HORIZON EUROPE Framework Programme; U.S. Fish and Wildlife Service; Fondation Segré; Consejo Mexiquense de Ciencia y Tecnología; Unitatea Executiva pentru Finantarea Invatamantului Superior, a Cercetarii, Dezvoltarii si Inovarii; University of Venda; Miljødirektoratet; Environment and Climate Change Canada; Academia Româna; Fundação de Apoio ao Desenvolvimento do Ensino, Ciência e Tecnologia do Estado de Mato Grosso do Sul; Consejo Nacional de Ciencia y Tecnología; Boise State University; Ministerio de Ciencia e Innovación; Fundação de Amparo à Pesquisa do Estado de Goiás; Ministarstvo Prosvete, Nauke i Tehnološkog Razvoja; Ministère des Transports; Lietuvos Mokslo Taryba; Fundação de Amparo à Pesquisa do Estado de São Paulo; European Commission; Sächsisches Staatsministerium für Wissenschaft und Kunst; Narodowym Centrum Nauki; National Geographic Society; King Mongkut's University of Technology Thonburi; Akademie Věd České Republiky; Ministry of Environment; National Research Foundation; U.S. Department of Agriculture; Ministério da Ciência, Tecnologia e Inovação; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior; Environmental Restoration and Conservation Agency; Royal Zoological Society of Scotland; National Natural Science Foundation of China; Idaho Transportation Department; Fundação de Amparo à Pesquisa e Inovação do Estado de Santa Catarina; Massachusetts Department of Fish and Game; Generalitat de Catalunya; Bundesministerium für Verkehr und Digitale Infrastruktur; Comunidad de Madrid; Ministry of Natural Resources; Biologické Centrum, Akademie Věd České Republiky; Conselho Nacional de Desenvolvimento Científico e Tecnológico; British Herpetological Society; Deutsche Bundesstiftung Umwelt; Junta de Andalucía; Fundação para a Ciência e a Tecnologia; Idaho Department of Fish and Game; McLean Foundation; Fundación Charles Darwin; Natural Sciences and Engineering Research Council of Canada; Bundesministerium für Bildung und Forschung; Panthera; Idea Wild; Mekelle University","keywords":"Collision; Vertebrate; Environmental science; Ecology; Biology; Geography; Computer science; Computer security","score_opus":0.04990942759797331,"score_gpt":0.32237358595060733,"score_spread":0.27246415835263404,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409031891","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006747178,0.00015443374,0.00029556404,0.000059983948,0.000021639968,0.000023799581,0.9913857,0.00018054378,0.0011311832],"genre_scores_gemma":[0.009501348,0.000108286535,0.00073012215,0.000025948078,0.000011434049,0.000086167805,0.9891144,0.00003260748,0.00038969138],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993606,0.00009074062,0.00013362912,0.00016012098,0.00016739809,0.00008752253],"domain_scores_gemma":[0.9975706,0.00044564056,0.00080701505,0.00035276395,0.00061566615,0.0002084273],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006130864,0.000814047,0.0006622454,0.0033239853,0.00035191435,0.00076604786,0.0011655444,0.00092654023,0.0071428535],"category_scores_gemma":[0.0031401122,0.0003555691,0.0007140505,0.0057265665,0.0002609939,0.00065499585,0.0012279418,0.00086758746,0.0070637213],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030286997,0.00011309705,0.13358079,0.0027604704,0.00039443778,0.00038719267,0.00049455586,0.005508032,0.0015115918,0.0019244851,0.8294363,0.02358609],"study_design_scores_gemma":[0.00020435214,0.000098365796,0.36703217,0.00069475383,0.00013856648,0.00044492836,0.0007838341,0.0037544502,0.0013845572,0.0011709842,0.6241883,0.00010481595],"about_ca_topic_score_codex":0.03244404,"about_ca_topic_score_gemma":0.04083111,"teacher_disagreement_score":0.03244404,"about_ca_system_score_codex":0.0005936437,"about_ca_system_score_gemma":0.0010968179,"threshold_uncertainty_score":0.064510345},"labels":[],"label_agreement":null},{"id":"W4409045544","doi":"10.1038/s41597-025-04678-6","title":"An chromosome-level haplotype-resolved genome assembly and annotation of pitaya (Selenicereus polyrhizus)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Botanical Research and Applications","field":"Agricultural and Biological Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"National Natural Science Foundation of China","keywords":"Contig; Haplotype; Biology; Genome; Sequence assembly; Gene; Genetics; Reference genome; Chromosome; Coding region; Gene prediction; Computational biology; Genotype; Transcriptome","score_opus":0.08317464501943966,"score_gpt":0.32675391724953634,"score_spread":0.24357927223009668,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409045544","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.85961956,0.0016534262,0.09311378,0.000278855,0.00010665679,0.00026475312,0.040441096,0.0015502638,0.0029715828],"genre_scores_gemma":[0.6357509,0.0014435989,0.19195932,0.000110505316,0.000047037778,0.00027836606,0.16563173,0.0006368248,0.00414168],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99988914,0.000010404535,0.000009690124,0.00005400157,0.00002428229,0.000012503529],"domain_scores_gemma":[0.99978286,0.000034381363,0.00004349288,0.00003442139,0.00006993186,0.000034943383],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019706493,0.00038957933,0.00040201738,0.0007350456,0.00042959774,0.00042899922,0.0002634786,0.0003346491,0.0012565306],"category_scores_gemma":[0.0004716979,0.00026450024,0.0006039828,0.0008210831,0.00011584596,0.0003331571,0.0003956293,0.00055244967,0.00074397045],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004342656,0.00008427458,0.011969533,0.0006306162,0.00009943363,0.00060271344,0.0005316975,0.0025498378,0.9340838,0.00047488036,0.0017717393,0.04676726],"study_design_scores_gemma":[0.00023919423,0.001119824,0.5624602,0.00028439602,0.0010369903,0.002941289,0.0009840825,0.05514176,0.26036102,0.0015692157,0.11365439,0.00020760494],"about_ca_topic_score_codex":0.0049685873,"about_ca_topic_score_gemma":0.008523445,"teacher_disagreement_score":0.0049685873,"about_ca_system_score_codex":0.00028545124,"about_ca_system_score_gemma":0.0006790998,"threshold_uncertainty_score":0.009879351},"labels":[],"label_agreement":null},{"id":"W4409158760","doi":"10.1038/s41597-025-04852-w","title":"StoichLife: A Global Dataset of Plant and Animal Elemental Content","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Isotope Analysis in Ecology","field":"Environmental Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Memorial University of Newfoundland","funders":"Australian Research Council; Natural Sciences and Engineering Research Council of Canada; National Science Foundation; Royal Society; Department of Education and Training; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Vlaamse regering; Fonds Wetenschappelijk Onderzoek; Ministério da Ciência, Tecnologia e Inovação; Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Deutsche Forschungsgemeinschaft; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Russian Science Foundation","keywords":"Content (measure theory); Information retrieval; Environmental science; Computer science; Mathematics","score_opus":0.04010059842137004,"score_gpt":0.2917755226563082,"score_spread":0.25167492423493815,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409158760","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009544913,0.0005646971,0.0012702898,0.000075130374,0.00002281502,0.00001844965,0.98534167,0.00085260416,0.0023094614],"genre_scores_gemma":[0.011178029,0.0002895476,0.004201991,0.000064794694,0.000013561593,0.00009202388,0.9833145,0.00019626967,0.0006492865],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993212,0.00007536929,0.00009179595,0.00029330308,0.00014868582,0.00006954749],"domain_scores_gemma":[0.9984102,0.00028217488,0.00041667002,0.00035743698,0.00037895335,0.00015445548],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006196517,0.0010015158,0.0010298437,0.0046243113,0.0005832929,0.001437932,0.0014309481,0.0011514489,0.009463879],"category_scores_gemma":[0.0029471323,0.00046086067,0.000973599,0.0073577557,0.000378615,0.0014313926,0.001943272,0.0009898419,0.00973182],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007764422,0.00018592666,0.16330823,0.007653697,0.0011457183,0.00060294557,0.0013200853,0.0057915524,0.016258871,0.008729263,0.7112622,0.08296505],"study_design_scores_gemma":[0.000104211395,0.000041658786,0.105127096,0.00038026067,0.00012947936,0.00031943506,0.0002846462,0.0016836906,0.0020598674,0.0033485335,0.88643545,0.0000855839],"about_ca_topic_score_codex":0.010846749,"about_ca_topic_score_gemma":0.022922771,"teacher_disagreement_score":0.010846749,"about_ca_system_score_codex":0.00069085805,"about_ca_system_score_gemma":0.001062021,"threshold_uncertainty_score":0.03165984},"labels":[],"label_agreement":null},{"id":"W4409185689","doi":"10.1038/s41597-025-04868-2","title":"An open-access database of nature-based carbon offset project boundaries","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Land Use and Ecosystem Services","field":"Environmental Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carbon Engineering (Canada)","funders":"","keywords":"Geospatial analysis; Reforestation; Digitization; Database; Deforestation (computer science); Computer science; Offset (computer science); Carbon footprint; Afforestation; Remote sensing; Georeference; Metadata; Environmental science; Environmental resource management; Geography; Physical geography; Greenhouse gas; Forestry; World Wide Web; Geology; Telecommunications","score_opus":0.04432664285794074,"score_gpt":0.36030916925412615,"score_spread":0.3159825263961854,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409185689","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018067973,0.00033329267,0.009536492,0.00014725869,0.00007051629,0.00030305708,0.94668245,0.0031193376,0.02173967],"genre_scores_gemma":[0.026812894,0.0003136038,0.02177901,0.00006021842,0.00003110593,0.0005912049,0.9465994,0.00046101655,0.0033515897],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99753237,0.00030086172,0.00075286586,0.0005040634,0.00075835834,0.00015148186],"domain_scores_gemma":[0.9815579,0.0041571916,0.004520077,0.003808378,0.0046641314,0.0012922664],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0030944394,0.0006058706,0.0009010797,0.0129916,0.0007627666,0.0028180706,0.001487149,0.0007828525,0.026620334],"category_scores_gemma":[0.015571312,0.0005175548,0.0003157188,0.023351107,0.00029765154,0.0023889428,0.0022810888,0.00074663426,0.01621128],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008460957,0.0004230813,0.09930348,0.0025943695,0.00020396235,0.00056352396,0.0016474035,0.004719336,0.005315097,0.02003353,0.60431165,0.2600385],"study_design_scores_gemma":[0.00013914375,0.000079477795,0.095329784,0.00057488313,0.00007381523,0.0003577045,0.0009549681,0.0032483393,0.0035532909,0.0046691447,0.8909135,0.00010593325],"about_ca_topic_score_codex":0.0085971495,"about_ca_topic_score_gemma":0.009793882,"teacher_disagreement_score":0.99851286,"about_ca_system_score_codex":0.0009839099,"about_ca_system_score_gemma":0.004047624,"threshold_uncertainty_score":0.08905387},"labels":[],"label_agreement":null},{"id":"W4409246424","doi":"10.1038/s41597-025-04921-0","title":"Whole Genome Sequence of the gut commensal protist Tritrichomonas musculus isolated from laboratory mice","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Hospital for Sick Children","funders":"NHLBI Division of Intramural Research; Canadian Institutes of Health Research; U.S. Department of Health and Human Services; National Institutes of Health; Government of Canada; National Institute of Allergy and Infectious Diseases; Division of Intramural Research, National Institute of Allergy and Infectious Diseases","keywords":"Biology; Genome; Gene; Axenic; Genetics; Giardia; Model organism; Protist; Entamoeba; Giardia lamblia; Whole genome sequencing; Computational biology; Entamoeba histolytica; Bacteria","score_opus":0.02464295819662421,"score_gpt":0.2698881851833915,"score_spread":0.24524522698676732,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409246424","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.43225908,0.0029670214,0.0138725955,0.00031602682,0.00012031997,0.00018419976,0.54499435,0.00091746286,0.0043688817],"genre_scores_gemma":[0.13147978,0.002027299,0.02648179,0.00019058438,0.000030933057,0.00020656927,0.8358652,0.00039688253,0.0033209135],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99987316,0.000013235298,0.000014128001,0.00004724566,0.00003335754,0.000018732631],"domain_scores_gemma":[0.99977475,0.000041051448,0.00006269031,0.00002243664,0.000057780348,0.000041287727],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002361268,0.000608943,0.00045203825,0.001330889,0.0005794888,0.0004225105,0.0003172634,0.00055322296,0.0028721853],"category_scores_gemma":[0.00038968294,0.00021956743,0.00065466017,0.0021090172,0.00017432396,0.00020045812,0.00023034064,0.00066396844,0.002267131],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008700307,0.00005561979,0.003472266,0.0008602715,0.00007266922,0.0003104703,0.00018551656,0.0007299299,0.97851974,0.00025402853,0.0034317311,0.011237683],"study_design_scores_gemma":[0.00030134682,0.0012073038,0.5176538,0.0005704777,0.0011866689,0.0021484818,0.00058834185,0.005212823,0.24124566,0.0011401623,0.22856173,0.00018317062],"about_ca_topic_score_codex":0.0059834714,"about_ca_topic_score_gemma":0.007142941,"teacher_disagreement_score":0.0059834714,"about_ca_system_score_codex":0.0003553236,"about_ca_system_score_gemma":0.0008486167,"threshold_uncertainty_score":0.011897266},"labels":[],"label_agreement":null},{"id":"W4409427820","doi":"10.1038/s41597-025-04970-5","title":"Establishing a comprehensive host-parasite stable isotope database to unravel trophic relationships","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Isotope Analysis in Ecology","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; University of Alberta","funders":"Agencia Estatal de Investigación; Marsden Fund; HORIZON EUROPE Framework Programme; Ministry of Business, Innovation and Employment; Royal Society Te Apārangi; European Commission","keywords":"Host (biology); Trophic level; Stable isotope ratio; Parasite hosting; Database; Biology; Ecology; Computer science; World Wide Web","score_opus":0.0569834263259266,"score_gpt":0.30193306094343164,"score_spread":0.24494963461750505,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409427820","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13922074,0.0060736015,0.060847845,0.00068212155,0.00036118226,0.0003291697,0.77145386,0.0031125867,0.01791891],"genre_scores_gemma":[0.09704271,0.0021380272,0.10819989,0.00036600593,0.000110859,0.0005158007,0.7885974,0.00064160366,0.0023876738],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980761,0.00027466583,0.0004083519,0.0006192698,0.00047850885,0.00014304469],"domain_scores_gemma":[0.9954691,0.0008951312,0.0009038423,0.0011190891,0.0010815616,0.0005313658],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029413595,0.0005833602,0.0013372806,0.009909154,0.0011880894,0.002120519,0.0016418385,0.001063403,0.0055271075],"category_scores_gemma":[0.00628933,0.000545476,0.00068678707,0.010334184,0.0005493779,0.002191831,0.0031030138,0.0012088951,0.0053952835],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009558793,0.00039231635,0.3508339,0.0074310917,0.0009954417,0.0015218983,0.0027413247,0.0058557936,0.10034129,0.014197861,0.17650644,0.33822685],"study_design_scores_gemma":[0.000064470536,0.00012825472,0.16470008,0.0007409231,0.00029726583,0.001302006,0.0010508359,0.00545264,0.01112748,0.00915384,0.805772,0.00021020765],"about_ca_topic_score_codex":0.0048489035,"about_ca_topic_score_gemma":0.009745989,"teacher_disagreement_score":0.009909154,"about_ca_system_score_codex":0.00064039783,"about_ca_system_score_gemma":0.0022085959,"threshold_uncertainty_score":0.018490016},"labels":[],"label_agreement":null},{"id":"W4409566488","doi":"10.1038/s41597-025-04907-y","title":"ROBIN: Reference observatory of basins for international hydrological climate change detection","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"U.S. Geological Survey; Natural Environment Research Council; Research Councils UK","keywords":"Streamflow; Climate change; Water cycle; Coupled model intercomparison project; Environmental science; Drainage basin; Climatology; Climate model; Global warming; Adaptation (eye); Global change; Scale (ratio); Identification (biology); Environmental resource management; Water resources; Hydrological modelling; Physical geography; Geography; Ecology; Geology; Cartography","score_opus":0.10340024084581281,"score_gpt":0.3016252393294802,"score_spread":0.1982249984836674,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409566488","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029166888,0.00027668066,0.018156897,0.0008732129,0.00027020613,0.0003575924,0.90621233,0.014660449,0.030025769],"genre_scores_gemma":[0.08273352,0.00022540167,0.04823403,0.00032614302,0.00012047808,0.0008389852,0.85553414,0.0038800256,0.00810732],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9981572,0.00034893368,0.00020310831,0.00047439005,0.00060001586,0.0002162483],"domain_scores_gemma":[0.99324703,0.00071158377,0.0012986498,0.0019466544,0.0018078788,0.0009881515],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030606862,0.00068596617,0.00070721185,0.0021016062,0.0006168296,0.0016217575,0.0016242671,0.0006216588,0.02256397],"category_scores_gemma":[0.008677634,0.00040172375,0.00030332716,0.0051697074,0.00043072482,0.0019094452,0.0022495997,0.0010424753,0.009736449],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003357366,0.000113817914,0.050431672,0.00038835048,0.000072691066,0.00010614875,0.00064040226,0.006324665,0.0038484468,0.0077926773,0.8557926,0.074152775],"study_design_scores_gemma":[0.0001850927,0.000051552674,0.07879303,0.00010692389,0.00002813516,0.00005641612,0.00027400278,0.013717672,0.002814988,0.0041892473,0.89970297,0.00007995783],"about_ca_topic_score_codex":0.056688927,"about_ca_topic_score_gemma":0.05261157,"teacher_disagreement_score":0.056688927,"about_ca_system_score_codex":0.0011679789,"about_ca_system_score_gemma":0.004509018,"threshold_uncertainty_score":0.11271793},"labels":[],"label_agreement":null},{"id":"W4409669866","doi":"10.1038/s41597-025-05005-9","title":"Author Correction: An open-access database of nature-based carbon offset project boundaries","year":2025,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Environmental Conservation and Management","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carbon Engineering (Canada)","funders":"","keywords":"Offset (computer science); Computer science; Database; World Wide Web; Open source; Information retrieval; Programming language; Software","score_opus":0.0788561484145464,"score_gpt":0.3757769037891152,"score_spread":0.2969207553745688,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409669866","genre_codex":"editorial","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003281454,0.0017403016,0.017670814,0.06778463,0.59765005,0.00020832756,0.18153204,0.007108534,0.12302385],"genre_scores_gemma":[0.040323183,0.0053688777,0.043220673,0.020963402,0.03366072,0.0005720427,0.18058835,0.01632353,0.6589792],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99578816,0.00042367444,0.0007148154,0.00048988493,0.0023772637,0.00020614776],"domain_scores_gemma":[0.9618132,0.009278786,0.0021216294,0.0038353645,0.021961473,0.0009896002],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003361513,0.0009435028,0.00085319235,0.004845444,0.0022342752,0.0051599853,0.001942121,0.002432714,0.17485696],"category_scores_gemma":[0.059161037,0.0007353094,0.0005856831,0.00786089,0.0009835061,0.0030790882,0.002755744,0.0045387354,0.082146995],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025745478,0.000004727437,0.00022734815,0.000088801615,0.0000049992964,0.00011405267,0.000056051413,0.00010690272,0.000078708836,0.002506331,0.98500514,0.0117813125],"study_design_scores_gemma":[0.000013470916,0.0000040982186,0.00050225866,0.00014714034,0.000007761752,0.00012703383,0.000073441784,0.00013647911,0.00025079903,0.0010191544,0.9977022,0.000016241838],"about_ca_topic_score_codex":0.020702468,"about_ca_topic_score_gemma":0.023196565,"teacher_disagreement_score":0.17485696,"about_ca_system_score_codex":0.0024488682,"about_ca_system_score_gemma":0.0072485963,"threshold_uncertainty_score":0.5849548},"labels":[],"label_agreement":null},{"id":"W4409766480","doi":"10.1038/s41597-025-05001-z","title":"HarvestStat Africa – Harmonized Subnational Crop Statistics for Sub-Saharan Africa","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate change impacts on agriculture","field":"Agricultural and Biological Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"U.S. Geological Survey; Bundesministerium für Bildung und Forschung; Goddard Space Flight Center; Wyoming-Montana Water Science Center; Volkswagen Foundation; Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Consortium of International Agricultural Research Centers; U.S. Department of the Interior; Canada Research Chairs; United States Agency for International Development; Deutsche Forschungsgemeinschaft; National Aeronautics and Space Administration","keywords":"Agriculture; Sorghum; Famine; Agricultural productivity; Yield (engineering); Agricultural economics; Scarcity; Geography; Crop; Crop yield; Agroforestry; Environmental science; Economics; Ecology; Forestry; Biology","score_opus":0.1283449131331734,"score_gpt":0.30388792199483927,"score_spread":0.17554300886166588,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409766480","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01181136,0.00034325125,0.0014824693,0.00017702076,0.000063544554,0.00009150443,0.98174804,0.0004200578,0.0038627095],"genre_scores_gemma":[0.025933107,0.0007233677,0.006315086,0.0001042268,0.000042242682,0.0006146813,0.9638596,0.00029233246,0.0021153605],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99924064,0.00015430278,0.00017517676,0.00013445159,0.00017931966,0.000116151714],"domain_scores_gemma":[0.9964799,0.00058454205,0.0011779326,0.0004667301,0.0010953634,0.0001954976],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013001659,0.0008950914,0.00055198075,0.0047980146,0.00028598378,0.0010800662,0.00088245206,0.00034896802,0.00927075],"category_scores_gemma":[0.0060905786,0.00039216812,0.0004299638,0.012889288,0.00018059691,0.0012191144,0.0011016672,0.00057072734,0.0043174727],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00094808504,0.00012065866,0.10941074,0.00394215,0.00042499037,0.00060352235,0.0010350995,0.006137937,0.0032513048,0.009890896,0.7518494,0.11238522],"study_design_scores_gemma":[0.00015308135,0.00003392595,0.14777648,0.0005714155,0.00006362584,0.00017124118,0.00059969816,0.0016540309,0.0015768714,0.0018886563,0.8454561,0.00005483614],"about_ca_topic_score_codex":0.024431402,"about_ca_topic_score_gemma":0.014631976,"teacher_disagreement_score":0.024431402,"about_ca_system_score_codex":0.0007326754,"about_ca_system_score_gemma":0.002156095,"threshold_uncertainty_score":0.04857838},"labels":[],"label_agreement":null},{"id":"W4409826571","doi":"10.1038/s41597-025-04898-w","title":"Airborne imaging spectroscopy surveys of Arctic and boreal Alaska and northwestern Canada 2017–2023","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate change and permafrost","field":"Earth and Planetary Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Canadian Polar Commission; Government of Northwest Territories; University of British Columbia","funders":"Jet Propulsion Laboratory; National Aeronautics and Space Administration","keywords":"Tundra; Permafrost; Taiga; Imaging spectrometer; Environmental science; Remote sensing; Arctic; Transect; Climate change; Physical geography; Boreal; Ecotone; Oceanography; Geography; Geology; Ecology; Forestry; Spectrometer; Habitat","score_opus":0.03355551431714082,"score_gpt":0.2582392655103902,"score_spread":0.2246837511932494,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409826571","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8410331,0.0024213735,0.001561603,0.0008348315,0.000077298464,0.0001000954,0.12228911,0.0003383924,0.031344287],"genre_scores_gemma":[0.8784592,0.0020451117,0.0067060087,0.00029296183,0.00004657441,0.0000905787,0.102125145,0.00006437163,0.010170086],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997403,0.00000968881,0.000012386931,0.0000395791,0.00014202074,0.000055973742],"domain_scores_gemma":[0.9988146,0.000031464835,0.00008686278,0.000030180701,0.0009155393,0.00012133732],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00030379574,0.00029664524,0.00012622049,0.0022478912,0.0009768412,0.0005965346,0.00034587915,0.00019085521,0.001514022],"category_scores_gemma":[0.00042834686,0.000098712626,0.00018392483,0.0034367905,0.00021596384,0.00024191616,0.0004346015,0.00023002611,0.0004115002],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014426876,0.00011257371,0.8246532,0.0001773467,0.00015233505,0.00029918997,0.0015576096,0.004066475,0.00824449,0.00081953587,0.04986962,0.10990345],"study_design_scores_gemma":[0.0000075222415,0.000016106447,0.9585163,0.00008155949,0.00002565226,0.00006514084,0.0023695137,0.0017817783,0.0012467694,0.00011703499,0.03574849,0.000024003843],"about_ca_topic_score_codex":0.9559981,"about_ca_topic_score_gemma":0.97956467,"teacher_disagreement_score":0.044001877,"about_ca_system_score_codex":0.0032040896,"about_ca_system_score_gemma":0.0073592914,"threshold_uncertainty_score":0.08852196},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4410037531","doi":"10.1038/s41597-025-05069-7","title":"Multimodal sensor dataset for monitoring older adults post lower limb fractures in community settings","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hip and Femur Fractures","field":"Medicine","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University Health Network; University of Toronto; Toronto Rehabilitation Institute","funders":"","keywords":"Physical medicine and rehabilitation; Lower limb; Computer science; Medicine; Surgery","score_opus":0.025410983150132807,"score_gpt":0.35694932403285506,"score_spread":0.33153834088272227,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410037531","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21778503,0.0014436437,0.00811052,0.0007762152,0.00030823544,0.00059102895,0.7649946,0.0015542762,0.0044363537],"genre_scores_gemma":[0.1903951,0.00058563444,0.009428362,0.0003013205,0.00011520668,0.0010308528,0.79575497,0.000054240307,0.002334302],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99961686,0.0000545883,0.000060999137,0.000100490804,0.00011335278,0.000053721185],"domain_scores_gemma":[0.9995623,0.000077072116,0.00006700772,0.00007235437,0.00016471352,0.000056634206],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003930381,0.0007793921,0.0006844952,0.0009610709,0.0003250216,0.00032439505,0.00089070905,0.0008981545,0.0019942864],"category_scores_gemma":[0.0012766072,0.00013350732,0.0005665103,0.0009783113,0.0001802085,0.000323002,0.0008784165,0.0005031106,0.0015565887],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018047494,0.0018068237,0.23449534,0.004025775,0.00083775865,0.0022222043,0.0007718434,0.020352649,0.015148964,0.0013816324,0.52210486,0.19504741],"study_design_scores_gemma":[0.00043034586,0.0011964807,0.71389663,0.0006396166,0.00033506993,0.0022832262,0.0018944149,0.050244525,0.009759097,0.0025918288,0.2164832,0.00024552672],"about_ca_topic_score_codex":0.016620433,"about_ca_topic_score_gemma":0.03635777,"teacher_disagreement_score":0.016620433,"about_ca_system_score_codex":0.0004654473,"about_ca_system_score_gemma":0.0005972141,"threshold_uncertainty_score":0.033047378},"labels":[],"label_agreement":null},{"id":"W4410056551","doi":"10.1038/s41597-025-05054-0","title":"A Dataset for Understanding Radiologist-Artificial Intelligence Collaboration","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Alfred P. Sloan Foundation","keywords":"Computer science; Data science; Artificial intelligence","score_opus":0.6273723659303506,"score_gpt":0.545717655039745,"score_spread":0.08165471089060561,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410056551","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.084677614,0.0028724223,0.020663317,0.002429855,0.0005113962,0.0013323354,0.8720954,0.0044573583,0.010960276],"genre_scores_gemma":[0.084402926,0.00045876732,0.037073202,0.00050355477,0.000114728035,0.0015001487,0.8739629,0.00020993066,0.0017737292],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9962291,0.0011657814,0.0005564245,0.0008645145,0.00095236185,0.00023183304],"domain_scores_gemma":[0.98729336,0.0070515675,0.0010720416,0.0021030211,0.0017993668,0.00068065396],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029970463,0.0011716612,0.00074062386,0.0034022583,0.0009062556,0.0017316386,0.0022823557,0.0026041414,0.0055611343],"category_scores_gemma":[0.01596862,0.0003512733,0.0011865066,0.0031942194,0.00063911773,0.0010170644,0.0022753347,0.00159082,0.0041808463],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001331192,0.0018476052,0.08725273,0.0043366193,0.0005688401,0.0015540778,0.0013277048,0.02192057,0.0061733974,0.007274043,0.72792035,0.13849278],"study_design_scores_gemma":[0.00075810915,0.00066246546,0.14514093,0.0010748138,0.00028259787,0.002572873,0.0020410228,0.07075663,0.00987768,0.014334942,0.7522434,0.00025455633],"about_ca_topic_score_codex":0.009683679,"about_ca_topic_score_gemma":0.023417683,"teacher_disagreement_score":0.009683679,"about_ca_system_score_codex":0.0017490825,"about_ca_system_score_gemma":0.0018807147,"threshold_uncertainty_score":0.019254625},"labels":[],"label_agreement":null},{"id":"W4410265518","doi":"10.1038/s41597-025-05090-w","title":"A database for igneous rocks of the Newfoundland Appalachians","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geological and Geochemical Analysis","field":"Earth and Planetary Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Igneous rock; Geology; Geochemistry; Database; Computer science","score_opus":0.02718396622174989,"score_gpt":0.241120765856028,"score_spread":0.2139367996342781,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410265518","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024374245,0.0008162018,0.0008887963,0.0001964891,0.000021278376,0.00006256651,0.9658441,0.0005244664,0.0072718826],"genre_scores_gemma":[0.029865548,0.0008664293,0.004592398,0.000050097497,0.000014724494,0.00016024504,0.9609731,0.000066843786,0.0034106101],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99942076,0.00003634792,0.0001376747,0.0001400034,0.00015406647,0.00011119313],"domain_scores_gemma":[0.99530643,0.00062523584,0.00091409544,0.0006098839,0.0020840035,0.00046041238],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000605655,0.0006492971,0.00051305053,0.010967834,0.0012955231,0.0017889352,0.0011488606,0.00049711607,0.010874559],"category_scores_gemma":[0.0035114633,0.00036196096,0.00037098132,0.015824027,0.0004465707,0.0012288796,0.001218339,0.00046427586,0.0033439174],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045469386,0.00011924214,0.19667147,0.0031079191,0.00028207284,0.0012944414,0.0020133746,0.0037832516,0.004094204,0.0056914757,0.61037546,0.17211238],"study_design_scores_gemma":[0.000042078987,0.000018336426,0.2465193,0.00053691503,0.00009937491,0.00029480562,0.0013351464,0.0009737897,0.0014432815,0.0005372372,0.7481395,0.000060311646],"about_ca_topic_score_codex":0.7184662,"about_ca_topic_score_gemma":0.81033796,"teacher_disagreement_score":0.28153378,"about_ca_system_score_codex":0.004748903,"about_ca_system_score_gemma":0.012035679,"threshold_uncertainty_score":0.56638354},"labels":[],"label_agreement":null},{"id":"W4410308840","doi":"10.1038/s41597-025-05087-5","title":"The SESAME Human-Earth Atlas","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric and Environmental Gas Dynamics","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; National Aeronautics and Space Administration","keywords":"Atlas (anatomy); Earth (classical element); Astrobiology; Geography; Geology; Biology; Astronomy; Physics; Paleontology","score_opus":0.016197631690290763,"score_gpt":0.2560904796216868,"score_spread":0.23989284793139604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410308840","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021066929,0.00067512185,0.0041518942,0.0004672171,0.00020835504,0.00008166292,0.96092963,0.0023142362,0.02906518],"genre_scores_gemma":[0.012453929,0.0011722494,0.010604182,0.0001895169,0.0000668615,0.00028806305,0.9652421,0.000664815,0.009318274],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999671,0.00006786403,0.000049246995,0.000071227725,0.00009310637,0.000047606307],"domain_scores_gemma":[0.9990669,0.0001516657,0.00011000513,0.0001353698,0.00044289813,0.00009308899],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006687798,0.0009045286,0.00057777885,0.004161062,0.0003643934,0.0012122212,0.0012027205,0.00047685395,0.0569964],"category_scores_gemma":[0.0021788753,0.0002827384,0.00052400335,0.008314088,0.00020204388,0.001556502,0.0011981045,0.0007262176,0.021861805],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000725612,0.00001635403,0.0053245756,0.00082214957,0.00007134052,0.000117518815,0.0001804176,0.0011779396,0.00036549533,0.010143648,0.9507114,0.030996593],"study_design_scores_gemma":[0.000025500498,0.0000060402454,0.007014243,0.0001421517,0.000019393761,0.000083692044,0.0001058867,0.0007071771,0.000165406,0.0033700326,0.9883442,0.000016156135],"about_ca_topic_score_codex":0.041631427,"about_ca_topic_score_gemma":0.038266137,"teacher_disagreement_score":0.0569964,"about_ca_system_score_codex":0.00083371956,"about_ca_system_score_gemma":0.0016849067,"threshold_uncertainty_score":0.19067198},"labels":[],"label_agreement":null},{"id":"W4410315829","doi":"10.1038/s41597-025-05133-2","title":"A beneficial arthropod dataset for agricultural landscapes in Western Canada, and adjacent mountain ecosystems","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ecology and Vegetation Dynamics Studies","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Agriculture and Agri-Food Canada; Ducks Unlimited Canada; Calgary Laboratory Services; Simon Fraser University; University of Calgary","funders":"Alberta Innovates; Alberta Innovates Bio Solutions; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada; Alberta Canola Producers Commission; TD Friends of the Environment Foundation; Institute for Wetland and Waterfowl Research, Ducks Unlimited Canada; Alberta Biodiversity Monitoring Institute; Saskatchewan Canola Development Commission; Alberta Conservation Association","keywords":"Arthropod; Agriculture; Ecosystem; Geography; Ecology; Ecosystem services; Agroforestry; Environmental resource management; Biology; Environmental science","score_opus":0.012400308234453119,"score_gpt":0.25043991261102744,"score_spread":0.23803960437657432,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410315829","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.035719637,0.0005311604,0.00025368412,0.00009555125,0.000012749683,0.00007330558,0.9607448,0.00012681937,0.0024422894],"genre_scores_gemma":[0.042633727,0.0003642667,0.0015497896,0.00009484964,0.000007812277,0.00015634885,0.9532947,0.000028675768,0.0018697656],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995285,0.000017775417,0.000043379514,0.000107092696,0.00018720543,0.0001160212],"domain_scores_gemma":[0.9978897,0.00008944222,0.00022437872,0.00012816429,0.0013721577,0.00029618642],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00032875943,0.000641727,0.00050038425,0.0041696327,0.001885848,0.0011625998,0.0012577723,0.00041222712,0.0049043396],"category_scores_gemma":[0.0019075269,0.00029035538,0.0004977413,0.009182876,0.00035075296,0.00041547616,0.0010141989,0.0005780584,0.0019916422],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003234692,0.00013309694,0.43009573,0.0011465754,0.0004106888,0.00040732572,0.0010379553,0.0017852171,0.0019519619,0.001359947,0.5207298,0.04061819],"study_design_scores_gemma":[0.00006856487,0.000021127473,0.7806079,0.0002947739,0.00008730527,0.00019758259,0.0011753268,0.0012763413,0.0004895792,0.00027597067,0.21545108,0.000054583204],"about_ca_topic_score_codex":0.98306686,"about_ca_topic_score_gemma":0.99380493,"teacher_disagreement_score":0.016933143,"about_ca_system_score_codex":0.010704297,"about_ca_system_score_gemma":0.018541219,"threshold_uncertainty_score":0.07766545},"labels":[],"label_agreement":null},{"id":"W4410414373","doi":"10.1038/s41597-025-05052-2","title":"A collection of FAIR Dutch Freedom of Information Act documents","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Canadian Institute of Steel Construction","keywords":"Metadata; Computer science; World Wide Web; Standardization; Information retrieval; Publishing; Quality (philosophy); Process (computing); Metadata repository; Political science","score_opus":0.1328692685136867,"score_gpt":0.42210031862309055,"score_spread":0.28923105010940386,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410414373","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043773693,0.00023770038,0.0011635562,0.00026483854,0.000101741025,0.00018340125,0.9843998,0.0004049011,0.008866701],"genre_scores_gemma":[0.00531531,0.00018458356,0.0022532218,0.000054641325,0.000021764188,0.0004984001,0.9857837,0.00025624203,0.0056322254],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99171364,0.0012396637,0.0013635041,0.0010170628,0.0041375146,0.00052864477],"domain_scores_gemma":[0.9771513,0.00662474,0.0031258056,0.0052089435,0.0067623826,0.0011270202],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.004790827,0.000829575,0.00093473855,0.012307914,0.0018993159,0.0050106766,0.0013811587,0.0011919831,0.032767683],"category_scores_gemma":[0.028553205,0.00083104626,0.00069099374,0.024245428,0.0011976041,0.0027672101,0.0023464833,0.0017044093,0.027644433],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017979724,0.00014193864,0.0052599935,0.0012054639,0.00003604432,0.00021658378,0.0010879757,0.0008530551,0.0010038447,0.0062648635,0.9512949,0.032455597],"study_design_scores_gemma":[0.00004421579,0.000015692842,0.012837346,0.0003618097,0.000011580163,0.00008491986,0.000656674,0.0005364601,0.0010652464,0.0020030942,0.9823278,0.00005513423],"about_ca_topic_score_codex":0.07337004,"about_ca_topic_score_gemma":0.07836478,"teacher_disagreement_score":0.99861884,"about_ca_system_score_codex":0.0043734577,"about_ca_system_score_gemma":0.0098669985,"threshold_uncertainty_score":0.14588594},"labels":[],"label_agreement":null},{"id":"W4410464763","doi":"10.1038/s41597-025-05188-1","title":"Transcriptome data from the interdigital webs of the Chinese Soft-shell Turtle (Pelodiscus sinensis)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Turtle Biology and Conservation","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Liaoning Revitalization Talents Program","keywords":"Turtle (robot); Shell (structure); Biology; Transcriptome; Ecology; Engineering; Gene","score_opus":0.026948280065522405,"score_gpt":0.26446882480081324,"score_spread":0.23752054473529083,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410464763","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98739594,0.00072109606,0.0022998673,0.000049377766,0.000018248791,0.000015091572,0.00871831,0.000041643187,0.0007403414],"genre_scores_gemma":[0.9552428,0.00073862995,0.004756438,0.00016126,0.000017680326,0.00011244061,0.036539983,0.000059081918,0.0023716306],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.999918,0.0000042219244,0.000007754933,0.000033268745,0.000022419192,0.000014409289],"domain_scores_gemma":[0.9998882,0.000019816795,0.000025822861,0.000009897107,0.0000378686,0.0000182997],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00009162916,0.00024596255,0.00025030653,0.00048233772,0.00041028883,0.00021943453,0.00008093784,0.00015931229,0.00091044465],"category_scores_gemma":[0.00014033631,0.00009845327,0.00036412725,0.00058385805,0.00018778058,0.00017899144,0.00029385908,0.0002659753,0.00026107754],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014686852,0.000027260987,0.03062144,0.0002480666,0.00005405707,0.00019249045,0.00026150877,0.0002797559,0.9589064,0.0000935538,0.00034798423,0.008820654],"study_design_scores_gemma":[0.000017455144,0.0001551398,0.92576057,0.000031451724,0.00022717076,0.0008047791,0.00050436216,0.0020896532,0.06304067,0.00018317386,0.0071569905,0.000028610913],"about_ca_topic_score_codex":0.0010320939,"about_ca_topic_score_gemma":0.0028940977,"teacher_disagreement_score":0.0010320939,"about_ca_system_score_codex":0.00010499675,"about_ca_system_score_gemma":0.00022798787,"threshold_uncertainty_score":0.0030456781},"labels":[],"label_agreement":null},{"id":"W4410573777","doi":"10.1038/s41597-025-05184-5","title":"Global Impacts Dataset of Invasive Alien Species (GIDIAS)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Armed Forces","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Alien; Biodiversity; Invasive species; Alien species; Taxon; Ecology; Ecosystem; Ecosystem services; Introduced species; Geography; Livelihood; Global change; Environmental resource management; Invertebrate; Biology; Climate change; Environmental science; Agriculture; Political science","score_opus":0.06371622324894954,"score_gpt":0.313654097728842,"score_spread":0.24993787447989246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410573777","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001637478,0.0002429732,0.00020625685,0.000098699784,0.000019902443,0.0000222053,0.9958269,0.00011659149,0.0018291225],"genre_scores_gemma":[0.0037060636,0.00022545429,0.0006890149,0.00004993731,0.000010123402,0.00008351194,0.9948389,0.000025694715,0.0003712323],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987997,0.00017371743,0.00023953809,0.0002280652,0.00044003292,0.00011884702],"domain_scores_gemma":[0.99714166,0.0008647261,0.0006491609,0.0003607896,0.000773166,0.00021058733],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011674962,0.0009982726,0.0007661296,0.0054224594,0.00046469207,0.001174943,0.0012231265,0.0012589607,0.008191396],"category_scores_gemma":[0.0046944916,0.0003111876,0.0010243035,0.009069331,0.00033413968,0.00075773295,0.0016702684,0.0010606232,0.005231144],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028740248,0.00011148242,0.04928153,0.0061670127,0.00045161063,0.0004412492,0.00041344768,0.0065115704,0.0022655542,0.0053590895,0.903441,0.025268983],"study_design_scores_gemma":[0.0001149726,0.0000443995,0.06337613,0.0007410825,0.000112840266,0.00020343739,0.00034251658,0.0014439807,0.0009693743,0.0015837722,0.93100923,0.000058240057],"about_ca_topic_score_codex":0.025645817,"about_ca_topic_score_gemma":0.03177604,"teacher_disagreement_score":0.025645817,"about_ca_system_score_codex":0.0010423627,"about_ca_system_score_gemma":0.0018547093,"threshold_uncertainty_score":0.050993085},"labels":[],"label_agreement":null},{"id":"W4410641159","doi":"10.1038/s41597-025-04833-z","title":"Open multi-center intracranial electroencephalography dataset with task probing conscious visual perception","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"EEG and Brain-Computer Interfaces","field":"Neuroscience","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research","funders":"National Institute of Neurological Disorders and Stroke; National Institute of Mental Health; National Institute of General Medical Sciences; Templeton World Charity Foundation; Max-Planck-Gesellschaft","keywords":"Electroencephalography; Computer science; Perception; Task (project management); Stimulus (psychology); Consciousness; Cognitive psychology; Brain activity and meditation; Visual perception; Psychology; Artificial intelligence; Neuroscience","score_opus":0.04741419624711331,"score_gpt":0.3439156638838501,"score_spread":0.2965014676367368,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410641159","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.115792915,0.0012573439,0.012667065,0.00074020046,0.00038762638,0.0007310977,0.86056644,0.0039014716,0.0039557475],"genre_scores_gemma":[0.060427275,0.00022826175,0.008316865,0.00017283173,0.00009869412,0.0009018848,0.9282312,0.00016173405,0.0014613209],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994429,0.000090521295,0.00008145801,0.00015832424,0.00014241203,0.000084390085],"domain_scores_gemma":[0.9986166,0.00030820773,0.00013261983,0.00044895482,0.00035526324,0.00013836412],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0007136759,0.0012965847,0.0008985894,0.0015960193,0.00046063226,0.0008289708,0.001553531,0.0014839217,0.0034169753],"category_scores_gemma":[0.0027895335,0.00024836432,0.0009006535,0.0015067812,0.0004070371,0.00040488422,0.0013257666,0.0009651253,0.0048830556],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022101232,0.0015337154,0.05977689,0.0028268432,0.00096066797,0.0038285195,0.00059378863,0.011360097,0.024838494,0.0018048136,0.76913095,0.12113517],"study_design_scores_gemma":[0.0016953679,0.0011584448,0.3335484,0.0005191864,0.0005443515,0.006958821,0.0008970663,0.034418687,0.023466432,0.007632965,0.5887307,0.00042948348],"about_ca_topic_score_codex":0.006360384,"about_ca_topic_score_gemma":0.0144292815,"teacher_disagreement_score":0.99844646,"about_ca_system_score_codex":0.00047229565,"about_ca_system_score_gemma":0.00086061325,"threshold_uncertainty_score":0.012646735},"labels":[],"label_agreement":null},{"id":"W4410831600","doi":"10.1038/s41597-025-05098-2","title":"Monthly hydropower generation data for Western Canada to support Western-US interconnect power system studies","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Pacific Northwest National Laboratory; Laboratory Directed Research and Development; U.S. Army Corps of Engineers; Bureau of Reclamation","keywords":"Hydropower; Hydroelectricity; Streamflow; Electricity generation; Environmental science; Surface runoff; Resource (disambiguation); Water resources; Water resource management; Environmental resource management; Hydrology (agriculture); Computer science; Geography; Power (physics); Engineering; Drainage basin; Ecology; Cartography","score_opus":0.06278394105957384,"score_gpt":0.31492168670902043,"score_spread":0.2521377456494466,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410831600","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07528757,0.00017962012,0.0018203813,0.00025998446,0.000027931457,0.00013009833,0.9076857,0.0006006113,0.0140080545],"genre_scores_gemma":[0.15564555,0.0004287188,0.004886562,0.00010704119,0.000015411644,0.00014640455,0.83129895,0.00012396762,0.0073473398],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99957687,0.000022063034,0.000025674974,0.00005788925,0.00022077735,0.00009681118],"domain_scores_gemma":[0.99643975,0.00013378491,0.00018708168,0.00018637649,0.0028013801,0.0002515381],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003138234,0.000538733,0.0003090983,0.003050434,0.0010614955,0.0010576864,0.0008808425,0.00025528617,0.00603951],"category_scores_gemma":[0.0020936313,0.00021341047,0.00035921182,0.008742805,0.00025295155,0.00033738997,0.0005132765,0.00048845087,0.0013335636],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003621751,0.0002540574,0.26964605,0.0006019264,0.0003813237,0.0005044954,0.00067227153,0.08657109,0.002051944,0.008041998,0.56359386,0.067318805],"study_design_scores_gemma":[0.0001756339,0.000039793948,0.47573113,0.00028623873,0.00012666806,0.00013875075,0.0015623512,0.06366735,0.0042499746,0.0017024151,0.45217508,0.00014459355],"about_ca_topic_score_codex":0.9922116,"about_ca_topic_score_gemma":0.9957028,"teacher_disagreement_score":0.014269922,"about_ca_system_score_codex":0.014269922,"about_ca_system_score_gemma":0.030916201,"threshold_uncertainty_score":0.10353601},"labels":[],"label_agreement":null},{"id":"W4410838461","doi":"10.1038/s41597-025-05178-3","title":"Haplotype resolved chromosome-level genome assembly of the gold barb (Barbodes semifasciolatus)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"Guangdong Academy of Sciences; National Natural Science Foundation of China","keywords":"Haplotype; Biology; Chromosome; Genome; Genetics; Evolutionary biology; Gene; Allele","score_opus":0.039900493297992064,"score_gpt":0.27230831852169896,"score_spread":0.2324078252237069,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410838461","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8642441,0.0014632446,0.095270105,0.00023381009,0.00007450077,0.00019320087,0.032053415,0.0015787065,0.004888901],"genre_scores_gemma":[0.69949955,0.001379181,0.18982275,0.00014426681,0.00002595942,0.00026681344,0.10265466,0.00079062657,0.0054161362],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99988115,0.000010769505,0.000008072281,0.000056406814,0.000028804563,0.000014746022],"domain_scores_gemma":[0.999824,0.000028823166,0.00003310781,0.000027685814,0.000060803137,0.000025562178],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002068654,0.00031181067,0.00041252963,0.0008597882,0.00038074524,0.00042801318,0.0002874616,0.00036031927,0.001561634],"category_scores_gemma":[0.0004368078,0.000255983,0.00045641421,0.0008040258,0.000121915415,0.0002348114,0.00035614532,0.00050583464,0.00095889735],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003155782,0.000038697526,0.0075057903,0.00034458737,0.00007335548,0.00038184147,0.00038660932,0.0061791926,0.9544796,0.0008207695,0.0013114983,0.02816237],"study_design_scores_gemma":[0.00023636529,0.0006929032,0.5003842,0.00028830508,0.0006346975,0.0013819356,0.00072687067,0.07389953,0.33839992,0.0019497015,0.08120288,0.00020270217],"about_ca_topic_score_codex":0.010649889,"about_ca_topic_score_gemma":0.018454658,"teacher_disagreement_score":0.010649889,"about_ca_system_score_codex":0.00041533954,"about_ca_system_score_gemma":0.00065880857,"threshold_uncertainty_score":0.021175802},"labels":[],"label_agreement":null},{"id":"W4411183390","doi":"10.1038/s41597-025-05143-0","title":"SEEDNet: Covariate-free multi-country settlement-level epidemiological estimates datasets for network analysis","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Health, Environment, Cognitive Aging","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University Health Centre; Hospital for Sick Children; Public Health Ontario; SickKids Foundation; Toronto Metropolitan University; University of Toronto","funders":"Global Affairs Canada","keywords":"Covariate; Settlement (finance); Geography; Statistics; Epidemiology; Econometrics; Computer science; Mathematics; Medicine; World Wide Web","score_opus":0.12200161197404019,"score_gpt":0.3688138084758872,"score_spread":0.246812196501847,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411183390","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006503656,0.00015645748,0.05336227,0.0004110435,0.000085081665,0.00030308333,0.9326748,0.0039349,0.0025686973],"genre_scores_gemma":[0.030738607,0.00027174564,0.073807195,0.00019405085,0.000043649845,0.002768305,0.88987124,0.0007827953,0.0015224864],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99769586,0.0010485272,0.00033461433,0.00042512582,0.00039526782,0.00010070312],"domain_scores_gemma":[0.9882918,0.004962367,0.0017068235,0.0034605775,0.001139163,0.00043929578],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004487057,0.00078347384,0.00072102185,0.0036671495,0.00053047034,0.0014523178,0.0029650708,0.0014350743,0.020890262],"category_scores_gemma":[0.03118041,0.00079802086,0.0012103423,0.0070765424,0.00042735942,0.0017316806,0.0030538682,0.0013590889,0.007113244],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006550025,0.00029491432,0.052801516,0.003999183,0.00090045744,0.0004273817,0.001347856,0.044551868,0.0020687652,0.035931576,0.76038456,0.0966368],"study_design_scores_gemma":[0.0005769966,0.0001556365,0.059086118,0.0011223091,0.00023915195,0.00044284822,0.00079070445,0.048743814,0.0032706454,0.05772505,0.8276411,0.00020559148],"about_ca_topic_score_codex":0.010697441,"about_ca_topic_score_gemma":0.01799907,"teacher_disagreement_score":0.020890262,"about_ca_system_score_codex":0.0009752575,"about_ca_system_score_gemma":0.0022488977,"threshold_uncertainty_score":0.0698849},"labels":[],"label_agreement":null},{"id":"W4411277392","doi":"10.1038/s41597-025-05333-w","title":"Spatial occurrence records and distributions of tropical Asian butterflies","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"The Scarborough Hospital","funders":"U.S. Geological Survey; University of Hong Kong; National Science Foundation","keywords":"Geography; Ecology; Tropical Asia; Climatology; Biology; Geology","score_opus":0.040857089169196587,"score_gpt":0.2903349810107173,"score_spread":0.24947789184152072,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411277392","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.64420795,0.000593993,0.0004178077,0.000114893075,0.00000913053,0.000020926536,0.3508457,0.00012959543,0.0036600064],"genre_scores_gemma":[0.6331425,0.00084558135,0.0018906628,0.000059398633,0.000018313514,0.00012996528,0.36262953,0.000044462155,0.0012395899],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9997578,0.00003246066,0.000044414606,0.00006968618,0.000043559743,0.00005207461],"domain_scores_gemma":[0.99861336,0.00020452728,0.00055053417,0.00020609917,0.0003054515,0.00011997033],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00044032556,0.0002600969,0.00023145216,0.0042910473,0.00024621055,0.00042341003,0.00038096381,0.00014368328,0.002877824],"category_scores_gemma":[0.0016222032,0.00015279397,0.00024504756,0.00609975,0.00018468627,0.00031601542,0.0006438795,0.00020589547,0.0007919869],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000108264845,0.00002381154,0.95540494,0.00040572902,0.00013376959,0.0001261613,0.00092760014,0.001860363,0.0011561424,0.00042279987,0.0191012,0.020329243],"study_design_scores_gemma":[0.000010164724,0.000010615219,0.98173493,0.0000964713,0.000041079795,0.00013918045,0.0007769591,0.00088134117,0.00037796638,0.00009202662,0.015824214,0.000015018564],"about_ca_topic_score_codex":0.10500429,"about_ca_topic_score_gemma":0.12661824,"teacher_disagreement_score":0.10500429,"about_ca_system_score_codex":0.00050062354,"about_ca_system_score_gemma":0.00054750615,"threshold_uncertainty_score":0.20878619},"labels":[],"label_agreement":null},{"id":"W4411377134","doi":"10.1038/s41597-025-05301-4","title":"The Austronesian and the Micronesian Comparative Dictionaries as CLDF datasets","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"Ministry of Education, India; Ministry of Education - Singapore","keywords":"Micronesian; Austronesian languages; Interoperability; Computer science; Comparative method; Linguistics; Comparative linguistics; Usability; Information retrieval; World Wide Web; History; Genealogy; Sociology; Philology","score_opus":0.0220802469138846,"score_gpt":0.3009350242765415,"score_spread":0.2788547773626569,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411377134","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03656388,0.0020309915,0.023612142,0.0023556177,0.00048634002,0.0005419749,0.8983931,0.005322211,0.030693766],"genre_scores_gemma":[0.044017892,0.0008603535,0.08502169,0.00052402506,0.000055430144,0.0014538397,0.8598603,0.0015554554,0.0066509508],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9969928,0.00054555846,0.0009343072,0.0005706977,0.0007415506,0.00021506143],"domain_scores_gemma":[0.98427737,0.004447724,0.0017150562,0.0055612247,0.003031372,0.0009672327],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040650475,0.00053720496,0.00061145716,0.013393391,0.0016743881,0.0031997475,0.001979699,0.0011116422,0.021758582],"category_scores_gemma":[0.028930273,0.00051646447,0.00061417656,0.019628769,0.0011502419,0.003323723,0.0061560953,0.001996515,0.010823768],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009877781,0.00023345333,0.03306412,0.005870494,0.00017722201,0.0016826083,0.00928383,0.002840149,0.01469454,0.09117221,0.61990714,0.22008641],"study_design_scores_gemma":[0.000052728235,0.000014676394,0.014402408,0.0005953204,0.000027746899,0.00035855095,0.0012656605,0.0009782248,0.0015778227,0.007905713,0.9727683,0.00005298097],"about_ca_topic_score_codex":0.023699006,"about_ca_topic_score_gemma":0.049351234,"teacher_disagreement_score":0.023699006,"about_ca_system_score_codex":0.0028041466,"about_ca_system_score_gemma":0.0071593677,"threshold_uncertainty_score":0.07278973},"labels":[],"label_agreement":null},{"id":"W4411455735","doi":"10.1038/s41597-025-05330-z","title":"The Open Aurignacian Project: 3D scanning and the digital preservation of the Italian Paleolithic record","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Pleistocene-Era Hominins and Archaeology","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Università degli Studi di Ferrara; Eberhard Karls Universität Tübingen; European Commission; Social Sciences and Humanities Research Council of Canada; Université de Montréal; Università degli Studi di Siena; Deutsche Forschungsgemeinschaft; Università degli Studi di Genova; Ministero della cultura","keywords":"Aurignacian; Upper Paleolithic; Archaeology; 3d scanning; Middle Paleolithic; 3d model; Geography; Focus (optics); Computer science; Geology; Artificial intelligence","score_opus":0.06706645450932129,"score_gpt":0.35285486443349146,"score_spread":0.28578840992417015,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411455735","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5472028,0.011863608,0.08363978,0.0028108961,0.00055679475,0.00086634397,0.26478025,0.007796292,0.08048314],"genre_scores_gemma":[0.6295539,0.0042962395,0.13644153,0.00028541117,0.00044355047,0.0020053273,0.2170762,0.0018441754,0.008053752],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982868,0.0004819227,0.00014745265,0.00036688065,0.00056316255,0.00015386613],"domain_scores_gemma":[0.99551857,0.0008315944,0.0006667109,0.0022944552,0.00037347412,0.000315117],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003106565,0.000721263,0.00043159322,0.009781173,0.0009750521,0.0028735606,0.0012283735,0.0005769251,0.006773283],"category_scores_gemma":[0.0053693154,0.00034702194,0.00061221846,0.008905452,0.0018464881,0.0015202007,0.004619207,0.00046094824,0.002416498],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008902797,0.00015402405,0.15996422,0.0019558792,0.0002591954,0.0011503234,0.010534401,0.0030376522,0.007728009,0.016735867,0.084027946,0.7135622],"study_design_scores_gemma":[0.0000691338,0.000103850834,0.40008393,0.00087211607,0.00015142058,0.0022694913,0.004051456,0.0033470558,0.004416677,0.009275473,0.5752378,0.00012162032],"about_ca_topic_score_codex":0.011510851,"about_ca_topic_score_gemma":0.019371074,"teacher_disagreement_score":0.011510851,"about_ca_system_score_codex":0.0008621575,"about_ca_system_score_gemma":0.0019736062,"threshold_uncertainty_score":0.022887707},"labels":[],"label_agreement":null},{"id":"W4411509208","doi":"10.1038/s41597-025-05332-x","title":"A minimum data standard for wildlife disease research and surveillance","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Zoonotic diseases and public health","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Science Foundation","keywords":"Metadata; Wildlife; Data sharing; Wildlife disease; Computer science; Transparency (behavior); Best practice; Data science; Table (database); Data element; Data mining; Information retrieval; World Wide Web; Ecology; Biology; Medicine","score_opus":0.23073925653850116,"score_gpt":0.47858487912309405,"score_spread":0.2478456225845929,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411509208","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reporting","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reporting","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021985399,0.0019207357,0.75514776,0.024976974,0.004086625,0.018309625,0.09709634,0.008636007,0.067840576],"genre_scores_gemma":[0.0539596,0.0017511852,0.79017717,0.0059362724,0.0013222586,0.027474644,0.11121537,0.0012451251,0.006918329],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.82077116,0.068508,0.06676156,0.008744347,0.03113307,0.0040818504],"domain_scores_gemma":[0.49036372,0.1591344,0.031076893,0.14009954,0.1721262,0.007199223],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.18053691,0.0012058044,0.0020758712,0.013704249,0.0041347574,0.011336718,0.007302872,0.0060733645,0.007647318],"category_scores_gemma":[0.29538757,0.0017230264,0.002566376,0.013137267,0.0034028292,0.0117079085,0.007110517,0.0072964444,0.009395492],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013027469,0.001009752,0.041556694,0.005501706,0.00027722828,0.00090968,0.0063633267,0.0073303827,0.016124465,0.29373115,0.34043375,0.2854592],"study_design_scores_gemma":[0.00026707826,0.00038973897,0.014142999,0.0049944464,0.00014556122,0.00079169736,0.0027122116,0.004197716,0.010269415,0.057360876,0.90441215,0.00031621827],"about_ca_topic_score_codex":0.007625676,"about_ca_topic_score_gemma":0.0042007896,"teacher_disagreement_score":0.8194631,"about_ca_system_score_codex":0.005596688,"about_ca_system_score_gemma":0.023935663,"threshold_uncertainty_score":0.95478183},"labels":[],"label_agreement":null},{"id":"W4411881942","doi":"10.1038/s41597-025-05365-2","title":"Cross-cultural data on romantic love and mate preferences from 117,293 participants across 175 countries","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Evolutionary Psychology and Human Behavior","field":"Psychology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University","funders":"Narodowe Centrum Nauki; Eesti Teadusagentuur; Fundacja na rzecz Nauki Polskiej","keywords":"Physical attractiveness; Social psychology; Psychology; Romance; Attractiveness; Jealousy; Generalizability theory; Friendship; Diversity (politics); Trait; Developmental psychology; Sociology","score_opus":0.20308440465524344,"score_gpt":0.482523738825341,"score_spread":0.2794393341700976,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411881942","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8871709,0.0008149918,0.0011168472,0.00015035458,0.0000249241,0.00035939206,0.10502239,0.000024151741,0.005316077],"genre_scores_gemma":[0.85179573,0.0011809833,0.005025067,0.00044145776,0.000033115804,0.0030221925,0.13493103,0.000040299416,0.003530081],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9987534,0.00044153084,0.00023265109,0.00021119464,0.00020601644,0.0001552231],"domain_scores_gemma":[0.9951332,0.0016831656,0.0010457059,0.0007669161,0.00096104166,0.0004099101],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019340328,0.0003505427,0.00050368113,0.0024086493,0.0010163743,0.00079485454,0.0006234991,0.00051491294,0.0065298234],"category_scores_gemma":[0.0054657846,0.0004016885,0.00032446464,0.0055010975,0.00032656093,0.0005706455,0.0014735627,0.000560835,0.0025799815],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024614585,0.0002070468,0.91845995,0.00089860643,0.00017220611,0.0004719096,0.010709237,0.00027462773,0.0014836398,0.00074461004,0.02361822,0.042713776],"study_design_scores_gemma":[0.00003019724,0.00009584883,0.96011144,0.00021597292,0.00006187197,0.00040070232,0.013928418,0.0001692603,0.00050716894,0.00021715283,0.024223492,0.000038346752],"about_ca_topic_score_codex":0.01780464,"about_ca_topic_score_gemma":0.02297834,"teacher_disagreement_score":0.01780464,"about_ca_system_score_codex":0.00039433193,"about_ca_system_score_gemma":0.0008888346,"threshold_uncertainty_score":0.03540206},"labels":[],"label_agreement":null},{"id":"W4411881951","doi":"10.1038/s41597-025-05257-5","title":"An Africa-wide agricultural production database to support policy and satellite-based measurement systems","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Agricultural Economics and Policy","field":"Agricultural and Biological Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"Hertz Foundation; Bill and Melinda Gates Foundation","keywords":"Agriculture; Agricultural productivity; Satellite; Production (economics); Database; Environmental resource management; Remote sensing; Environmental science; Computer science; Geography; Archaeology; Engineering; Economics","score_opus":0.08821594327494711,"score_gpt":0.27868338972732887,"score_spread":0.19046744645238176,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411881951","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018504534,0.00029298675,0.011092402,0.00025727472,0.000039370163,0.00023954948,0.9610585,0.0024257684,0.006089669],"genre_scores_gemma":[0.061021753,0.00051970984,0.034123052,0.000073860276,0.000029055613,0.00071763195,0.9016042,0.00029664917,0.0016140469],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992003,0.00013819516,0.00018308165,0.0002205612,0.00020458641,0.000053314205],"domain_scores_gemma":[0.99486685,0.0011865824,0.0009802581,0.0010733848,0.0015582991,0.0003347645],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001418183,0.00077332626,0.00060654985,0.0053030616,0.00035591214,0.0010797087,0.0011245603,0.00060964504,0.009861648],"category_scores_gemma":[0.0077685225,0.0003645939,0.00031085967,0.013318291,0.00014027505,0.0016473723,0.001029625,0.00048708668,0.0054648337],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00072916894,0.00037167265,0.10429551,0.0028412098,0.0003586346,0.0007925546,0.00094534503,0.03439216,0.00915154,0.010614269,0.55640507,0.27910283],"study_design_scores_gemma":[0.00024840806,0.00007603203,0.12879965,0.0004699521,0.00011917953,0.00028232494,0.0009543441,0.043850604,0.0059312354,0.0053995294,0.8137546,0.00011413793],"about_ca_topic_score_codex":0.02002621,"about_ca_topic_score_gemma":0.016093312,"teacher_disagreement_score":0.02002621,"about_ca_system_score_codex":0.0010361898,"about_ca_system_score_gemma":0.0018420069,"threshold_uncertainty_score":0.0398193},"labels":[],"label_agreement":null},{"id":"W4411883243","doi":"10.1038/s41597-025-05192-5","title":"De novo Genome Assembly and Annotation of 12 Fungi Associated with Fruit Tree Decline Syndrome in ON, Canada","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Plant Pathogens and Fungal Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Agriculture and Agri-Food Canada","funders":"Agriculture and Agri-Food Canada; Hospital for Sick Children; Ministry of Agriculture, Food and Rural Affairs; McGill University; Génome Québec; Ontario Ministry of Agriculture, Food and Rural Affairs; Government of Canada","keywords":"Biology; Genome; Abiotic component; Ecology; Gene; Genetics","score_opus":0.01485561878516139,"score_gpt":0.23740830898332904,"score_spread":0.22255269019816765,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411883243","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3747222,0.010590077,0.018413678,0.00086726685,0.0002801648,0.0005602919,0.5762873,0.0031440924,0.015134858],"genre_scores_gemma":[0.23494726,0.004810229,0.049262114,0.00041785897,0.000042733882,0.00030718363,0.700806,0.0011863665,0.008220202],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993104,0.000028195385,0.00003755506,0.00023298782,0.00023554388,0.0001554328],"domain_scores_gemma":[0.99898916,0.00011240293,0.00012910161,0.000069984715,0.0005112251,0.0001881178],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00056710566,0.0011029793,0.0010579868,0.0030169438,0.0031438472,0.0024660772,0.0008697908,0.00084666576,0.0025283487],"category_scores_gemma":[0.0016270941,0.0005529502,0.001061672,0.0049827164,0.0005016255,0.0004781637,0.0010688804,0.0011333282,0.0014630683],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002925386,0.0002791983,0.10504519,0.0076127136,0.00093926,0.0049300236,0.0074593606,0.010574651,0.60080415,0.004511528,0.100858875,0.15405977],"study_design_scores_gemma":[0.00025485363,0.00018658298,0.48880464,0.001560068,0.0011113846,0.0014904939,0.0033026657,0.011838605,0.028641632,0.0016607558,0.46088436,0.00026390224],"about_ca_topic_score_codex":0.59789467,"about_ca_topic_score_gemma":0.80130875,"teacher_disagreement_score":0.40210533,"about_ca_system_score_codex":0.0052063493,"about_ca_system_score_gemma":0.018255807,"threshold_uncertainty_score":0.80894685},"labels":[],"label_agreement":null},{"id":"W4411885790","doi":"10.1038/s41597-025-05216-0","title":"A Harmonized Dataset of High-Resolution Embodied Life Cycle Assessment Results for Buildings in North America","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Sustainable Building Design and Assessment","field":"Engineering","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Center for the Built Environment; ClimateWorks Foundation; University of Washington; Alfred P. Sloan Foundation","keywords":"Life-cycle assessment; Reuse; Sustainability; Environmental impact assessment; Building design; Computer science; Built environment; Environmental resource management; Set (abstract data type); Impact assessment; Architectural engineering; Environmental science; Civil engineering; Engineering; Production (economics); Ecology","score_opus":0.0279553434322536,"score_gpt":0.30075236889562457,"score_spread":0.272797025463371,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411885790","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023160059,0.0004303986,0.0027152668,0.00020858894,0.0000391116,0.00007027203,0.9672779,0.00086785253,0.0052305902],"genre_scores_gemma":[0.031373,0.00020436666,0.005630929,0.00008012667,0.0000154608,0.00027077438,0.9613928,0.00012477074,0.00090762775],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998722,0.0001947506,0.00011236837,0.00038807868,0.0004453978,0.0001373088],"domain_scores_gemma":[0.9969137,0.0005205049,0.00033575567,0.00066997326,0.0013826742,0.00017741152],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00090071396,0.0007689144,0.00064999925,0.0028877296,0.00075151614,0.0010962737,0.0015826388,0.0010595107,0.005399965],"category_scores_gemma":[0.003285899,0.00040056757,0.0009204284,0.0072492543,0.0004942257,0.00084846665,0.0013674697,0.0008143208,0.0032854413],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027234506,0.0004116551,0.093039446,0.0021544155,0.00044890182,0.0003329443,0.00076001044,0.028266307,0.0033002235,0.005005685,0.81041586,0.05559219],"study_design_scores_gemma":[0.00012318384,0.00004817678,0.20044684,0.0004771517,0.000112162095,0.0001388723,0.0012081952,0.013950049,0.002980383,0.0027591912,0.77763444,0.00012140945],"about_ca_topic_score_codex":0.15561883,"about_ca_topic_score_gemma":0.24926706,"teacher_disagreement_score":0.15561883,"about_ca_system_score_codex":0.002192731,"about_ca_system_score_gemma":0.0031635736,"threshold_uncertainty_score":0.30942607},"labels":[],"label_agreement":null},{"id":"W4411886742","doi":"10.1038/s41597-025-04941-w","title":"A global assemblage of regional prescribed burn records — GlobalRx","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Parks Canada","funders":"European Commission; Research Councils UK","keywords":"Environmental science; Climate change; Environmental resource management; Global warming; Scale (ratio); Geography; Climatology; Ecology; Cartography","score_opus":0.024384418077484607,"score_gpt":0.28147021078266327,"score_spread":0.2570857927051787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411886742","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24481566,0.002278787,0.005706611,0.00018570585,0.00010828688,0.00013507773,0.737387,0.0012889969,0.008093858],"genre_scores_gemma":[0.11783316,0.0006790039,0.010702171,0.00009005863,0.00005774899,0.00019508193,0.86864156,0.00030940922,0.0014917351],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99916756,0.00009541543,0.00016663202,0.00030961927,0.00016466698,0.00009617186],"domain_scores_gemma":[0.99739933,0.00034748085,0.0007236755,0.0006981723,0.00066402374,0.00016735053],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00095161627,0.0006492439,0.00059324293,0.0052304165,0.0003339594,0.0008486196,0.0006193296,0.00041814544,0.003282746],"category_scores_gemma":[0.0028040358,0.00031256516,0.00064427423,0.007846629,0.00023525444,0.0007098618,0.0012419369,0.00053228106,0.002624743],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037560397,0.00010897517,0.73398626,0.0017933655,0.0011595833,0.0004008699,0.0009364474,0.0073357536,0.009693636,0.001850391,0.11820421,0.12415493],"study_design_scores_gemma":[0.000041394804,0.00005644321,0.7981297,0.00021307681,0.00020877647,0.00047986838,0.0005137204,0.0025558504,0.0024983624,0.00045451312,0.19479498,0.000053327552],"about_ca_topic_score_codex":0.021023361,"about_ca_topic_score_gemma":0.025507677,"teacher_disagreement_score":0.021023361,"about_ca_system_score_codex":0.0003987198,"about_ca_system_score_gemma":0.00077383313,"threshold_uncertainty_score":0.04180199},"labels":[],"label_agreement":null},{"id":"W4411887324","doi":"10.1038/s41597-025-05423-9","title":"Characterizing organisms from three domains of life with universal primers from throughout the global ocean","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Microbial Community Ecology and Physiology","field":"Environmental Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Francis Xavier University","funders":"Marine National Facility; University of California, Santa Barbara; University of Southern California; Australian Antarctic Division; University of Tasmania; Simons Foundation; Commonwealth Scientific and Industrial Research Organisation; National Science Foundation","keywords":"Archaea; Biology; Amplicon; Ribosomal RNA; Plankton; Bacteria; Evolutionary biology; Gene; Genetics; Ecology; Polymerase chain reaction","score_opus":0.01839102698263867,"score_gpt":0.2514513170836531,"score_spread":0.23306029010101442,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411887324","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5345341,0.008950492,0.33659604,0.0008975524,0.0005300185,0.0010035335,0.085662924,0.0042132377,0.027612038],"genre_scores_gemma":[0.23835613,0.0033572451,0.6576221,0.0015870408,0.00008321409,0.0025506702,0.08728256,0.0012736372,0.007887284],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9985936,0.00015202239,0.00018004722,0.0006400754,0.0002694398,0.00016484603],"domain_scores_gemma":[0.99924195,0.00011020125,0.0002401154,0.00013623179,0.00020047551,0.000070986716],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011724622,0.0011300225,0.000650642,0.002531989,0.0010218458,0.0011969574,0.00070421206,0.0007728103,0.002578227],"category_scores_gemma":[0.0017854603,0.0005732466,0.00075231475,0.0026044224,0.0005035167,0.0007131892,0.00169118,0.0009046286,0.002613776],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031945595,0.00007517016,0.045328353,0.0008577332,0.0001296575,0.00021116683,0.0010863646,0.00054227136,0.81176704,0.0019510239,0.004662543,0.13306926],"study_design_scores_gemma":[0.00012604111,0.0010009854,0.22332491,0.000906689,0.00060399884,0.0023768246,0.0021583028,0.0038689524,0.41582313,0.0054646926,0.34417358,0.00017185326],"about_ca_topic_score_codex":0.001651405,"about_ca_topic_score_gemma":0.007767284,"teacher_disagreement_score":0.002578227,"about_ca_system_score_codex":0.0004228792,"about_ca_system_score_gemma":0.0012312339,"threshold_uncertainty_score":0.0086250305},"labels":[],"label_agreement":null},{"id":"W4411924796","doi":"10.1038/s41597-025-04674-w","title":"An ecological trait matrix of Neotropical freshwater fishes","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fish biology, ecology, and behavior","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal Ontario Museum; University of Toronto","funders":"","keywords":"Ecology; Fauna; Habitat; Biodiversity; Biology; Trophic level; Geography","score_opus":0.03420529127750205,"score_gpt":0.30513441445103523,"score_spread":0.2709291231735332,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411924796","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9894196,0.00009585512,0.00047693605,0.000025086058,0.0000017815051,0.000018605697,0.009120641,0.000020532541,0.0008210577],"genre_scores_gemma":[0.9869631,0.00011376787,0.0017654074,0.00001174634,0.0000030756748,0.00008212331,0.010653753,0.000009174077,0.0003977449],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99964595,0.000059347705,0.00006395073,0.000102307546,0.000081243656,0.000047124504],"domain_scores_gemma":[0.9984969,0.00031016127,0.00052243023,0.0001493429,0.00038262713,0.0001384526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005344241,0.0002792892,0.00018586619,0.0043792375,0.00034533403,0.0004730156,0.00020572612,0.00011522406,0.0025191873],"category_scores_gemma":[0.0016896334,0.00013188407,0.00025183617,0.0030823438,0.00029831205,0.00027769344,0.0007971856,0.00016682624,0.0002995795],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000041238025,0.000015147807,0.9801886,0.00005916343,0.000090287904,0.00007546334,0.00062279374,0.00040974803,0.002521596,0.00014461689,0.0007724196,0.015058946],"study_design_scores_gemma":[0.0000010001673,0.0000063645334,0.9986532,0.0000048137827,0.000004135896,0.000045640805,0.00029201803,0.00025854394,0.0000435095,0.000048636724,0.00063914515,0.000002932191],"about_ca_topic_score_codex":0.023984628,"about_ca_topic_score_gemma":0.060912047,"teacher_disagreement_score":0.023984628,"about_ca_system_score_codex":0.00044329072,"about_ca_system_score_gemma":0.00044503808,"threshold_uncertainty_score":0.047690094},"labels":[],"label_agreement":null},{"id":"W4411976768","doi":"10.1038/s41597-025-05281-5","title":"A Public Dataset of Annotated Orcinus orca Acoustic Signals for Detection and Ecotype Classification","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Animal Vocal Communication and Behavior","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Fisheries and Oceans Canada; Dalhousie University; Ocean Networks Canada Society; University of Victoria; Simon Fraser University","funders":"","keywords":"Ecotype; Geography; Biology; Ecology","score_opus":0.15728338917184548,"score_gpt":0.3649071657068553,"score_spread":0.2076237765350098,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411976768","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010865182,0.00063379004,0.001167745,0.00010421088,0.00010972182,0.000074606534,0.98262054,0.0017338822,0.0026903637],"genre_scores_gemma":[0.006180584,0.00014941458,0.0016252252,0.000053436204,0.000017905817,0.00014814103,0.99099696,0.00007789855,0.00075049314],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99898213,0.00010298795,0.00011984171,0.00035464673,0.00029055332,0.00014988302],"domain_scores_gemma":[0.99813026,0.0005327662,0.00021508186,0.0003756285,0.00053892826,0.00020728809],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006398079,0.0016177916,0.0009974289,0.0045220456,0.0010739052,0.0011313946,0.0018640566,0.0018711542,0.0112176025],"category_scores_gemma":[0.0028551382,0.00035296567,0.00083268544,0.004213527,0.000588742,0.00074730767,0.002298791,0.0010226347,0.012121751],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00080915465,0.00028923093,0.04064739,0.005420554,0.00027950414,0.0012070462,0.000593936,0.0032799311,0.016854174,0.0012401259,0.8500764,0.07930256],"study_design_scores_gemma":[0.00017627366,0.00013344773,0.19605535,0.001255914,0.00022319831,0.0009764534,0.0009506514,0.0076905596,0.0045632287,0.0014611706,0.7862755,0.0002383591],"about_ca_topic_score_codex":0.03785373,"about_ca_topic_score_gemma":0.08322587,"teacher_disagreement_score":0.03785373,"about_ca_system_score_codex":0.0009635446,"about_ca_system_score_gemma":0.0017725377,"threshold_uncertainty_score":0.07526678},"labels":[],"label_agreement":null},{"id":"W4412152332","doi":"10.1038/s41597-025-05460-4","title":"100 m soil landscape grids of Canada","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Geostatistics and Mapping","field":"Environmental Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Health Canada; Parks Canada; Agriculture and Agri-Food Canada","funders":"Government of Canada","keywords":"Environmental science; Geography; Soil science; Earth science; Geology","score_opus":0.015791123441894428,"score_gpt":0.23615779145484425,"score_spread":0.22036666801294982,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412152332","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19281386,0.001375359,0.038053256,0.0008867833,0.00013745665,0.0006660189,0.6430826,0.005439687,0.11754504],"genre_scores_gemma":[0.47489017,0.0010403541,0.07251104,0.0001737127,0.000012616222,0.0004919469,0.41776314,0.00045224724,0.032664705],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9997353,0.000021416148,0.000014791896,0.000051211286,0.00010177327,0.000075537835],"domain_scores_gemma":[0.99941814,0.000023155699,0.000018231674,0.00005021614,0.00044494338,0.00004528938],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00012749631,0.00032325985,0.00029547358,0.0015771369,0.0006955426,0.0011169837,0.0007923431,0.00019693756,0.016853113],"category_scores_gemma":[0.0010037558,0.0002630346,0.00039622924,0.0057113827,0.00022139231,0.0002768441,0.0005569511,0.0003143852,0.002856014],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005101346,0.00015499005,0.097964644,0.000548569,0.00021425476,0.00043060715,0.0006685383,0.15075266,0.006854864,0.02710952,0.40694046,0.30785072],"study_design_scores_gemma":[0.00014495914,0.000032622633,0.2512769,0.00014909069,0.0000616165,0.00023664662,0.0011747967,0.13266149,0.0035964022,0.0054551573,0.60505587,0.00015445748],"about_ca_topic_score_codex":0.9591213,"about_ca_topic_score_gemma":0.9750308,"teacher_disagreement_score":0.040878713,"about_ca_system_score_codex":0.007143763,"about_ca_system_score_gemma":0.009474984,"threshold_uncertainty_score":0.08223885},"labels":[],"label_agreement":null},{"id":"W4412154136","doi":"10.1038/s41597-025-05520-9","title":"A formulation dataset of poly(lactide-co-glycolide) nanoparticles for small molecule delivery","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Nanoparticle-Based Drug Delivery","field":"Materials Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium; University of Toronto","funders":"Canada First Research Excellence Fund; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Lactide; Nanoparticle; Molecule; PLGA; Chemistry; Chemical engineering; Nanotechnology; Materials science; Combinatorial chemistry; Organic chemistry; Polymer; Engineering; Copolymer","score_opus":0.04680186697591612,"score_gpt":0.30648978730054166,"score_spread":0.2596879203246255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412154136","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02591251,0.010577742,0.003906765,0.0011513066,0.00022076434,0.00048416466,0.94913095,0.0017275077,0.0068882653],"genre_scores_gemma":[0.021722423,0.0030327814,0.010182279,0.0005254967,0.000044444387,0.0006240743,0.9618492,0.00015438614,0.0018649246],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985436,0.00028522214,0.00023293654,0.0003566258,0.0004906026,0.0000910115],"domain_scores_gemma":[0.9977968,0.0008656412,0.00028925945,0.00028602255,0.00060208817,0.00016018863],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015557365,0.0012546768,0.0010644363,0.003792041,0.0008143725,0.0013510191,0.0018436831,0.0026636866,0.0050393017],"category_scores_gemma":[0.0040206793,0.00031211134,0.0013847749,0.004241408,0.0005340167,0.00077606254,0.0010754049,0.0013029638,0.005542272],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016221211,0.0021944556,0.021593813,0.02030921,0.0007215976,0.0019026557,0.00033314235,0.015222293,0.030956328,0.004741383,0.780177,0.12022604],"study_design_scores_gemma":[0.0003846251,0.0004713433,0.020333357,0.00088333146,0.0002941282,0.0011105493,0.0003127639,0.00847927,0.015283755,0.002289344,0.95003784,0.00011967824],"about_ca_topic_score_codex":0.010700964,"about_ca_topic_score_gemma":0.02573075,"teacher_disagreement_score":0.010700964,"about_ca_system_score_codex":0.0014909464,"about_ca_system_score_gemma":0.0028857642,"threshold_uncertainty_score":0.021277368},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4412156013","doi":"10.1038/s41597-025-05453-3","title":"Wave 2 of the Multilingual Eye-Movement Corpus (MECO): New text reading data across languages","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Reading and Literacy Development","field":"Psychology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Social Sciences and Humanities Research Council of Canada; Agencia Nacional de Investigación y Desarrollo; Det Obelske Familiefond; Ministry of Education, India; Azrieli Foundation; Ministarstvo Prosvete, Nauke i Tehnološkog Razvoja; Norges Forskningsråd; Israel Science Foundation; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Aalborg Universitet; Bundesministerium für Bildung und Forschung; National Research University Higher School of Economics; Canada Research Chairs; Government of Canada","keywords":"Reading (process); Computer science; Eye movement; Movement (music); Natural language processing; Linguistics; Artificial intelligence; Art","score_opus":0.08484770351577375,"score_gpt":0.42346698270580246,"score_spread":0.3386192791900287,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412156013","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6781751,0.00062618224,0.011055662,0.0012096511,0.00030254404,0.0049870852,0.27198145,0.0007974581,0.030864915],"genre_scores_gemma":[0.5118466,0.0006475306,0.03902734,0.0012861496,0.0002533946,0.034881048,0.3769293,0.0011111012,0.03401752],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9974226,0.0007020651,0.0003612974,0.00056391227,0.00071917597,0.0002309727],"domain_scores_gemma":[0.98393446,0.004134382,0.0015657826,0.0033434848,0.006020403,0.0010016126],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038695177,0.00059318915,0.0008171382,0.0032287824,0.0015503707,0.0017951679,0.0011349131,0.0008406034,0.008957439],"category_scores_gemma":[0.015020771,0.00047661242,0.0004107724,0.0039831894,0.0007926981,0.0017058095,0.004076326,0.0012707054,0.003924568],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027977023,0.001394542,0.29435807,0.0033438432,0.00033364445,0.0017237203,0.049391117,0.0010136791,0.02596787,0.0062573114,0.24051529,0.37290323],"study_design_scores_gemma":[0.00017683386,0.00023481908,0.8266843,0.00022743738,0.00007665085,0.000552176,0.007435632,0.0003240265,0.0047327145,0.0010508476,0.15835154,0.00015296722],"about_ca_topic_score_codex":0.028791081,"about_ca_topic_score_gemma":0.084390074,"teacher_disagreement_score":0.028791081,"about_ca_system_score_codex":0.0013514683,"about_ca_system_score_gemma":0.0022230325,"threshold_uncertainty_score":0.057246983},"labels":[],"label_agreement":null},{"id":"W4412363010","doi":"10.1038/s41597-025-05540-5","title":"A Cross-Species Brain Magnetic Resonance Imaging and Histology Database of Vertebrates","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"MRI in cancer diagnosis","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"Japan Society for the Promotion of Science; Ministry of Education, Culture, Sports, Science and Technology; Mochida Memorial Foundation for Medical and Pharmaceutical Research; Japan Agency for Medical Research and Development; Takeda Science Foundation","keywords":"Magnetic resonance imaging; Histology; Functional magnetic resonance imaging; Nuclear magnetic resonance; Biology; Neuroscience; Medicine; Pathology; Physics; Radiology","score_opus":0.03809466685632366,"score_gpt":0.34249371533526757,"score_spread":0.3043990484789439,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412363010","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.044059005,0.008669066,0.029547645,0.00068613084,0.00019396862,0.00041132807,0.88539964,0.006317119,0.024716247],"genre_scores_gemma":[0.043499067,0.0033713493,0.053045813,0.00043191537,0.00011457455,0.000903688,0.8930134,0.0012170934,0.004403145],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99856,0.00012727274,0.00048024367,0.00039807564,0.00033357623,0.00010093213],"domain_scores_gemma":[0.99442154,0.0008772993,0.0013659955,0.0012624505,0.0014894941,0.00058321084],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021021443,0.00084715197,0.0011802196,0.014005722,0.0010108973,0.0014300864,0.0018633283,0.0013450012,0.019904131],"category_scores_gemma":[0.005875713,0.0006756503,0.0005876581,0.0074246856,0.0006702844,0.001516436,0.0025871878,0.0009799676,0.011799457],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014738138,0.00023976938,0.057616614,0.010229658,0.00084392057,0.0030261958,0.002397138,0.0013521542,0.09672069,0.009388098,0.48519006,0.3315219],"study_design_scores_gemma":[0.00008656257,0.00009891781,0.13777286,0.0011082998,0.0004743045,0.0040126396,0.0003739877,0.00060790894,0.0069918497,0.0037010934,0.8446476,0.00012394939],"about_ca_topic_score_codex":0.007197136,"about_ca_topic_score_gemma":0.014758398,"teacher_disagreement_score":0.019904131,"about_ca_system_score_codex":0.0008074666,"about_ca_system_score_gemma":0.002829649,"threshold_uncertainty_score":0.06658596},"labels":[],"label_agreement":null},{"id":"W4412392585","doi":"10.1038/s41597-025-05350-9","title":"In vivo submillimeter diffusion MRI dataset of 9 macaque brains curated for tractography","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"HORIZON EUROPE Framework Programme; Université de Bordeaux; Alliance de recherche numérique du Canada; Conseil Régional Aquitaine; European Commission","keywords":"Macaque; Tractography; Diffusion MRI; Neuroscience; Computer science; Biology; Nuclear magnetic resonance; Magnetic resonance imaging; Medicine; Physics; Radiology","score_opus":0.11451306423913672,"score_gpt":0.4241875787324237,"score_spread":0.309674514493287,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412392585","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4679831,0.006059244,0.073525764,0.0013426845,0.00038560326,0.0007028115,0.43080038,0.01123228,0.007968216],"genre_scores_gemma":[0.2349556,0.0016323683,0.086798996,0.00037605548,0.00014291702,0.0012424173,0.6679913,0.0014178791,0.005442491],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996124,0.00004316805,0.00004396392,0.00015759737,0.00009486349,0.000048032132],"domain_scores_gemma":[0.9991073,0.00012371046,0.00009250193,0.00027246465,0.00033283216,0.0000712328],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000552525,0.0012894493,0.0009862211,0.0020305165,0.0009624404,0.0006862238,0.00084288523,0.0011439191,0.0029924035],"category_scores_gemma":[0.001763072,0.00034941142,0.0011873941,0.0017607395,0.0005420072,0.00040673435,0.0011628866,0.0009557679,0.0029714308],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014038201,0.00070028903,0.033411015,0.003925073,0.0014996574,0.0052944943,0.0018305775,0.021827187,0.3199619,0.003952481,0.3054076,0.30078587],"study_design_scores_gemma":[0.00034739226,0.0006698927,0.23588082,0.00069269683,0.00097536785,0.011903077,0.00085864135,0.040470283,0.099842094,0.0099455025,0.5980451,0.00036913788],"about_ca_topic_score_codex":0.017194832,"about_ca_topic_score_gemma":0.038073275,"teacher_disagreement_score":0.017194832,"about_ca_system_score_codex":0.0005840464,"about_ca_system_score_gemma":0.0016174536,"threshold_uncertainty_score":0.034189522},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4412430220","doi":"10.1038/s41597-025-05510-x","title":"Publication of data sharing statements in clinical trials by cardiovascular journals: a quantitative and qualitative analysis","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; St. Joseph’s Healthcare Hamilton; Impact","funders":"Natural Science Foundation of Guangdong Province; National Natural Science Foundation of China","keywords":"Clinical trial; Qualitative analysis; Data sharing; Computer science; Information retrieval; Data science; Qualitative research; Medicine; Alternative medicine; Internal medicine; Sociology; Pathology; Social science","score_opus":0.9769527842570832,"score_gpt":0.7718171202364945,"score_spread":0.2051356640205887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412430220","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9472048,0.0020483418,0.025453936,0.009098564,0.00022672494,0.004889657,0.004313204,0.00011534181,0.006649381],"genre_scores_gemma":[0.95179594,0.0014547207,0.030361684,0.0026043837,0.00011505587,0.011147333,0.0011683559,0.00009474081,0.0012578301],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.70962834,0.23171741,0.026236448,0.0059775445,0.02187694,0.0045634196],"domain_scores_gemma":[0.30152547,0.58804923,0.049281072,0.009181357,0.04706434,0.004898503],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.18790564,0.00063967926,0.0013608088,0.0110923685,0.003845763,0.004788627,0.00162973,0.0014366105,0.002901026],"category_scores_gemma":[0.36666587,0.0008660895,0.001080799,0.011901483,0.006035341,0.006545799,0.00562963,0.0019159183,0.0004322887],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008152173,0.0004215966,0.10921321,0.008973412,0.00031898057,0.00044967022,0.7888261,0.0004240058,0.0033008144,0.0055800327,0.0041755764,0.07750134],"study_design_scores_gemma":[0.0002562491,0.00079888146,0.07796004,0.0051835347,0.00024975205,0.0005511063,0.8643961,0.001989759,0.0041955933,0.007487119,0.036574118,0.00035771474],"about_ca_topic_score_codex":0.00215198,"about_ca_topic_score_gemma":0.002509991,"teacher_disagreement_score":0.98890764,"about_ca_system_score_codex":0.008207966,"about_ca_system_score_gemma":0.013998965,"threshold_uncertainty_score":0.9937519},"labels":[],"label_agreement":null},{"id":"W4412479567","doi":"10.1038/s41597-025-05571-y","title":"Correction: An ecological trait matrix of Neotropical freshwater fishes","year":2025,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Fish biology, ecology, and behavior","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal Ontario Museum; University of Toronto","funders":"","keywords":"Trait; Ecology; Geography; Biology; Computer science","score_opus":0.03664765482473346,"score_gpt":0.2940788143301705,"score_spread":0.25743115950543705,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412479567","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006461042,0.00077918626,0.0026993626,0.036434557,0.9477879,0.000029845087,0.004795541,0.00079040596,0.0060371556],"genre_scores_gemma":[0.052733693,0.009566231,0.02752241,0.05619947,0.28922713,0.00035957247,0.022317816,0.008600195,0.53347355],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9963961,0.00049326423,0.00065501936,0.0005661872,0.0016924189,0.00019703839],"domain_scores_gemma":[0.9502803,0.009543958,0.0019256005,0.0032710088,0.03385916,0.0011200459],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037873227,0.0021689227,0.0013366026,0.0040100836,0.0029190513,0.0032532243,0.0026961467,0.0026789147,0.058616467],"category_scores_gemma":[0.06866866,0.0009579348,0.00093744544,0.0037887802,0.001786983,0.002608182,0.0023161704,0.006369857,0.027070219],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021552356,0.000003334694,0.0001491699,0.00008889317,0.000008729765,0.00014730988,0.00005637346,0.00006656838,0.00006039166,0.0009299644,0.9909094,0.007558301],"study_design_scores_gemma":[0.000022066824,0.000010760832,0.0012637415,0.0003362571,0.00003452726,0.0006255732,0.00013392422,0.0003266889,0.00040124467,0.0017605141,0.9950422,0.000042479136],"about_ca_topic_score_codex":0.02982392,"about_ca_topic_score_gemma":0.028892297,"teacher_disagreement_score":0.058616467,"about_ca_system_score_codex":0.0031157443,"about_ca_system_score_gemma":0.005780403,"threshold_uncertainty_score":0.19609159},"labels":[],"label_agreement":null},{"id":"W4412490415","doi":"10.1038/s41597-025-05582-9","title":"A Machine Learning-Reconstructed Dataset of River Discharge, Temperature, and Heat Flux into the Arctic Ocean","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate change and permafrost","field":"Earth and Planetary Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"European Centre for Medium-Range Weather Forecasts; National Natural Science Foundation of China; Grains Research and Development Corporation; ArcticNet","keywords":"Arctic; Environmental science; Discharge; Climatology; Climate change; Arctic vegetation; Marine ecosystem; Oceanography; Global warming; Ecosystem; Geology; Geography; Ecology","score_opus":0.028587561589018144,"score_gpt":0.2568026545127386,"score_spread":0.22821509292372044,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412490415","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47409979,0.0011079342,0.027422544,0.0010679385,0.0005173626,0.00009949341,0.48458576,0.005519168,0.0055800206],"genre_scores_gemma":[0.37408862,0.00042951907,0.02545122,0.00017622217,0.00009811393,0.00013240904,0.59790653,0.00025231286,0.0014650224],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995763,0.00009604801,0.00004842853,0.00014088057,0.000087140514,0.000051260988],"domain_scores_gemma":[0.9991973,0.00016047299,0.00008330265,0.00019017569,0.000310104,0.00005876756],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009065881,0.0005272854,0.0004403096,0.0011201933,0.00041823974,0.0006330491,0.0007418279,0.00064634584,0.0014677563],"category_scores_gemma":[0.002670399,0.00020125041,0.0009599806,0.0018974737,0.00036163002,0.00043559738,0.00072800566,0.00096245314,0.001414067],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008567772,0.00076005957,0.29754075,0.0008275368,0.0013150179,0.00083177414,0.00043314506,0.33365136,0.013199412,0.004748939,0.22521374,0.12062147],"study_design_scores_gemma":[0.0004257888,0.00019456178,0.30862448,0.0003990223,0.00023379544,0.00045341125,0.0006287106,0.50118804,0.015396717,0.005829769,0.16637795,0.00024780133],"about_ca_topic_score_codex":0.04122421,"about_ca_topic_score_gemma":0.061733305,"teacher_disagreement_score":0.04122421,"about_ca_system_score_codex":0.00057949044,"about_ca_system_score_gemma":0.0019023075,"threshold_uncertainty_score":0.081968546},"labels":[],"label_agreement":null},{"id":"W4412520569","doi":"10.1038/s41597-025-05597-2","title":"A multi-source melt pool compilation for vision-based analytics applications in additive manufacturing","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Additive Manufacturing Materials and Processes","field":"Engineering","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Alliance de recherche numérique du Canada; Mitacs; McGill University","keywords":"Analytics; Computer science; Process (computing); Process analytical technology; Standardization; Process engineering; Data mining; Work in process","score_opus":0.032198137681128336,"score_gpt":0.3008100488377761,"score_spread":0.2686119111566478,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412520569","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021230657,0.0035403392,0.04018309,0.00039199734,0.0005304043,0.00049244607,0.8364753,0.08847226,0.008683395],"genre_scores_gemma":[0.014608479,0.00064062595,0.042463847,0.00012505098,0.000039463208,0.0004817209,0.93739843,0.0023023672,0.0019400481],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988821,0.00007540056,0.000102714184,0.0003713304,0.00044997619,0.000118398195],"domain_scores_gemma":[0.99917763,0.00011726304,0.00007045542,0.000326525,0.00025283443,0.000055151304],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008820414,0.0030178889,0.0012062765,0.0042539164,0.0008907871,0.0018507191,0.0028884383,0.0015934107,0.011275064],"category_scores_gemma":[0.0026627649,0.00078650494,0.0026011558,0.0037933148,0.00046938125,0.0018971015,0.0026547823,0.0015844392,0.014828206],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00080235954,0.00052175997,0.006648421,0.004083892,0.00044606958,0.00049691525,0.00025190605,0.02056167,0.027930195,0.00308222,0.7836322,0.15154229],"study_design_scores_gemma":[0.0003273222,0.0002711135,0.019596353,0.00061191764,0.00019166939,0.0006181538,0.00027504328,0.07285865,0.0455231,0.006139762,0.8533554,0.00023152928],"about_ca_topic_score_codex":0.008042267,"about_ca_topic_score_gemma":0.020907303,"teacher_disagreement_score":0.011275064,"about_ca_system_score_codex":0.0010942448,"about_ca_system_score_gemma":0.0016705133,"threshold_uncertainty_score":0.037718832},"labels":[],"label_agreement":null},{"id":"W4412571399","doi":"10.1038/s41597-025-05575-8","title":"A high-resolution database of historical and future climate for Africa developed with deep neural networks","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Government of British Columbia; University of British Columbia; University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Downscaling; Climate model; Orographic lift; Climate change; Computer science; Database; Precipitation; Climatology; Grid; Artificial neural network; Meteorology; Environmental science; Data mining; Remote sensing; Geography; Machine learning; Geology","score_opus":0.03540132981286964,"score_gpt":0.25168378708427913,"score_spread":0.2162824572714095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412571399","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2313646,0.0010807733,0.055904116,0.0008222862,0.00024806216,0.000189117,0.6943062,0.005800293,0.010284506],"genre_scores_gemma":[0.44129965,0.0010812479,0.08687888,0.00008813623,0.000068774294,0.00039645986,0.4671961,0.00037808227,0.0026126346],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99981135,0.00003332419,0.000024813586,0.000058478305,0.000047858048,0.00002412658],"domain_scores_gemma":[0.9993338,0.00013937154,0.000097139586,0.00017307591,0.00021070639,0.00004587087],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00048840896,0.00046860983,0.0003110835,0.0026852207,0.00033313295,0.000657653,0.0009225545,0.0004753708,0.006176676],"category_scores_gemma":[0.002409989,0.0002629599,0.00039522708,0.004790309,0.00014030514,0.0009892592,0.0006826444,0.00058555166,0.0017691338],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003651779,0.0003874116,0.096210234,0.0008962201,0.00040637224,0.0013237117,0.00041625276,0.4596621,0.0075992155,0.008665084,0.14119537,0.28287286],"study_design_scores_gemma":[0.0002050312,0.00010352021,0.10909203,0.0004101097,0.00017551889,0.0004064478,0.0008315429,0.5953602,0.017902644,0.011711882,0.26362878,0.00017225835],"about_ca_topic_score_codex":0.023725798,"about_ca_topic_score_gemma":0.026986212,"teacher_disagreement_score":0.023725798,"about_ca_system_score_codex":0.00067725556,"about_ca_system_score_gemma":0.0009638256,"threshold_uncertainty_score":0.047175407},"labels":[],"label_agreement":null},{"id":"W4412630680","doi":"10.1038/s41597-025-05638-w","title":"Mesopelagic Mesozooplankton and Micronekton Database","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine Invertebrate Physiology and Ecology","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Broadcom (Canada); University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Mesopelagic zone; Oceanography; Database; Environmental science; Geology; Pelagic zone; Computer science","score_opus":0.024746571007223696,"score_gpt":0.250270855309319,"score_spread":0.22552428430209528,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412630680","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022213897,0.00069888344,0.0004226544,0.00007741193,0.000025472467,0.000081927516,0.9721048,0.00013774843,0.0042371987],"genre_scores_gemma":[0.02812854,0.00061251543,0.0018556791,0.000051542313,0.00002338755,0.00041688542,0.9671931,0.000033167227,0.001685098],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99951684,0.000039698956,0.00014880672,0.00012697912,0.000101304,0.00006636143],"domain_scores_gemma":[0.998781,0.00010009893,0.00033878212,0.00014911767,0.00047445795,0.00015660637],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005976013,0.0006054764,0.0007222074,0.004533811,0.0003740136,0.0007928439,0.0009931982,0.00034060952,0.009493871],"category_scores_gemma":[0.002151603,0.00030689366,0.00044941824,0.005172509,0.0001385229,0.0006200156,0.0014214666,0.00051851175,0.0042968984],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00092539185,0.0002128653,0.43680555,0.009665974,0.0007943344,0.00081449165,0.0016228175,0.0021483796,0.0055112843,0.006316914,0.4470022,0.08817975],"study_design_scores_gemma":[0.00009316177,0.000046205634,0.5081225,0.0008643219,0.00021074369,0.00021066974,0.0008127876,0.00083555526,0.0011058314,0.0007704568,0.48687387,0.000053882173],"about_ca_topic_score_codex":0.03721766,"about_ca_topic_score_gemma":0.04876298,"teacher_disagreement_score":0.03721766,"about_ca_system_score_codex":0.0007242009,"about_ca_system_score_gemma":0.0022079505,"threshold_uncertainty_score":0.07400209},"labels":[],"label_agreement":null},{"id":"W4412761082","doi":"10.1038/s41597-025-05619-z","title":"High-Resolution Ultrasound Data for AI-Based Segmentation in Mouse Brain Tumor","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Izaak Walton Killam Health Centre; Dalhousie University","funders":"Canadian Institutes of Health Research; Dalhousie University","keywords":"Segmentation; Ultrasound; Artificial intelligence; Computer science; Computational biology; Pattern recognition (psychology); Biology; Medicine; Radiology","score_opus":0.021195710511228493,"score_gpt":0.32581800951468665,"score_spread":0.30462229900345816,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412761082","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.50001115,0.0067490744,0.2660012,0.0017486162,0.00053458253,0.00057371805,0.18938242,0.026192322,0.008806911],"genre_scores_gemma":[0.50115335,0.004360062,0.24173264,0.000478064,0.00010827833,0.00079790596,0.24199772,0.0035503486,0.0058216853],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9996018,0.00004774081,0.000051788465,0.00012230275,0.00012494985,0.000051381037],"domain_scores_gemma":[0.9992186,0.00018994755,0.00012957302,0.00014011771,0.00025874458,0.000063037645],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008048498,0.00089347846,0.0005302567,0.0028048027,0.0003392539,0.00086298137,0.0007203814,0.0011420106,0.0030739664],"category_scores_gemma":[0.0013944608,0.00038723776,0.00086314016,0.0018358661,0.00040408433,0.00042436956,0.0006213555,0.00080593495,0.002141183],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018666851,0.0007586137,0.017234415,0.0024709639,0.00042649807,0.0011512022,0.00046491955,0.09271721,0.6007608,0.0037618887,0.060303554,0.2180833],"study_design_scores_gemma":[0.00015323797,0.00074641354,0.064834446,0.00049236923,0.0004027736,0.003540852,0.00058514084,0.37791905,0.42640913,0.0070730206,0.11748973,0.0003538327],"about_ca_topic_score_codex":0.0069407346,"about_ca_topic_score_gemma":0.011124573,"teacher_disagreement_score":0.0069407346,"about_ca_system_score_codex":0.0007098138,"about_ca_system_score_gemma":0.0008039967,"threshold_uncertainty_score":0.013800621},"labels":[],"label_agreement":null},{"id":"W4412779950","doi":"10.1038/s41597-025-05653-x","title":"Sea lice infestation dataset for wild and farmed salmon populations on the Pacific coast of Canada (2001–2023)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Parasite Biology and Host Interactions","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto; Health PEI; University of Prince Edward Island","funders":"","keywords":"Infestation; Fishery; Biology; Aquaculture; Lepeophtheirus; Juvenile; Fishing; Geography; Ecology; Fish <Actinopterygii>","score_opus":0.056472108171808345,"score_gpt":0.35384665693180073,"score_spread":0.29737454875999236,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412779950","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023592895,0.00019154239,0.000104472936,0.00006646196,0.000011946775,0.000042065647,0.9744685,0.00009325083,0.0014289015],"genre_scores_gemma":[0.041342475,0.00022851009,0.00038541152,0.00007425701,0.0000061710202,0.000106911524,0.9558152,0.000020936963,0.0020200573],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995609,0.000017246986,0.00005143266,0.00008330546,0.00016160184,0.00012551942],"domain_scores_gemma":[0.997334,0.0000764958,0.00034220683,0.00012564678,0.0018268698,0.00029485757],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039741743,0.0007201099,0.0005536888,0.0031289107,0.0010771399,0.0009693594,0.0010617732,0.00041502883,0.0055661364],"category_scores_gemma":[0.0016035838,0.00032946662,0.0005136066,0.007525154,0.00024781554,0.0002920882,0.0006612905,0.0007170198,0.0025503496],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004275178,0.000099451645,0.52936584,0.00086495985,0.00048045992,0.0002868444,0.00034643398,0.001820819,0.0009396565,0.00047647028,0.44656575,0.018325774],"study_design_scores_gemma":[0.00008489912,0.00003056753,0.9056162,0.00019054124,0.000083203595,0.00014066306,0.00056605675,0.0012201624,0.00043095538,0.000096769385,0.09150076,0.000039229053],"about_ca_topic_score_codex":0.97948575,"about_ca_topic_score_gemma":0.9869802,"teacher_disagreement_score":0.02051425,"about_ca_system_score_codex":0.008230522,"about_ca_system_score_gemma":0.012723768,"threshold_uncertainty_score":0.05971688},"labels":[],"label_agreement":null},{"id":"W4412810094","doi":"10.1038/s41597-025-05669-3","title":"Genome-wide analysis of host-encoded microRNAs modulating SARS-CoV-2 infection","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"SARS-CoV-2 and COVID-19 Research","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Peter MacCallum Cancer Centre; Australian Government; Commonwealth Scientific and Industrial Research Organisation; Canada Excellence Research Chairs, Government of Canada; Australian Cancer Research Foundation; University of Melbourne","keywords":"microRNA; Genome; Biology; Host (biology); Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Computational biology; Coronavirus disease 2019 (COVID-19); Virology; Genetics; Gene; Medicine; Infectious disease (medical specialty)","score_opus":0.08601688917498375,"score_gpt":0.3898502404357454,"score_spread":0.30383335126076166,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412810094","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97670424,0.001062696,0.004662533,0.00013958362,0.000028892056,0.00006267942,0.014341615,0.00037815422,0.002619555],"genre_scores_gemma":[0.9546121,0.0008990377,0.014559829,0.0003726118,0.000014346222,0.00009497863,0.026906513,0.00018400414,0.002356673],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99971205,0.000024698535,0.000017650018,0.00010248531,0.00010835874,0.000034833374],"domain_scores_gemma":[0.9997657,0.000074770374,0.00006733959,0.000020252397,0.000041729363,0.000030249457],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00038244188,0.0002622896,0.00035965093,0.00045383954,0.0002218513,0.00042244888,0.00014268741,0.00031267037,0.0012255523],"category_scores_gemma":[0.00036233128,0.00016144325,0.0003683004,0.00036373307,0.00015767303,0.00013272795,0.0002496561,0.0003677257,0.00042952958],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031516035,0.000044854463,0.006100507,0.00008745788,0.00005020234,0.000079638594,0.0000273824,0.00074554246,0.9859881,0.00013447995,0.0004406674,0.005986099],"study_design_scores_gemma":[0.00009214787,0.00084601145,0.20889665,0.000030478855,0.00032416696,0.0009597301,0.00016290814,0.011083385,0.7580674,0.00039192004,0.019079551,0.00006562641],"about_ca_topic_score_codex":0.0010974241,"about_ca_topic_score_gemma":0.0028229686,"teacher_disagreement_score":0.0012255523,"about_ca_system_score_codex":0.00026460568,"about_ca_system_score_gemma":0.00039531826,"threshold_uncertainty_score":0.0040999055},"labels":[],"label_agreement":null},{"id":"W4412843976","doi":"10.1038/s41597-025-05617-1","title":"Tsunami Runup Survey Data From The Taan Fjord Landslide Event","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Landslides and related hazards","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Selkirk College","funders":"Division of Civil, Mechanical and Manufacturing Innovation; National Park Service; National Science Foundation","keywords":"Fjord; Geology; Landslide; Event (particle physics); Oceanography; Geography; Seismology; Physics","score_opus":0.05290228076373086,"score_gpt":0.30273011749594375,"score_spread":0.2498278367322129,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412843976","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.59936637,0.0002853992,0.00048054566,0.00012839989,0.00004584679,0.00012017515,0.3916854,0.00027957762,0.0076083015],"genre_scores_gemma":[0.28801867,0.00022104905,0.002015726,0.0000732945,0.000028922084,0.00017718985,0.70455736,0.000036476453,0.004871365],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99967,0.000022548273,0.000045125005,0.00006180298,0.00013159073,0.00006889356],"domain_scores_gemma":[0.99904126,0.000060901082,0.00019192704,0.00012894739,0.00045738101,0.000119601485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00030807714,0.00033686627,0.000212877,0.0021719248,0.00053438003,0.000566614,0.00039115123,0.00032629637,0.0018445643],"category_scores_gemma":[0.00088153523,0.00014304597,0.0001694647,0.002544461,0.0001728258,0.00019863484,0.00052231294,0.00026004086,0.0012349196],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038360924,0.00042008262,0.7555084,0.00040816786,0.00018268565,0.0011282039,0.0021425653,0.0042777816,0.006226919,0.000818976,0.17212956,0.056372967],"study_design_scores_gemma":[0.000018432758,0.000031249594,0.94616127,0.000055567856,0.00001965558,0.00006519381,0.00076938566,0.0014182762,0.0008297297,0.000046211702,0.050567802,0.000017315182],"about_ca_topic_score_codex":0.28363857,"about_ca_topic_score_gemma":0.4608746,"teacher_disagreement_score":0.28363857,"about_ca_system_score_codex":0.0008274872,"about_ca_system_score_gemma":0.0017048141,"threshold_uncertainty_score":0.5639752},"labels":[],"label_agreement":null},{"id":"W4412958423","doi":"10.1038/s41597-025-05131-4","title":"A global dataset of terrestrial biological nitrogen fixation","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Carbon and Nitrogen Dynamics","field":"Agricultural and Biological Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Dalhousie University","funders":"Leverhulme Trust","keywords":"Biome; Terrestrial ecosystem; Abundance (ecology); Terrestrial plant; Environmental science; Lichen; Litter; Nitrogen fixation; Ecosystem; Ecology; Relative species abundance; Nitrogen; Biology; Chemistry","score_opus":0.060609636333438084,"score_gpt":0.2954699647597412,"score_spread":0.2348603284263031,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412958423","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0047280914,0.00029283637,0.00037943939,0.00004343017,0.00001538517,0.000010600981,0.9937039,0.00017926712,0.0006470385],"genre_scores_gemma":[0.0072422037,0.00015428565,0.00096905016,0.000038279974,0.000009078002,0.00007470599,0.9912656,0.000040829065,0.0002060025],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986657,0.00019590232,0.00017115346,0.00055485574,0.000275009,0.000137352],"domain_scores_gemma":[0.99713385,0.0006640705,0.00061241956,0.000498275,0.00085636444,0.00023501249],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010398652,0.0013187068,0.0011579742,0.004491599,0.00045382162,0.0011341646,0.001414139,0.0011558463,0.0062884],"category_scores_gemma":[0.0031916774,0.00048399987,0.0014513964,0.01171021,0.00031336778,0.000886935,0.0012871266,0.00097594823,0.0063151815],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005803046,0.0002759979,0.1504269,0.007816093,0.0019509464,0.00044482056,0.00037576363,0.031657528,0.0083655,0.0049703154,0.741031,0.05210483],"study_design_scores_gemma":[0.00030048483,0.00008458187,0.20521046,0.0006966814,0.00031954644,0.00021968767,0.00025214473,0.005830433,0.002631699,0.0024362996,0.7818966,0.00012147633],"about_ca_topic_score_codex":0.021845765,"about_ca_topic_score_gemma":0.018305527,"teacher_disagreement_score":0.021845765,"about_ca_system_score_codex":0.0009679054,"about_ca_system_score_gemma":0.0016600033,"threshold_uncertainty_score":0.043437183},"labels":[],"label_agreement":null},{"id":"W4413054501","doi":"10.1038/s41597-025-05750-x","title":"Correction: Global Impacts Dataset of Invasive Alien Species (GIDIAS)","year":2025,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Armed Forces","funders":"","keywords":"Alien; Ecology; Alien species; Invasive species; Geography; Data science; Biology; Computer science; Political science","score_opus":0.05863488910597562,"score_gpt":0.30234846505438717,"score_spread":0.24371357594841156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413054501","genre_codex":"editorial","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00074678665,0.00066801725,0.001989524,0.035271633,0.54083693,0.00013452848,0.40050748,0.0032882344,0.016556818],"genre_scores_gemma":[0.020740887,0.0028693043,0.016177015,0.046500135,0.046995394,0.0013271743,0.4958421,0.012429303,0.3571187],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9947724,0.0006394804,0.0009553224,0.0007081474,0.0024837959,0.00044093712],"domain_scores_gemma":[0.94954187,0.009277969,0.0024951794,0.005380477,0.031617224,0.0016873297],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0051643224,0.002374469,0.0018247447,0.008310511,0.0021863978,0.005409539,0.003231876,0.0032969194,0.17573537],"category_scores_gemma":[0.0727659,0.0014542665,0.0016145018,0.0076508424,0.001297581,0.0031815683,0.0034432772,0.0062703574,0.094357066],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015645686,0.000002570595,0.00009277948,0.000050312174,0.0000063232137,0.000017847957,0.000005720894,0.00004538261,0.000009657651,0.00015480621,0.99833727,0.0012615614],"study_design_scores_gemma":[0.00007440964,0.0000059789963,0.0016070628,0.00027017668,0.00002626677,0.000057050762,0.0000660346,0.0002419572,0.00016200672,0.0010518167,0.99640936,0.000027888822],"about_ca_topic_score_codex":0.09051948,"about_ca_topic_score_gemma":0.08430195,"teacher_disagreement_score":0.17573537,"about_ca_system_score_codex":0.004209521,"about_ca_system_score_gemma":0.009659989,"threshold_uncertainty_score":0.58789337},"labels":[],"label_agreement":null},{"id":"W4413054850","doi":"10.1038/s41597-025-05543-2","title":"MRS-BIDS, an extension to the Brain Imaging Data Structure for magnetic resonance spectroscopy","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced MRI Techniques and Applications","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital; Western University","funders":"National Institute on Aging; U.S. Department of Health and Human Services; National Institutes of Health; Austrian Science Fund; National Institute of Biomedical Imaging and Bioengineering; Wellcome Trust; Wellcome","keywords":"Extension (predicate logic); Nuclear magnetic resonance; Magnetic resonance imaging; Spectroscopy; Nuclear magnetic resonance spectroscopy; Materials science; Physics; Computer science; Medicine; Astronomy; Radiology","score_opus":0.048309771907065105,"score_gpt":0.40685938455419296,"score_spread":0.35854961264712787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413054850","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025892772,0.0020929507,0.65182924,0.0020756477,0.0011503602,0.0015188507,0.097766384,0.22900765,0.011969694],"genre_scores_gemma":[0.024504982,0.002346311,0.68530923,0.0025496157,0.0004927404,0.0038277428,0.21871705,0.05291537,0.009336915],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9909738,0.0020219148,0.0016611717,0.0014000679,0.0033248444,0.00061822543],"domain_scores_gemma":[0.98545897,0.0038252955,0.00113137,0.0049737953,0.0036901,0.0009204326],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011492055,0.002692645,0.0022774737,0.0044893986,0.002015578,0.0069599156,0.006432505,0.0022419624,0.036751136],"category_scores_gemma":[0.04185942,0.0023687806,0.0038638606,0.004801361,0.0015554242,0.011357441,0.016315306,0.0055214562,0.04056727],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002405406,0.00028736232,0.0041152877,0.0032725078,0.0005524111,0.0006827488,0.00097379176,0.006581526,0.008926655,0.091926195,0.62071335,0.25956285],"study_design_scores_gemma":[0.0003838937,0.00016004092,0.0015455917,0.00052463793,0.00010701126,0.0006539107,0.00019702537,0.021056145,0.00893758,0.072027266,0.8942006,0.00020633612],"about_ca_topic_score_codex":0.005266547,"about_ca_topic_score_gemma":0.0063617406,"teacher_disagreement_score":0.036751136,"about_ca_system_score_codex":0.0016225289,"about_ca_system_score_gemma":0.0074286815,"threshold_uncertainty_score":0.12294477},"labels":[],"label_agreement":null},{"id":"W4413055262","doi":"10.1038/s41597-025-05699-x","title":"Baseline high-resolution maps of soil nutrients in Morocco to support sustainable agriculture","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Geostatistics and Mapping","field":"Environmental Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Agriculture; Baseline (sea); Sustainability; Environmental science; Precision agriculture; Bootstrapping (finance); Vegetation (pathology); Digital soil mapping; Fertilizer; Environmental resource management; Soil map; Agricultural engineering; Computer science; Soil science; Geography; Mathematics; Soil water; Ecology; Econometrics; Engineering","score_opus":0.012705279275881866,"score_gpt":0.24755638078252454,"score_spread":0.23485110150664268,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413055262","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9403784,0.0021358558,0.0077027883,0.0008009392,0.000066387496,0.000078900885,0.03369114,0.0007136892,0.014431844],"genre_scores_gemma":[0.9821315,0.00033705414,0.01054156,0.000027029999,0.0000127942985,0.000045375604,0.0061588557,0.000024848821,0.00072092196],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998186,0.000040067658,0.0000119098295,0.0000355551,0.00004357324,0.00005023626],"domain_scores_gemma":[0.99950993,0.00007730348,0.00006858029,0.00006179391,0.000259286,0.000023034216],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004996341,0.00027733573,0.00018320161,0.0017321092,0.00031110217,0.00064898777,0.00033555258,0.00027926764,0.0015174111],"category_scores_gemma":[0.0016339677,0.000087595334,0.00019178611,0.0021731837,0.00016742255,0.0003750551,0.0005253722,0.00013681712,0.0002913895],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008551283,0.00023592006,0.4592787,0.00094112696,0.0002867736,0.0015329522,0.0016492415,0.10132073,0.018526522,0.005882967,0.020312084,0.38917783],"study_design_scores_gemma":[0.000052439627,0.0000716449,0.8599161,0.0002820395,0.00009608804,0.000259056,0.0015976834,0.06923111,0.0055480124,0.0018634055,0.061016284,0.00006601585],"about_ca_topic_score_codex":0.10664722,"about_ca_topic_score_gemma":0.14148149,"teacher_disagreement_score":0.10664722,"about_ca_system_score_codex":0.0015437668,"about_ca_system_score_gemma":0.0012456066,"threshold_uncertainty_score":0.21205294},"labels":[],"label_agreement":null},{"id":"W4413112481","doi":"10.1038/s41597-025-05652-y","title":"Phased genome assemblies and pangenome graphs of human populations of Japan and Saudi Arabia","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Institute of Genetics; Japan Science and Technology Agency; National Bioscience Database Center; King Salman Center for Disability Research; Dalian Institute of Chemical Physics; King Abdullah University of Science and Technology; Japan Society for the Promotion of Science; Ministry of Education, Culture, Sports, Science and Technology; King Fahad Medical City; Japan Agency for Medical Research and Development","keywords":"Biology; Haplotype; Reference genome; Genome; Population; Graph; Human genome; Computational biology; Gene; Genetics; Computer science; Allele; Theoretical computer science; Medicine","score_opus":0.039620647400108455,"score_gpt":0.31301324952601484,"score_spread":0.2733926021259064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413112481","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.78914404,0.0025931613,0.14378889,0.00033216673,0.0001054532,0.00016967367,0.057273008,0.0024898788,0.004103885],"genre_scores_gemma":[0.6821992,0.0016110949,0.19244088,0.00019655717,0.000029858771,0.0002505914,0.120805606,0.00086587423,0.0016002778],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99957556,0.00010326005,0.00004553473,0.00017819119,0.00006899389,0.000028454064],"domain_scores_gemma":[0.99863356,0.000439064,0.00023394826,0.0002720865,0.00036542653,0.00005590607],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007904608,0.00043773526,0.0002834963,0.002354905,0.00047866555,0.0005563829,0.0005185402,0.00039179611,0.002246221],"category_scores_gemma":[0.0038508207,0.00028642832,0.00061407883,0.0030754765,0.00019939081,0.0004644547,0.0006966468,0.0003200788,0.0005007906],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003098926,0.0001350835,0.23129576,0.0031836932,0.0014206589,0.0017780106,0.004981965,0.069489904,0.27885506,0.014401337,0.01952994,0.37182972],"study_design_scores_gemma":[0.0001987006,0.0005187364,0.5526791,0.0005413174,0.0012265573,0.0031445762,0.003078379,0.13725117,0.07978911,0.021743054,0.19956045,0.00026882393],"about_ca_topic_score_codex":0.010022191,"about_ca_topic_score_gemma":0.015819913,"teacher_disagreement_score":0.010022191,"about_ca_system_score_codex":0.00044528124,"about_ca_system_score_gemma":0.0005532554,"threshold_uncertainty_score":0.01992774},"labels":[],"label_agreement":null},{"id":"W4413116177","doi":"10.1038/s41597-025-05792-1","title":"A dataset of high-resolution plantar pressures for gait analysis across varying footwear and walking speeds","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Diabetic Foot Ulcer Assessment and Management","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of New Brunswick","funders":"New Brunswick Innovation Foundation; Natural Sciences and Engineering Research Council of Canada; Atlantic Canada Opportunities Agency; Government of Canada","keywords":"Gait; Gait analysis; Plantar pressure; Physical medicine and rehabilitation; Plantar flexion; Preferred walking speed; Computer science; Medicine; Ankle; Engineering; Anatomy","score_opus":0.04221187040446684,"score_gpt":0.3595110626952025,"score_spread":0.31729919229073567,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413116177","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.39212996,0.0020197493,0.01173542,0.0005703313,0.00032620825,0.00052242144,0.5865551,0.0027068576,0.0034340317],"genre_scores_gemma":[0.2066999,0.00055547047,0.014309888,0.00019768432,0.00007963059,0.00061232946,0.7755368,0.00007010052,0.0019382832],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99951875,0.00006625721,0.00006533527,0.00013351483,0.00015073566,0.00006541901],"domain_scores_gemma":[0.99933237,0.00012032023,0.00009458257,0.00011865371,0.00021761928,0.00011641024],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041421028,0.0011034128,0.00074432185,0.0015654027,0.0003646954,0.0005252075,0.0011031064,0.0012238567,0.002146754],"category_scores_gemma":[0.0014047354,0.00018566425,0.0006887374,0.0014298373,0.00029348306,0.00029809028,0.0008387104,0.00058000605,0.0022227596],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003123364,0.0036025292,0.16004565,0.004035128,0.001081187,0.0024450587,0.0004237635,0.020009805,0.031222604,0.0010298651,0.48097214,0.2920089],"study_design_scores_gemma":[0.00073315465,0.0017980022,0.76146406,0.0009477796,0.00038300187,0.003349325,0.0011404277,0.08317256,0.019791208,0.002831981,0.12408904,0.00029955615],"about_ca_topic_score_codex":0.005820594,"about_ca_topic_score_gemma":0.013144781,"teacher_disagreement_score":0.005820594,"about_ca_system_score_codex":0.00024621384,"about_ca_system_score_gemma":0.0006360033,"threshold_uncertainty_score":0.011573434},"labels":[],"label_agreement":null},{"id":"W4413312032","doi":"10.1038/s41597-025-05791-2","title":"Hierarchical Event Descriptor library schema for EEG data annotation","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"EEG and Brain-Computer Interfaces","field":"Neuroscience","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Neurological Institute and Hospital","funders":"National Institutes of Health; University of Electronic Science and Technology of China; National Institute of Mental Health; U.S. Department of Health and Human Services","keywords":"Schema (genetic algorithms); Annotation; Computer science; Information retrieval; Electroencephalography; Natural language processing; Artificial intelligence; Psychology; Neuroscience","score_opus":0.11280935658473884,"score_gpt":0.35039420488544903,"score_spread":0.2375848483007102,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413312032","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0039899405,0.0004632114,0.8537861,0.000871912,0.00026749153,0.0013170458,0.076465584,0.045238584,0.017600024],"genre_scores_gemma":[0.07074548,0.0012107133,0.5866597,0.0018115699,0.00020292093,0.0038212864,0.30065557,0.0112985745,0.023594137],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99646026,0.0007198133,0.0012059025,0.0006607303,0.0007767114,0.00017657953],"domain_scores_gemma":[0.9931967,0.0017189407,0.0004650346,0.002009718,0.0023880631,0.00022142673],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047542998,0.0012127063,0.00073869666,0.004273363,0.0009749018,0.0044348133,0.0024034705,0.0013453779,0.02519402],"category_scores_gemma":[0.011260062,0.00069650775,0.0018209101,0.0038945875,0.0010267333,0.0053995843,0.004166007,0.0018771081,0.013807827],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00080410397,0.00027736096,0.0059681237,0.0022750604,0.00019434252,0.0010960981,0.0029583138,0.011552346,0.01501395,0.29860905,0.3832703,0.2779809],"study_design_scores_gemma":[0.00009949326,0.00009283534,0.002183925,0.00039343032,0.000067362125,0.0008226972,0.0006186028,0.027878178,0.015938452,0.06395555,0.88781095,0.00013849045],"about_ca_topic_score_codex":0.010425189,"about_ca_topic_score_gemma":0.008964939,"teacher_disagreement_score":0.02519402,"about_ca_system_score_codex":0.0028172506,"about_ca_system_score_gemma":0.004201845,"threshold_uncertainty_score":0.0842824},"labels":[],"label_agreement":null},{"id":"W4413341034","doi":"10.1038/s41597-026-07298-w","title":"VO: The Vaccine Ontology","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research","funders":"","keywords":"Ontology; Vaccine adjuvant; Compendium; Computer science; Adjuvant; Medicine; DNA vaccination; Immune system; Immunology; Immunization","score_opus":0.03379012968567952,"score_gpt":0.33427289494916207,"score_spread":0.30048276526348255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413341034","genre_codex":"methods","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008454188,0.0059335986,0.7141721,0.01594586,0.002800804,0.0018268902,0.079233535,0.014468178,0.15716481],"genre_scores_gemma":[0.09016801,0.013307503,0.6908387,0.008611961,0.0012398652,0.002370149,0.15548833,0.003933816,0.034041762],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.996599,0.00075869064,0.00064035464,0.0005483966,0.0011110343,0.0003425415],"domain_scores_gemma":[0.99665415,0.0010458285,0.00036658155,0.00067750487,0.0009146077,0.00034133057],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030923693,0.00093869097,0.00081255933,0.005727623,0.0023114067,0.0049498947,0.0020326637,0.0024655752,0.009039899],"category_scores_gemma":[0.008029468,0.0007261141,0.002014787,0.0058193,0.0020776559,0.0107481545,0.0038882573,0.003169674,0.005040557],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008167126,0.00008875974,0.0014868295,0.0010568971,0.00006806224,0.0003575974,0.0011645783,0.0029304954,0.0025913524,0.7385876,0.13186915,0.11971698],"study_design_scores_gemma":[0.000009752066,0.000012388014,0.00039684065,0.00032113562,0.000022347049,0.00029801208,0.00027744213,0.0021054735,0.00052792707,0.074356616,0.92164946,0.000022629192],"about_ca_topic_score_codex":0.018887002,"about_ca_topic_score_gemma":0.014053475,"teacher_disagreement_score":0.018887002,"about_ca_system_score_codex":0.0030398776,"about_ca_system_score_gemma":0.010175506,"threshold_uncertainty_score":0.037554145},"labels":[],"label_agreement":null},{"id":"W4413398865","doi":"10.1038/s41597-025-05798-9","title":"Whole genome sequencing and annotations of Trametes sanguinea ZHSJ","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Insect and Arachnid Ecology and Behavior","field":"Biochemistry, Genetics and Molecular Biology","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"Natural Science Foundation of Shandong Province; National Natural Science Foundation of China","keywords":"Biology; Genome; Computational biology; DNA sequencing; Whole genome sequencing; Evolutionary biology; Genetics; Gene","score_opus":0.029939659394598944,"score_gpt":0.29708225111537623,"score_spread":0.26714259172077726,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413398865","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9016342,0.0053557577,0.010949506,0.00021708138,0.00013260962,0.00018888905,0.07521111,0.000807515,0.005503204],"genre_scores_gemma":[0.68265784,0.002530504,0.02330611,0.00021893786,0.00005317294,0.00016877774,0.2818155,0.0003090836,0.008940109],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99982953,0.000013921843,0.000020183388,0.00006018783,0.000047479683,0.000028728758],"domain_scores_gemma":[0.9998355,0.000022637721,0.00003261225,0.000021836797,0.00005807633,0.000029287175],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022824755,0.00049615203,0.0005080376,0.0014452292,0.00046873317,0.00047908118,0.0002839283,0.000276199,0.0019988518],"category_scores_gemma":[0.0002825659,0.00017204533,0.0008214637,0.0018781224,0.00016059006,0.0002950415,0.000381774,0.00038616554,0.0008315603],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00086797186,0.0000549852,0.01209779,0.0011169154,0.00008881684,0.0006338456,0.00078288594,0.00069965364,0.9481223,0.0003113128,0.0018122654,0.03341134],"study_design_scores_gemma":[0.00018208372,0.0008861071,0.7066287,0.0003485436,0.0010031649,0.0030398036,0.0017799566,0.005766009,0.113445126,0.0006925218,0.16607793,0.00015010974],"about_ca_topic_score_codex":0.0053244038,"about_ca_topic_score_gemma":0.00837921,"teacher_disagreement_score":0.0053244038,"about_ca_system_score_codex":0.0003059078,"about_ca_system_score_gemma":0.0005356026,"threshold_uncertainty_score":0.010586798},"labels":[],"label_agreement":null},{"id":"W4413450650","doi":"10.1038/s41597-025-05123-4","title":"High-resolution Canada domain disturbance forcings suitable for land surface modeling applications","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cryospheric studies and observations","field":"Earth and Planetary Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Canadian Forest Service; Natural Resources Canada; Carleton University; Environment and Climate Change Canada","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Disturbance (geology); Carbon cycle; Environmental science; Climatology; Geography; Computer science; Environmental resource management; Physical geography; Ecology; Geology; Ecosystem","score_opus":0.028931366953676554,"score_gpt":0.23325200072650848,"score_spread":0.20432063377283194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413450650","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6836301,0.00047938203,0.1525352,0.0014064349,0.0003773702,0.0004684127,0.10692187,0.014040206,0.04014093],"genre_scores_gemma":[0.89989036,0.0002453159,0.060497962,0.00014812053,0.000048615904,0.00022584693,0.035054922,0.00085188274,0.0030369894],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9998109,0.000018234074,0.00001024734,0.000050814077,0.00006410686,0.000045725887],"domain_scores_gemma":[0.9994716,0.00008091195,0.00003961955,0.00008261472,0.00025845418,0.00006678927],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000298567,0.00058591005,0.00035050424,0.00078700646,0.0010379099,0.00088176836,0.0013238589,0.0005579242,0.0042532217],"category_scores_gemma":[0.0019365416,0.00038906513,0.00080386334,0.0019657004,0.0003572086,0.00034097844,0.00047476447,0.00092922867,0.00063236465],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000069718626,0.00007877961,0.025593258,0.00006321956,0.000095588875,0.000109722656,0.000057193043,0.9372842,0.002725256,0.0033220241,0.01310996,0.017491173],"study_design_scores_gemma":[0.00007709707,0.000006893404,0.017729776,0.000019035024,0.000018070032,0.000023809012,0.000050824878,0.9712583,0.0013456027,0.00086442806,0.008563811,0.000042463063],"about_ca_topic_score_codex":0.87944037,"about_ca_topic_score_gemma":0.891192,"teacher_disagreement_score":0.12055963,"about_ca_system_score_codex":0.0045638494,"about_ca_system_score_gemma":0.008668291,"threshold_uncertainty_score":0.24253923},"labels":[],"label_agreement":null},{"id":"W4413491796","doi":"10.1038/s41597-025-05509-4","title":"Plant traits and associated ecological data from global change experiments and climate gradients in Norway","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Senter for Internasjonalisering av Utdanning; Norges Forskningsråd; Universitetet i Bergen","keywords":"Ecology; Plant functional type; Environmental science; Ecosystem; Climate change; Plant community; Plant ecology; Global change; Global warming; Environmental gradient; Physical geography; Geography; Biology; Ecological succession","score_opus":0.06794931172273355,"score_gpt":0.2938196451764213,"score_spread":0.22587033345368773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413491796","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9972574,0.00009084565,0.0003333273,0.0000047177746,0.0000047819794,0.0000075344824,0.0020422407,0.0000098403225,0.00024929803],"genre_scores_gemma":[0.9902001,0.000115058814,0.001489461,0.000025349611,0.0000041388926,0.000064939515,0.00786486,0.000011908598,0.000224209],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9995453,0.000112457885,0.00005955297,0.00014655934,0.000077170385,0.00005894589],"domain_scores_gemma":[0.9988845,0.0003766974,0.0003033534,0.00022520297,0.00012732939,0.00008285592],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011774147,0.0002454743,0.00035397915,0.0007372585,0.00028759948,0.0003461115,0.00023500223,0.00018034705,0.0004642533],"category_scores_gemma":[0.0006954902,0.00015454304,0.00034590389,0.0007902873,0.0005003156,0.00032344236,0.00039235508,0.00020421794,0.0000713979],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015582853,0.00035740755,0.819725,0.00021429948,0.0003376781,0.00034892402,0.0010861272,0.00714115,0.14814225,0.00020976728,0.0007990832,0.020080104],"study_design_scores_gemma":[0.000012755631,0.000084788655,0.99640095,0.0000053213207,0.00002263487,0.0000369702,0.00015535895,0.00050933537,0.0020181003,0.00003231445,0.00071152905,0.000009924229],"about_ca_topic_score_codex":0.038711604,"about_ca_topic_score_gemma":0.06340632,"teacher_disagreement_score":0.038711604,"about_ca_system_score_codex":0.00066344655,"about_ca_system_score_gemma":0.00031077163,"threshold_uncertainty_score":0.076972544},"labels":[],"label_agreement":null},{"id":"W4413494925","doi":"10.1038/s41597-025-05866-0","title":"Full Field Digital Mammography Dataset from a Population Screening Program","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"AI in cancer detection","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"St. John’s Health Sciences Centre; Newfoundland and Labrador Centre for Applied Health Research; Memorial University of Newfoundland","funders":"Memorial University of Newfoundland","keywords":"Mammography; Field (mathematics); Population; Digital mammography; Mammography screening; Data science; Computer science; Medical physics; Medicine; Environmental health; Mathematics; Breast cancer; Internal medicine; Cancer","score_opus":0.03314364068906861,"score_gpt":0.3164268363211404,"score_spread":0.28328319563207177,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413494925","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010187993,0.0007370886,0.0005523299,0.000393822,0.00006408681,0.00012667607,0.9846501,0.0009553557,0.0023325428],"genre_scores_gemma":[0.0087821,0.00021019588,0.0013532747,0.00011399155,0.000019181081,0.00008744587,0.9884567,0.000027089081,0.00095004647],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991418,0.00010764402,0.00007719207,0.00022041077,0.0003062191,0.00014673761],"domain_scores_gemma":[0.99858046,0.0002771758,0.00009669125,0.00027319734,0.0006077696,0.00016473526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00067363394,0.0014006017,0.0010759066,0.002720654,0.00079821906,0.0010770018,0.0024769856,0.0016580651,0.009802539],"category_scores_gemma":[0.0032143248,0.00035249942,0.0010776079,0.0038683638,0.00044502981,0.00039773335,0.0010266585,0.001047942,0.010410156],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006535833,0.0005127129,0.024682026,0.0014406623,0.00028380877,0.0005381226,0.00009261786,0.0035008562,0.0015000575,0.0005404885,0.92748874,0.03876635],"study_design_scores_gemma":[0.0007028237,0.00027670577,0.18346064,0.00069774076,0.00036155223,0.0017918021,0.00051458675,0.017009426,0.0041186498,0.002364022,0.7885066,0.00019548115],"about_ca_topic_score_codex":0.19935566,"about_ca_topic_score_gemma":0.32091007,"teacher_disagreement_score":0.19935566,"about_ca_system_score_codex":0.0024194138,"about_ca_system_score_gemma":0.0034104313,"threshold_uncertainty_score":0.39639056},"labels":[],"label_agreement":null},{"id":"W4413910953","doi":"10.1038/s41597-025-05445-3","title":"The Indo-European Cognate Relationships dataset","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Language and cultural evolution","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Lexeme; Cognate; Metadata; Computer science; Benchmark (surveying); Lexicon; Language family; Natural language processing; Interoperability; Linguistics; Artificial intelligence; Information retrieval; World Wide Web; Geography; Cartography","score_opus":0.10680086555281473,"score_gpt":0.37512981886630353,"score_spread":0.2683289533134888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413910953","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008432833,0.00041791936,0.00074828946,0.00020960814,0.00004276811,0.000054706514,0.9835434,0.0004268399,0.00612365],"genre_scores_gemma":[0.0049355924,0.00010141777,0.001490068,0.0000728086,0.000012082064,0.00013583474,0.99221814,0.00006485329,0.0009691172],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986864,0.00025033447,0.00018783423,0.00040531412,0.00030218283,0.00016799338],"domain_scores_gemma":[0.9984542,0.00046975032,0.00017987414,0.00035356128,0.00037488923,0.0001677312],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010701132,0.0009896707,0.00070868805,0.0057388702,0.0011968816,0.0018823445,0.0020432295,0.0014701855,0.016651245],"category_scores_gemma":[0.005460748,0.00025750705,0.0008415338,0.0085527,0.000555508,0.001386466,0.002899858,0.0012054008,0.015343],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033900654,0.00016756674,0.02046454,0.0023017647,0.00014485786,0.0007129987,0.00096335274,0.0013640522,0.0024339573,0.011534618,0.92777056,0.03180275],"study_design_scores_gemma":[0.000064313834,0.00002184442,0.023827722,0.00033699183,0.000045106484,0.00036217718,0.0006859989,0.0007840388,0.0009064771,0.002233851,0.9706898,0.00004174812],"about_ca_topic_score_codex":0.01140051,"about_ca_topic_score_gemma":0.019915866,"teacher_disagreement_score":0.016651245,"about_ca_system_score_codex":0.0010635488,"about_ca_system_score_gemma":0.0016995636,"threshold_uncertainty_score":0.055703938},"labels":[],"label_agreement":null},{"id":"W4414357588","doi":"10.1038/s41597-025-05962-1","title":"A dataset from a coordinated multi-site laboratory study investigating the Hue-Heat-Hypothesis","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Building Energy and Comfort Optimization","field":"Engineering","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"HORIZON EUROPE Framework Programme; Magyar Tudományos Akadémia Számítástechnikai és Automatizálási Kutatóintézet; Magyar Tudományos Akadémia; European Commission; Ministério da Ciência, Tecnologia e Inovação; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Villum Fonden; National Science Foundation","keywords":"Metadata; Documentation; Perception; Protocol (science); Test (biology); Resource (disambiguation)","score_opus":0.046110205236155985,"score_gpt":0.26668603584685574,"score_spread":0.22057583061069974,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414357588","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030688168,0.00083074643,0.00575337,0.00040031824,0.00013050227,0.0004980549,0.9545822,0.0008584973,0.006258207],"genre_scores_gemma":[0.03665942,0.0003637002,0.008759785,0.00026480743,0.000047173926,0.0018389396,0.95019186,0.00013132981,0.001743071],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9975527,0.00075570174,0.00034537265,0.00055410306,0.0005983642,0.00019377991],"domain_scores_gemma":[0.9913981,0.0032379767,0.0011390225,0.0016641621,0.0019953325,0.0005653509],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024048358,0.00084924203,0.0011688904,0.0030374685,0.00070359017,0.0013352844,0.0023696271,0.0018476194,0.011080626],"category_scores_gemma":[0.009246212,0.0003192882,0.0012528731,0.005403744,0.000434812,0.00074223516,0.00198228,0.00081526814,0.008046828],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033416578,0.0016205468,0.11572059,0.012634217,0.00094249,0.0014513456,0.0017636928,0.005828124,0.007907505,0.0036299098,0.72851944,0.116640486],"study_design_scores_gemma":[0.0006808281,0.0007462858,0.35098034,0.0015515583,0.0005097721,0.0008919771,0.0029184937,0.0047956733,0.0046612383,0.005311871,0.62667984,0.00027208126],"about_ca_topic_score_codex":0.013594479,"about_ca_topic_score_gemma":0.02921993,"teacher_disagreement_score":0.013594479,"about_ca_system_score_codex":0.0010889547,"about_ca_system_score_gemma":0.0014476699,"threshold_uncertainty_score":0.037068367},"labels":[],"label_agreement":null},{"id":"W4414418843","doi":"10.1038/s41597-025-05840-w","title":"A regional ocean database for the Coastal China Sea","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Oceanographic and Atmospheric Processes","field":"Earth and Planetary Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Dalhousie University; National Key Research and Development Program of China; National Aeronautics and Space Administration","keywords":"NetCDF; Biogeochemical cycle; Longitude; China; Relational database; Range (aeronautics); Seawater; Particulates","score_opus":0.03477501683651092,"score_gpt":0.260077505257286,"score_spread":0.2253024884207751,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414418843","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034134095,0.0003742608,0.00287848,0.00014658918,0.000078303055,0.0001580084,0.98137087,0.0030061074,0.008573924],"genre_scores_gemma":[0.0048798644,0.00026702412,0.004677469,0.000052931693,0.00001912151,0.00023325578,0.9868896,0.0002465033,0.0027341405],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99933237,0.000061011568,0.00018351684,0.0001464077,0.00019647706,0.00008028565],"domain_scores_gemma":[0.99820185,0.000117181145,0.00020030161,0.00044759808,0.00083045196,0.00020266985],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012229531,0.0010359712,0.00084126444,0.006111255,0.0005980381,0.0016179589,0.001883599,0.0006561161,0.03493086],"category_scores_gemma":[0.0027614934,0.0003403341,0.00046805892,0.012713494,0.00021310445,0.001585836,0.0018077416,0.00062955037,0.022085363],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037418745,0.00005805915,0.02068437,0.00141879,0.00012631278,0.00024707642,0.00024716553,0.0017657075,0.0025427013,0.009141075,0.8531182,0.11027646],"study_design_scores_gemma":[0.0001081682,0.000021263031,0.020836964,0.00017055508,0.000062461084,0.00010296638,0.00015729528,0.001688526,0.0011485799,0.0017866353,0.97387016,0.000046476485],"about_ca_topic_score_codex":0.043846816,"about_ca_topic_score_gemma":0.024511687,"teacher_disagreement_score":0.043846816,"about_ca_system_score_codex":0.0012260211,"about_ca_system_score_gemma":0.0064931186,"threshold_uncertainty_score":0.11685532},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414465667","doi":"10.1038/s41597-025-05428-4","title":"Quantum mechanical dataset of 836k neutral closed-shell molecules with up to 5 heavy atoms from C, N, O, F, Si, P, S, Cl, Br","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Structural Genomics Consortium; Vector Institute; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Basic Energy Sciences; Office of Science; Canada First Research Excellence Fund; University of Toronto; Argonne National Laboratory; U.S. Department of Energy; European Commission; Compute Canada; Government of Canada; Canadian Institute for Advanced Research; Laboratory Directed Research and Development","keywords":"Conformational isomerism; Quantum Monte Carlo; Molecule; Multipole expansion; Quantum; Quantum chemistry; Quantum chemical; Diffusion Monte Carlo","score_opus":0.027421716630805924,"score_gpt":0.30673603416352735,"score_spread":0.2793143175327214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414465667","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13959546,0.004625269,0.00544914,0.0009145011,0.00022250974,0.00018408059,0.8284735,0.0034099033,0.017125562],"genre_scores_gemma":[0.10108799,0.0010140913,0.007582097,0.0003393232,0.00004527788,0.0002572305,0.88685995,0.00027731198,0.002536774],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994949,0.000085586755,0.000028382552,0.00014403352,0.00017492262,0.000072167015],"domain_scores_gemma":[0.999627,0.00009501354,0.000043108696,0.00011344769,0.00007738297,0.000044044464],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00043233714,0.0016481072,0.0012416809,0.00076615735,0.00080750434,0.00076692685,0.0031940842,0.0019028326,0.013215067],"category_scores_gemma":[0.0012169477,0.00035704058,0.0011649211,0.0020753003,0.00052377017,0.0009425401,0.00085392647,0.001485057,0.0072176456],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010354163,0.0009546037,0.016882623,0.0059981244,0.0006991313,0.00092366035,0.00013610984,0.13122775,0.020027606,0.021496989,0.7540394,0.046578582],"study_design_scores_gemma":[0.0012925988,0.0006663206,0.016376954,0.0003022928,0.00022371385,0.00073677953,0.00030160343,0.20674981,0.020796537,0.02094313,0.73141414,0.00019606497],"about_ca_topic_score_codex":0.009336267,"about_ca_topic_score_gemma":0.015847733,"teacher_disagreement_score":0.013215067,"about_ca_system_score_codex":0.0010051132,"about_ca_system_score_gemma":0.0011917662,"threshold_uncertainty_score":0.044208825},"labels":[],"label_agreement":null},{"id":"W4414533842","doi":"10.1038/s41597-025-05759-2","title":"A global database of soil microbial phospholipid fatty acids and enzyme activities","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Carbon and Nitrogen Dynamics","field":"Agricultural and Biological Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Estonian Research Competency Council; Fundação para a Ciência e a Tecnologia; Natural Environment Research Council; Deutsches Zentrum für integrative Biodiversitätsforschung Halle-Jena-Leipzig; Japan Society for the Promotion of Science; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Ministério da Ciência, Tecnologia e Inovação; Generalitat Valenciana; China Postdoctoral Science Foundation; Austrian Science Fund; Eidgenössische Technische Hochschule Zürich; Academy of Finland; Eesti Teadusagentuur; Svenska Forskningsrådet Formas; Universität Kassel; Ministerstvo Školství, Mládeže a Tělovýchovy; Vetenskapsrådet; U.S. Department of Energy; National Natural Science Foundation of China; Agencia Estatal de Investigación; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Institute of Food and Agriculture; Ministerio de Ciencia, Tecnología e Innovación Productiva; Sight Research UK; Deutsche Forschungsgemeinschaft; European Commission; Division of Environmental Biology; Research Councils UK; Canadian Institute for Advanced Research; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Agence Nationale de la Recherche; Illinois Nutrient Research and Education Council; King Abdullah University of Science and Technology; Styrelsen för Internationellt Utvecklingssamarbete; Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa; Qatar Petroleum; U.S. Department of Agriculture; National Science Foundation","keywords":"Metadata; Ecosystem; Biomass (ecology); Georeference; Enzyme; Microbial ecology; Soil test","score_opus":0.023116330271086262,"score_gpt":0.25035119321967,"score_spread":0.22723486294858372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414533842","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021026807,0.0022929115,0.0047777533,0.00010060064,0.000038490038,0.00008317115,0.966062,0.0009798064,0.004638402],"genre_scores_gemma":[0.016187701,0.0015027768,0.0097947195,0.000059974096,0.000022112528,0.00016495354,0.9713652,0.00010703761,0.0007954296],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99823296,0.00017148114,0.00049216667,0.00048624328,0.0004803967,0.00013667849],"domain_scores_gemma":[0.99216616,0.0012052988,0.0017928228,0.0013752573,0.0026152157,0.00084534683],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018967729,0.0013303421,0.0018196814,0.016370468,0.00058722665,0.0022821147,0.0017639297,0.0011182922,0.007474731],"category_scores_gemma":[0.0061018807,0.0005512925,0.00076175795,0.028374167,0.00031292473,0.0024089757,0.0019536847,0.00088297686,0.007580222],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015393564,0.00054163905,0.24832228,0.013739503,0.0012108241,0.0013173841,0.0012657191,0.007150723,0.046398167,0.0067987926,0.23112012,0.44059545],"study_design_scores_gemma":[0.00013158911,0.0001743314,0.25599796,0.0009162074,0.0004862851,0.000851615,0.0006210293,0.0022527967,0.008425954,0.0033288077,0.72666615,0.00014727264],"about_ca_topic_score_codex":0.0071657836,"about_ca_topic_score_gemma":0.0077982154,"teacher_disagreement_score":0.016370468,"about_ca_system_score_codex":0.00070640055,"about_ca_system_score_gemma":0.003169359,"threshold_uncertainty_score":0.02500546},"labels":[],"label_agreement":null},{"id":"W4414606635","doi":"10.1038/s41597-025-05869-x","title":"Description of the monthly spatial dynamics of commercial and bycatch demersal species in the Bay of Biscay","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and fisheries research","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Fisheries and Oceans Canada; Memorial University of Newfoundland","funders":"Institut Français de Recherche pour l'Exploitation de la Mer","keywords":"Demersal zone; Bycatch; Bay; Fishing; Demersal fish; Cuckoo; Key (lock); Carpentaria","score_opus":0.03688700128157619,"score_gpt":0.26206275582751365,"score_spread":0.22517575454593747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414606635","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93444926,0.00036348397,0.0013885213,0.00014377365,0.000014761827,0.00003282002,0.060161687,0.0002292222,0.0032164038],"genre_scores_gemma":[0.9565078,0.00030334038,0.0035149883,0.000027240505,0.000013767398,0.000057446832,0.038084295,0.000029817946,0.0014611715],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9998869,0.000009978001,0.0000106356565,0.00004505278,0.000026253429,0.000021212736],"domain_scores_gemma":[0.9995695,0.00005549716,0.00012621685,0.00003590493,0.00013483244,0.00007801881],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022283218,0.00023171071,0.00016146818,0.00205245,0.0002542686,0.0005669535,0.00024104587,0.00020788687,0.0021953564],"category_scores_gemma":[0.0006978226,0.00012341082,0.0002019418,0.0022617339,0.00018764171,0.00028556233,0.0003917707,0.00022503927,0.00032870928],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000090630085,0.000049364557,0.95080495,0.00014165613,0.00010866549,0.00011711137,0.0006561025,0.016272154,0.0018810565,0.0006337976,0.0061116535,0.023132777],"study_design_scores_gemma":[0.000002647313,0.0000068926674,0.98921853,0.000012906115,0.000006157072,0.00002434173,0.00017632353,0.0064920727,0.00011675009,0.000078330144,0.0038545795,0.000010422049],"about_ca_topic_score_codex":0.2655698,"about_ca_topic_score_gemma":0.34276766,"teacher_disagreement_score":0.2655698,"about_ca_system_score_codex":0.00088504725,"about_ca_system_score_gemma":0.0005514459,"threshold_uncertainty_score":0.52804804},"labels":[],"label_agreement":null},{"id":"W4414664181","doi":"10.1038/s41597-025-05877-x","title":"Soil carbon stock densities in mangrove and forested wetland ecosystems of Panama","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Coastal wetland ecosystem dynamics","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Horizon 2020 Framework Programme; Smithsonian Tropical Research Institute; Secretaria Nacional de Ciencia y Tecnología; European Commission; Inter-American Development Bank; Ministry of Environment; Department for Environment, Food and Rural Affairs, UK Government; Smithsonian Institution","keywords":"Wetland; Soil carbon; Mangrove; Ecosystem; Peat; Blue carbon; Histosol; Carbon stock; Soil organic matter; Carbon sequestration","score_opus":0.013690570068821151,"score_gpt":0.22979700522102864,"score_spread":0.21610643515220748,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414664181","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99945086,0.00003417476,0.000055434244,0.0000059088534,3.6524736e-7,0.0000022042657,0.00022055613,0.000002925926,0.00022763244],"genre_scores_gemma":[0.9994355,0.000032911204,0.0001390864,0.0000033432968,8.996363e-7,0.000006564725,0.00028548992,9.947397e-7,0.00009502379],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99991405,0.000015157015,0.000008800236,0.000029716239,0.000015144229,0.000017083701],"domain_scores_gemma":[0.9996506,0.000055238957,0.00014155029,0.00003557449,0.000062438514,0.000054614102],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019750884,0.00016959014,0.00010633159,0.0009866736,0.0002753211,0.00037747648,0.00016631624,0.00011709942,0.0007275531],"category_scores_gemma":[0.0005969893,0.000116689676,0.00008856223,0.00090786186,0.00021460703,0.00043273685,0.00045194133,0.00010617099,0.00006368966],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000035190522,0.000013300745,0.9922031,0.0000085765005,0.000028305029,0.000041358056,0.00039128072,0.00012826879,0.0013635162,0.000036420155,0.000056121033,0.0056945374],"study_design_scores_gemma":[7.5717236e-7,0.0000064735646,0.9993773,0.0000012762262,0.000004426222,0.000026512342,0.00017356759,0.00019930048,0.000099980985,0.000013137234,0.0000961406,0.0000010746041],"about_ca_topic_score_codex":0.037171166,"about_ca_topic_score_gemma":0.07536587,"teacher_disagreement_score":0.037171166,"about_ca_system_score_codex":0.00040055456,"about_ca_system_score_gemma":0.00014973899,"threshold_uncertainty_score":0.07390958},"labels":[],"label_agreement":null},{"id":"W4414786104","doi":"10.1038/s41597-026-07582-9","title":"Persistent hindrances to data re-use in single-cell genomics","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada; National Institutes of Health","keywords":"Raw data; Genomics; USable; Data sharing; Transcriptome; Functional genomics","score_opus":0.0914354641283294,"score_gpt":0.2841214691485495,"score_spread":0.19268600502022012,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414786104","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22028284,0.0163552,0.60822123,0.051939923,0.0057088207,0.003981093,0.023696873,0.0423182,0.027495816],"genre_scores_gemma":[0.34370884,0.004300794,0.55538076,0.018209528,0.0016154347,0.0045652706,0.033553928,0.028286109,0.010379316],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.63937515,0.1754138,0.065464154,0.031301226,0.082911566,0.0055341],"domain_scores_gemma":[0.12553337,0.41386455,0.028454848,0.34162,0.085265234,0.0052620308],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.41316515,0.0018406712,0.0028628143,0.015714295,0.005432634,0.018422838,0.0112270685,0.0032265163,0.0041593215],"category_scores_gemma":[0.63208723,0.0026703733,0.003480537,0.016275238,0.008855316,0.015845384,0.02142892,0.007495039,0.005354253],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022119232,0.0007065687,0.0933141,0.012486689,0.0023891313,0.004404759,0.07170833,0.0063698194,0.04825684,0.040850118,0.11325292,0.6040489],"study_design_scores_gemma":[0.00020957892,0.0004695346,0.037708707,0.008065603,0.00090794335,0.0042193704,0.019147625,0.00993143,0.054775637,0.078888364,0.7849085,0.0007676859],"about_ca_topic_score_codex":0.006833476,"about_ca_topic_score_gemma":0.008406269,"teacher_disagreement_score":0.58683485,"about_ca_system_score_codex":0.0058881426,"about_ca_system_score_gemma":0.017624259,"threshold_uncertainty_score":0.72367185},"labels":[],"label_agreement":null},{"id":"W4414964998","doi":"10.1038/s41597-025-06046-w","title":"Aerodynamic dataset for selected doubly curved membrane canopy structures","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Wind and Air Flow Studies","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Western University","funders":"HORIZON EUROPE Framework Programme; Horizon 2020 Framework Programme; European Commission","keywords":"Aerodynamics; Wind tunnel; Ridge; Flow (mathematics); Scale (ratio); Arch; Boundary (topology); Canopy; Conical surface","score_opus":0.02408599901505776,"score_gpt":0.28398071527915625,"score_spread":0.2598947162640985,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414964998","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03667324,0.00043383572,0.0030885749,0.00017307556,0.00010771347,0.00010627659,0.9518437,0.0033056664,0.0042679603],"genre_scores_gemma":[0.031439688,0.00014428153,0.0035905386,0.000044838074,0.0000260524,0.00019812772,0.9635894,0.00013629955,0.00083075376],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992749,0.0000718475,0.00007448853,0.00017997326,0.0002865725,0.00011225016],"domain_scores_gemma":[0.99908864,0.00019603033,0.000094185285,0.00024501994,0.00029694868,0.000079194724],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005328412,0.0012677268,0.0009775809,0.001965795,0.0006170289,0.0008754585,0.0014920556,0.0018499232,0.008168988],"category_scores_gemma":[0.0015631634,0.00032119697,0.001168659,0.0019636902,0.0003390903,0.00071676774,0.0010414653,0.00096632104,0.0119534535],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044101547,0.00067588425,0.06910556,0.0020898976,0.00024883702,0.00088371744,0.00034141963,0.024562648,0.0099779675,0.00197785,0.8149453,0.07475003],"study_design_scores_gemma":[0.00029856936,0.0003482543,0.22591762,0.0005582294,0.000108761196,0.00088280934,0.0010606636,0.056344043,0.008293258,0.0029000842,0.7030629,0.00022490627],"about_ca_topic_score_codex":0.0095346,"about_ca_topic_score_gemma":0.027098631,"teacher_disagreement_score":0.0095346,"about_ca_system_score_codex":0.00048314733,"about_ca_system_score_gemma":0.0006400589,"threshold_uncertainty_score":0.027328014},"labels":[],"label_agreement":null},{"id":"W4415040514","doi":"10.1038/s41597-025-05924-7","title":"CanCPLD: Convective Parameters and Lightning Data to Support Future Thunderstorm Projections in North America","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Environment and Climate Change Canada","funders":"Environment and Climate Change Canada","keywords":"Thunderstorm; Lightning (connector); Climate change; Climate model; Nowcasting; Convection; Severe weather; Statistical model","score_opus":0.03515086346564959,"score_gpt":0.28131441501360893,"score_spread":0.24616355154795933,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415040514","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.32386088,0.0001845745,0.001555156,0.00021049938,0.000047565263,0.00007217034,0.66803914,0.001678988,0.0043510585],"genre_scores_gemma":[0.30206662,0.00016157587,0.0054731434,0.000058938742,0.000030324947,0.00014997578,0.6907943,0.00019308504,0.0010720418],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99985254,0.000021170657,0.000020764546,0.00004388541,0.00004078557,0.000020904386],"domain_scores_gemma":[0.9989889,0.00011521109,0.00014595682,0.00014850067,0.00043349815,0.0001678668],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004435902,0.00040018873,0.00026552303,0.0013704227,0.0003045259,0.00049561675,0.00060518534,0.00040092968,0.0030788921],"category_scores_gemma":[0.0011343354,0.00021514892,0.0002695678,0.0017926941,0.0001048027,0.0005458476,0.000378893,0.0003446399,0.00085485214],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00049858977,0.00043686904,0.5578411,0.0008294382,0.0003405654,0.0005302002,0.0006049244,0.1024097,0.007726781,0.0011300741,0.2669965,0.06065536],"study_design_scores_gemma":[0.00021740592,0.00007190005,0.7511816,0.00013781231,0.00008378452,0.00010542923,0.00060640683,0.12286292,0.004535125,0.0007269617,0.119374245,0.00009640008],"about_ca_topic_score_codex":0.10352602,"about_ca_topic_score_gemma":0.12534954,"teacher_disagreement_score":0.896474,"about_ca_system_score_codex":0.0006760406,"about_ca_system_score_gemma":0.0013679373,"threshold_uncertainty_score":0.20584685},"labels":[],"label_agreement":null},{"id":"W4415265664","doi":"10.1038/s41597-025-05854-4","title":"Lung Ultrasound Imaging Dataset for Accurate Detection and Localization of LUS Vertical Artifact","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ultrasound in Clinical Applications","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Makerere University; International Development Research Centre","keywords":"Artifact (error); Lung ultrasound; Ultrasound; Pattern recognition (psychology); Economic shortage; Deep learning","score_opus":0.05532807895029387,"score_gpt":0.3994944132357846,"score_spread":0.34416633428549076,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415265664","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1499577,0.007801019,0.03248911,0.002136888,0.00071026443,0.0013889861,0.78233117,0.013092912,0.010091944],"genre_scores_gemma":[0.0982656,0.0013592239,0.033199858,0.00047031525,0.00014709172,0.0008025679,0.862128,0.0003997944,0.0032274532],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993494,0.00008687383,0.00008023635,0.00016367076,0.00021709337,0.00010268074],"domain_scores_gemma":[0.9992231,0.00018476124,0.00007399664,0.00014847839,0.000293097,0.000076508375],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007618705,0.0013708574,0.0008976878,0.0022113863,0.00057645736,0.0008435702,0.0014374199,0.0017203024,0.0052095777],"category_scores_gemma":[0.0021419975,0.0003356471,0.0009942345,0.0012785048,0.0004035362,0.00036746272,0.0012283495,0.0010279146,0.005761828],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016228413,0.0012843629,0.040862095,0.0053221276,0.0005682821,0.0041971905,0.00039891465,0.012454542,0.061026115,0.0014606366,0.61409545,0.25670752],"study_design_scores_gemma":[0.0008100514,0.00090936443,0.17852126,0.0016590342,0.0005778743,0.014270732,0.0011208167,0.084415264,0.07145554,0.003935611,0.6419476,0.00037684993],"about_ca_topic_score_codex":0.010681088,"about_ca_topic_score_gemma":0.027443785,"teacher_disagreement_score":0.010681088,"about_ca_system_score_codex":0.00067996664,"about_ca_system_score_gemma":0.0012632441,"threshold_uncertainty_score":0.02123785},"labels":[],"label_agreement":null},{"id":"W4415648821","doi":"10.1038/s41597-025-05976-9","title":"A global database on land use and management change effects on soil KMnO4-oxidisable organic carbon (POXC)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Soil Carbon and Nitrogen Dynamics","field":"Agricultural and Biological Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Smiths Detection (Canada)","funders":"Agence Nationale de la Recherche; European Commission","keywords":"Soil carbon; Arable land; Land use; Land use, land-use change and forestry; Soil health; Land management; Soil management; Soil organic matter; Soil quality","score_opus":0.04532416565877685,"score_gpt":0.25302777723991243,"score_spread":0.2077036115811356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415648821","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009517093,0.00068959314,0.0002278715,0.00005503846,0.000018072484,0.000015512884,0.9970534,0.00015033662,0.0008385135],"genre_scores_gemma":[0.0047094305,0.0012180865,0.0013764246,0.00009177704,0.000021373413,0.00015579376,0.99184895,0.00009115832,0.00048699093],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99853134,0.00019674616,0.00044014025,0.00038553917,0.0003231975,0.00012295679],"domain_scores_gemma":[0.99092454,0.0030095342,0.002466867,0.0009432837,0.0020952825,0.00056057936],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001717043,0.0011864812,0.0015744136,0.014096261,0.00032824016,0.0016771987,0.0013682663,0.0012703738,0.017918108],"category_scores_gemma":[0.008271328,0.0005311964,0.0011196652,0.025311846,0.00030500765,0.0014950352,0.0019046275,0.00080641237,0.00964347],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006719816,0.00012859704,0.05612679,0.041462213,0.0014466256,0.0004728139,0.00064424955,0.0028009438,0.0062294174,0.006486845,0.79681516,0.08671447],"study_design_scores_gemma":[0.00013834302,0.000045881032,0.06724425,0.0028129632,0.00035770627,0.00016827892,0.0002848711,0.0003628002,0.00119577,0.0017814226,0.92554176,0.00006600243],"about_ca_topic_score_codex":0.012249904,"about_ca_topic_score_gemma":0.014419939,"teacher_disagreement_score":0.017918108,"about_ca_system_score_codex":0.00088778127,"about_ca_system_score_gemma":0.0024461974,"threshold_uncertainty_score":0.059942067},"labels":[],"label_agreement":null},{"id":"W4415657680","doi":"10.1038/s41597-026-07496-6","title":"Characterization of high-resolution AI data center training workloads on single and multiple GPU nodes","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cloud Computing and Resource Management","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; Independent Electricity System Operator","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Session (web analytics); Resilience (materials science); Grid; Data center; Set (abstract data type); Energy consumption; Data set; Training set; Training (meteorology)","score_opus":0.06182922118262564,"score_gpt":0.2691192533714195,"score_spread":0.20729003218879388,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415657680","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9652938,0.00032764208,0.008785044,0.0003529561,0.00018609255,0.00014046868,0.018721236,0.0020569386,0.0041359398],"genre_scores_gemma":[0.9480125,0.0001412549,0.006243196,0.000121778976,0.00006496929,0.00010323445,0.04392627,0.000121357654,0.0012653684],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9995401,0.00005046684,0.000027937123,0.00014747946,0.00011995851,0.0001141519],"domain_scores_gemma":[0.99925345,0.00014674336,0.000054855223,0.00016803622,0.00025208795,0.00012479842],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039860766,0.00060275587,0.0003994984,0.00078443455,0.00045917608,0.0004595676,0.0007201047,0.0004643894,0.000930032],"category_scores_gemma":[0.0017762942,0.00017134087,0.0002825263,0.00093004847,0.00031545013,0.0005128989,0.00058012625,0.00081922684,0.0006050605],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030181324,0.003926326,0.34866315,0.00086129474,0.00043464382,0.0012864071,0.0013315546,0.2086498,0.08714094,0.0030761634,0.1502704,0.19134124],"study_design_scores_gemma":[0.00014001851,0.00070297,0.43110883,0.000069697184,0.00007059721,0.0006856957,0.0009127672,0.51226103,0.027289264,0.0019933227,0.024677247,0.00008863407],"about_ca_topic_score_codex":0.007943722,"about_ca_topic_score_gemma":0.016883226,"teacher_disagreement_score":0.007943722,"about_ca_system_score_codex":0.0004623664,"about_ca_system_score_gemma":0.00050941744,"threshold_uncertainty_score":0.015794992},"labels":[],"label_agreement":null},{"id":"W4415702507","doi":"10.1038/s41597-025-06212-0","title":"Author Correction: ROBIN: Reference observatory of basins for international hydrological climate change detection","year":2025,"lang":"en","type":"erratum","venue":"Scientific Data","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Climate change; Change detection; Observatory; Drainage basin; Global change; General Circulation Model","score_opus":0.10853584463885262,"score_gpt":0.30733881501489174,"score_spread":0.1988029703760391,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415702507","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006118034,0.0007809883,0.002607056,0.065240175,0.9080748,0.00003577443,0.004505407,0.0012511696,0.01689295],"genre_scores_gemma":[0.028612118,0.0047196355,0.015394389,0.061202653,0.1411329,0.00016454083,0.014026625,0.008558418,0.72618884],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.996366,0.00044561917,0.0005592537,0.00059529097,0.0018324128,0.00020154225],"domain_scores_gemma":[0.9758247,0.0036929382,0.0010364003,0.0015976392,0.017190097,0.00065832015],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024425723,0.0014500681,0.0008900987,0.0027177187,0.0030102276,0.0032930642,0.0017517203,0.0028171944,0.06201314],"category_scores_gemma":[0.04252499,0.0006346612,0.000633174,0.0026393386,0.0013525718,0.0020982688,0.0016003477,0.006685532,0.04285679],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012847368,0.0000024842493,0.000068817884,0.000029908055,0.0000024542535,0.00009293796,0.0000308422,0.000030005247,0.000048123562,0.00075808953,0.99450725,0.004416256],"study_design_scores_gemma":[0.00000900506,0.000005641341,0.00045873472,0.00012477612,0.000008473996,0.0002521006,0.00008498248,0.00015271017,0.0003847026,0.00069639384,0.99780566,0.000016843233],"about_ca_topic_score_codex":0.036855627,"about_ca_topic_score_gemma":0.043562487,"teacher_disagreement_score":0.06201314,"about_ca_system_score_codex":0.002537472,"about_ca_system_score_gemma":0.005064754,"threshold_uncertainty_score":0.20745462},"labels":[],"label_agreement":null},{"id":"W4415707275","doi":"10.1038/s41597-025-06005-5","title":"A Video Dataset for Nearshore Wave Breaking Type Classification","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ocean Waves and Remote Sensing","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"National Key Research and Development Program of China; National Natural Science Foundation of China","keywords":"Breaking wave; Workflow; Breaking strength; Limiting; Dissipation; Frame (networking); Wave height","score_opus":0.11903734315670375,"score_gpt":0.315366750120919,"score_spread":0.19632940696421528,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415707275","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02896604,0.0015038416,0.006920746,0.00040084278,0.0007135611,0.0007351777,0.94856113,0.005358238,0.006840417],"genre_scores_gemma":[0.013031264,0.00026435274,0.0077101546,0.00008280332,0.000058764523,0.00034713733,0.9764846,0.00010102388,0.0019198956],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990496,0.00007415488,0.00009202445,0.0002764919,0.0003447133,0.00016299245],"domain_scores_gemma":[0.99901533,0.0001107917,0.00009179997,0.00024119504,0.0004208771,0.00012000775],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00053738547,0.0021805584,0.0009448508,0.0028395634,0.00091804663,0.000895123,0.002165701,0.0016308732,0.00949149],"category_scores_gemma":[0.0018816199,0.00025563277,0.0009997693,0.0029275934,0.00044300113,0.0009859717,0.0012925632,0.0015788873,0.011431803],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007042653,0.00086590304,0.0066641467,0.0018489597,0.00017394722,0.00082926167,0.0001611413,0.0042681056,0.016725812,0.0012847176,0.8178224,0.14865133],"study_design_scores_gemma":[0.0005003837,0.0008081585,0.09404242,0.0011218581,0.000262689,0.002254141,0.0011415578,0.06448017,0.03295931,0.003587564,0.7985141,0.00032765008],"about_ca_topic_score_codex":0.02725047,"about_ca_topic_score_gemma":0.062596515,"teacher_disagreement_score":0.02725047,"about_ca_system_score_codex":0.0011957068,"about_ca_system_score_gemma":0.0014862615,"threshold_uncertainty_score":0.05418372},"labels":[],"label_agreement":null},{"id":"W4416143213","doi":"10.1038/s41597-025-06045-x","title":"Plant traits and associated ecological data from Afromontane grasslands of Maloti-Drakensberg, South Africa","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Senter for Internasjonalisering av Utdanning; Universitetet i Bergen","keywords":"Trait; Ecosystem; Vegetation (pathology); Vascular plant; Taxon; Climate change; Plant community; Biodiversity; Terrestrial plant","score_opus":0.04068062986498387,"score_gpt":0.23800735297200298,"score_spread":0.1973267231070191,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416143213","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9984676,0.00010122516,0.00005636877,0.000007902042,5.8348434e-7,0.0000127974445,0.0011048345,0.0000016684703,0.00024700575],"genre_scores_gemma":[0.99635255,0.00014327497,0.0005952695,0.00001784284,0.0000022063405,0.000058314028,0.0026037227,0.0000030195313,0.00022375253],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9996983,0.00004663495,0.000028357805,0.00009155041,0.000038865463,0.00009630842],"domain_scores_gemma":[0.9995117,0.000096393625,0.00018803323,0.00007101068,0.000065282184,0.00006753598],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00027411027,0.00035582852,0.00024044725,0.0022849045,0.00058301305,0.00035325988,0.00027529593,0.00021637729,0.00093056966],"category_scores_gemma":[0.0006408793,0.00016669458,0.0001583619,0.0022835375,0.0004965478,0.0002639201,0.00067101495,0.00014266156,0.00019460966],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020913065,0.00008188362,0.9493506,0.0001839511,0.00008244466,0.0006258743,0.0059202113,0.00028145683,0.025207939,0.00013352947,0.00038240073,0.017540503],"study_design_scores_gemma":[0.000004227592,0.000015158645,0.9980627,0.000011383603,0.0000085035535,0.000116620686,0.00083144114,0.00005609087,0.00018076265,0.000007712432,0.0007022893,0.000003171851],"about_ca_topic_score_codex":0.069881506,"about_ca_topic_score_gemma":0.21750936,"teacher_disagreement_score":0.069881506,"about_ca_system_score_codex":0.00065602886,"about_ca_system_score_gemma":0.00039966946,"threshold_uncertainty_score":0.13894951},"labels":[],"label_agreement":null},{"id":"W4416386449","doi":"10.1038/s41597-025-06115-0","title":"Mouse Hippocampal Sharp-Wave Ripple Dataset Curated From Public Neuropixels Datasets","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Neural dynamics and brain function","field":"Neuroscience","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"CIHR Skin Research Training Centre; Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Hippocampal formation; Task (project management); Hippocampus; Field (mathematics); Raw data; Electrophysiology; Process (computing)","score_opus":0.14175337180313524,"score_gpt":0.3148659871587131,"score_spread":0.17311261535557787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416386449","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006424546,0.00053405226,0.0036899152,0.00011043843,0.00007754087,0.00007170667,0.97978437,0.008293653,0.0010137454],"genre_scores_gemma":[0.0034834445,0.00017523176,0.0037262398,0.00006439626,0.000012580689,0.0002405075,0.99143726,0.00048805107,0.0003723153],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999151,0.00007922521,0.0001471727,0.00026744884,0.0002516183,0.00010349414],"domain_scores_gemma":[0.99845743,0.00027938726,0.00028523747,0.00045890154,0.00033306231,0.00018611169],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012957539,0.0019973323,0.0014051012,0.0042931437,0.0007681557,0.0011921722,0.0023234289,0.0014013969,0.00826182],"category_scores_gemma":[0.0034797778,0.00046633836,0.0016180004,0.0033566214,0.0004560624,0.00097526214,0.002027011,0.0013743928,0.011285121],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014847369,0.0002518927,0.016682934,0.0053639445,0.0007511339,0.0008052206,0.0003102054,0.0042262385,0.037601855,0.0028450496,0.8971753,0.032501463],"study_design_scores_gemma":[0.0007449885,0.0004038587,0.07114983,0.0009086344,0.0007304461,0.0017047152,0.00024583543,0.008233161,0.031829257,0.0065082004,0.87723076,0.00031030976],"about_ca_topic_score_codex":0.004850112,"about_ca_topic_score_gemma":0.010783265,"teacher_disagreement_score":0.00826182,"about_ca_system_score_codex":0.000643774,"about_ca_system_score_gemma":0.0017678965,"threshold_uncertainty_score":0.027638555},"labels":[],"label_agreement":null},{"id":"W4416386476","doi":"10.1038/s41597-025-06284-y","title":"14,441 Genomics-Based Validated Automated Comprehensive Clinicopathologic Correlations for Myeloid Neoplasms","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Chronic Myeloid Leukemia Treatments","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Nova Scotia Health Authority; Spinologics (Canada)","funders":"","keywords":"Myeloid; Myeloid leukemia; Dysplasia; Peripheral blood; Myeloid sarcoma; Blood cancer","score_opus":0.08826511586548723,"score_gpt":0.3756117539573349,"score_spread":0.2873466380918477,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416386476","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1401324,0.0037518814,0.004639525,0.00036873764,0.000111887915,0.0002832583,0.84516746,0.001621179,0.003923712],"genre_scores_gemma":[0.057899892,0.00047636713,0.003764479,0.00012954343,0.000045166205,0.0002705079,0.9368417,0.000053220167,0.00051913405],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9983211,0.00026049477,0.0003214335,0.000585945,0.00035146324,0.00015951508],"domain_scores_gemma":[0.9970294,0.0010501053,0.0004940152,0.0005016168,0.00068390847,0.00024093493],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017198331,0.0009930785,0.000930997,0.0046468102,0.00042887474,0.0011043309,0.0011381909,0.0009393485,0.0056211906],"category_scores_gemma":[0.0065311785,0.00031071092,0.00080822565,0.0030810237,0.00025150145,0.0004971775,0.0014110489,0.000518424,0.0043783793],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017463014,0.00047823522,0.45047528,0.004067867,0.0009745397,0.0030199636,0.00030496944,0.0056063966,0.0121316975,0.0013340535,0.354986,0.16487476],"study_design_scores_gemma":[0.0009747091,0.0004611985,0.5587831,0.0010322034,0.0009338718,0.006479561,0.00043673706,0.018051177,0.012532818,0.0023852235,0.3977902,0.00013920123],"about_ca_topic_score_codex":0.005217331,"about_ca_topic_score_gemma":0.008485175,"teacher_disagreement_score":0.0056211906,"about_ca_system_score_codex":0.0007886825,"about_ca_system_score_gemma":0.0018861478,"threshold_uncertainty_score":0.018804789},"labels":[],"label_agreement":null},{"id":"W4416389477","doi":"10.1038/s41597-025-06130-1","title":"Global dataset for realized thermal and aridity niche limits for terrestrial vertebrates","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"Natural Sciences and Engineering Research Council of Canada; University of Ottawa","keywords":"Niche; Arid; Climate change; Range (aeronautics); Macroecology; Extinction (optical mineralogy); Ecological niche; Population; Environmental change","score_opus":0.09821407305433746,"score_gpt":0.3473951902289647,"score_spread":0.24918111717462724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416389477","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025225277,0.000108590575,0.00030526545,0.00006511052,0.0000141009195,0.00001266548,0.99574053,0.000192516,0.0010386123],"genre_scores_gemma":[0.0050000297,0.000075216514,0.0012335507,0.000056046694,0.000009620627,0.00010513647,0.9931063,0.000045722172,0.0003683635],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994492,0.00006627238,0.00010346739,0.00019196457,0.00012263129,0.000066518696],"domain_scores_gemma":[0.9979504,0.0004125055,0.000620337,0.00030939668,0.00050996756,0.00019743988],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00079550897,0.0008264517,0.00071252784,0.0027125257,0.0004322858,0.00090111356,0.0012447531,0.0010220113,0.015761709],"category_scores_gemma":[0.003811806,0.00038039894,0.00072255125,0.0050488645,0.00023412176,0.00080661644,0.0013286948,0.0011053107,0.0111476295],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020427523,0.00008932759,0.082843155,0.0019585143,0.00033665693,0.0001607159,0.00030219706,0.0030185895,0.0017619109,0.0025131258,0.8859476,0.020863928],"study_design_scores_gemma":[0.00019180622,0.000039663933,0.18661927,0.00045806752,0.000097838325,0.00027815558,0.00031059008,0.002355227,0.00074800616,0.0023559472,0.8064697,0.00007570634],"about_ca_topic_score_codex":0.015348008,"about_ca_topic_score_gemma":0.03046087,"teacher_disagreement_score":0.015761709,"about_ca_system_score_codex":0.0006305204,"about_ca_system_score_gemma":0.00096349086,"threshold_uncertainty_score":0.052728176},"labels":[],"label_agreement":null},{"id":"W4416389566","doi":"10.1038/s41597-025-06106-1","title":"A dataset on metal-related production activities and their socio-environmental impacts in Canada","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Extraction and Separation Processes","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université du Québec à Montréal; Polytechnique Montréal","funders":"Haute école Spécialisée de Suisse Occidentale; Environment and Climate Change Canada; Government of Canada; École Polytechnique Fédérale de Lausanne; Polytechnique Montréal; Université du Québec à Montréal","keywords":"Traceability; Production (economics); Sustainability; Climate change; Environmental impact assessment; Ecological footprint; Code (set theory)","score_opus":0.015998062201415225,"score_gpt":0.2355920694080879,"score_spread":0.2195940072066727,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416389566","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006978278,0.0002499905,0.00021305736,0.0001242391,0.000008735234,0.000026892767,0.99042726,0.00012031623,0.0018511628],"genre_scores_gemma":[0.017413411,0.0003307078,0.00093290466,0.000050053837,0.000004624747,0.00006470039,0.9796928,0.00002460992,0.0014861209],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990086,0.000044257435,0.00009897547,0.00020588422,0.00040841155,0.00023393637],"domain_scores_gemma":[0.9969311,0.00035496918,0.00022402404,0.00020996477,0.001957829,0.0003221138],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041070464,0.0008467097,0.00060557853,0.0067452327,0.0016057487,0.0018379514,0.0015554626,0.00073853415,0.006953334],"category_scores_gemma":[0.0032716603,0.00035598697,0.00084854395,0.02031944,0.0005320464,0.000578286,0.0011495141,0.00075915747,0.002516439],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036158724,0.00016454235,0.12857282,0.0024239048,0.00035546298,0.0005873051,0.00079388527,0.010640749,0.0017028524,0.005842014,0.8004294,0.048125543],"study_design_scores_gemma":[0.000079082405,0.000021410862,0.24375829,0.0006902623,0.00011608253,0.0001736232,0.0014445224,0.0049931905,0.0014400308,0.0011731717,0.74599683,0.00011349151],"about_ca_topic_score_codex":0.98224497,"about_ca_topic_score_gemma":0.98837984,"teacher_disagreement_score":0.020167224,"about_ca_system_score_codex":0.020167224,"about_ca_system_score_gemma":0.035354532,"threshold_uncertainty_score":0.14632416},"labels":[],"label_agreement":null},{"id":"W4416414271","doi":"10.1038/s41597-025-06068-4","title":"The ICGC ARGO data dictionary for standardizing global cancer clinical data","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Indoc Research; University of Toronto; Ontario Institute for Cancer Research","funders":"Government of Ontario","keywords":"Interoperability; Cancer; Clinical Oncology; Argo; Precision oncology; Germline; Genome; Genomics; Metadata","score_opus":0.12173710700112084,"score_gpt":0.456465764786295,"score_spread":0.3347286577851742,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416414271","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reporting","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":"reporting","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004166595,0.0016100727,0.35367587,0.0073815766,0.0017790422,0.0030572058,0.57026786,0.015917724,0.04214407],"genre_scores_gemma":[0.014951313,0.0015125474,0.35806328,0.003718401,0.0002892358,0.0040530073,0.6079631,0.004339768,0.0051093725],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9854387,0.0040239957,0.005821689,0.001552953,0.0026410827,0.00052167167],"domain_scores_gemma":[0.94412583,0.016813438,0.00455374,0.017901618,0.014289439,0.0023159853],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.018159436,0.0010381725,0.0017164749,0.009949848,0.0017215795,0.0071187853,0.0038793779,0.0025049248,0.013846937],"category_scores_gemma":[0.062404837,0.0010604872,0.0015287953,0.015747601,0.0017607581,0.0051823473,0.006140187,0.0045223827,0.013872237],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005972551,0.00012640869,0.0070877415,0.002979337,0.00011702684,0.00039005635,0.0016990621,0.006233294,0.003980491,0.23288724,0.6083346,0.1355676],"study_design_scores_gemma":[0.0000627918,0.000025414425,0.0012551859,0.00088300474,0.000027051141,0.00014690665,0.0002575176,0.0015638749,0.001245759,0.014652746,0.979829,0.0000508137],"about_ca_topic_score_codex":0.027102321,"about_ca_topic_score_gemma":0.020569814,"teacher_disagreement_score":0.98184055,"about_ca_system_score_codex":0.005093812,"about_ca_system_score_gemma":0.020442758,"threshold_uncertainty_score":0.09603739},"labels":[],"label_agreement":null},{"id":"W4416475311","doi":"10.1038/s41597-025-06286-w","title":"AmphiTherm: a comprehensive database of amphibian thermal tolerance and preference","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Amphibian and Reptile Biology","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; Environment and Climate Change Canada","funders":"Fondo Nacional de Desarrollo Científico y Tecnológico; University of New South Wales; Department of Education and Training; Deutsche Forschungsgemeinschaft; Wenner-Gren Foundation","keywords":"Ectotherm; Amphibian; Biodiversity; Preference; Trait; Resource (disambiguation)","score_opus":0.04953360695434261,"score_gpt":0.2723470102566372,"score_spread":0.2228134033022946,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416475311","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.033233058,0.028167307,0.0064165415,0.00035526548,0.00008951378,0.00022335495,0.9195539,0.0018924025,0.0100687165],"genre_scores_gemma":[0.053034835,0.01683551,0.033363078,0.0004901666,0.00011863446,0.0010403552,0.8923808,0.0006943537,0.0020423008],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99874604,0.00025584747,0.0004221427,0.00027471734,0.00023366822,0.00006752042],"domain_scores_gemma":[0.9942967,0.0023192859,0.0016941061,0.0005053901,0.0008776432,0.00030682064],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013069466,0.00077052513,0.0011052575,0.013094328,0.0005152131,0.0012214588,0.0009591993,0.0006565452,0.010628898],"category_scores_gemma":[0.008214886,0.000438334,0.0009641911,0.010593199,0.00034122332,0.0021410864,0.0020960364,0.00077925314,0.003959172],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012894187,0.00016437037,0.13024189,0.09119473,0.00184663,0.0012196107,0.0040334817,0.0024753367,0.019372866,0.00616346,0.28790775,0.4540904],"study_design_scores_gemma":[0.00007413968,0.00009522067,0.14276421,0.0055351895,0.00089703844,0.0010518313,0.0004927557,0.0005529483,0.0015643592,0.0019126458,0.8449467,0.00011296481],"about_ca_topic_score_codex":0.0040831785,"about_ca_topic_score_gemma":0.014008835,"teacher_disagreement_score":0.013094328,"about_ca_system_score_codex":0.00045265636,"about_ca_system_score_gemma":0.0019411582,"threshold_uncertainty_score":0.03555727},"labels":[],"label_agreement":null},{"id":"W4416536103","doi":"10.1038/s41597-025-05513-8","title":"Twenty years (2000–2020) of butterfly monitoring data across the contiguous United States","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Lepidoptera: Biology and Taxonomy","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Cascades (Canada)","funders":"U.S. Geological Survey; U.S. Fish and Wildlife Service; Israel Institute for Biological Research; U.S. Department of the Interior; National Science Foundation","keywords":"Butterfly; Taxonomy (biology); Biodiversity; Population; Taxonomic rank; Occupancy","score_opus":0.04054657016238197,"score_gpt":0.32129642095769834,"score_spread":0.28074985079531634,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416536103","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09780091,0.0016609015,0.00138047,0.0009930956,0.0001427131,0.000058805417,0.8922485,0.00038670626,0.005327955],"genre_scores_gemma":[0.09036713,0.0010907999,0.002744132,0.000505454,0.000074216056,0.00023795615,0.9030937,0.000046172332,0.0018403921],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99935263,0.00008006562,0.000097761855,0.00017943163,0.00020231784,0.000087739194],"domain_scores_gemma":[0.99737895,0.0002559565,0.000824355,0.00021927053,0.0010156657,0.00030593923],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00095210236,0.0005209173,0.00037568976,0.0029928007,0.00038896722,0.00064702507,0.0007625282,0.0004574524,0.0035405685],"category_scores_gemma":[0.0029412867,0.00029336137,0.0004253908,0.005993539,0.00014836535,0.00060587295,0.001082161,0.000543372,0.0017676954],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027092555,0.00014642709,0.5118684,0.0008957489,0.00053188356,0.0002813593,0.00037857727,0.004091043,0.0010265111,0.0011196499,0.4137384,0.065651156],"study_design_scores_gemma":[0.000058112364,0.000067208864,0.79289126,0.00040584643,0.00016308091,0.00021568805,0.0004993753,0.0030974925,0.00066334987,0.00066888425,0.20122999,0.000039801445],"about_ca_topic_score_codex":0.15013054,"about_ca_topic_score_gemma":0.14748414,"teacher_disagreement_score":0.15013054,"about_ca_system_score_codex":0.00088595896,"about_ca_system_score_gemma":0.0019044353,"threshold_uncertainty_score":0.29851335},"labels":[],"label_agreement":null},{"id":"W4416648216","doi":"10.1038/s41597-025-06269-x","title":"Historical insect disturbance maps from 1985 onwards for Canadian forests derived using earth observation data","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Remote Sensing in Agriculture","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval; Natural Resources Canada; Canadian Forest Service","funders":"Government of Canada","keywords":"Disturbance (geology); Taiga; Canopy; Satellite imagery; Boreal; Vegetation (pathology); Spruce budworm; Earth observation; Insect pest","score_opus":0.09896318917980036,"score_gpt":0.271258103074513,"score_spread":0.17229491389471263,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416648216","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.68680924,0.0011850141,0.002068036,0.00012250773,0.000028188739,0.00019887718,0.29605463,0.00035152904,0.013181957],"genre_scores_gemma":[0.7965255,0.001086014,0.0076852664,0.00003787121,0.0000101942205,0.00011898225,0.18940656,0.000043334257,0.005086226],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99976367,0.000007485437,0.000017291728,0.000037886788,0.00010476102,0.00006890915],"domain_scores_gemma":[0.99880755,0.000059702234,0.00018371228,0.000043125765,0.00078218954,0.00012381589],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00025379972,0.00042239812,0.00013371676,0.005023435,0.00084824045,0.000697777,0.0003815779,0.00013585355,0.0019712509],"category_scores_gemma":[0.0010917797,0.00015299326,0.00029434246,0.0070490744,0.00019952774,0.0002442307,0.00032988592,0.00025180352,0.00032851254],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036539973,0.000115974544,0.80690503,0.0005082356,0.00021660845,0.00044136946,0.0013961405,0.021458512,0.0036310754,0.001574141,0.04249351,0.120894045],"study_design_scores_gemma":[0.0000054507905,0.000010717629,0.98159224,0.000038293412,0.000019715031,0.000048304293,0.0003311191,0.002657162,0.00034451357,0.000036162102,0.01490216,0.000014085828],"about_ca_topic_score_codex":0.98816675,"about_ca_topic_score_gemma":0.9966467,"teacher_disagreement_score":0.0118332505,"about_ca_system_score_codex":0.011062584,"about_ca_system_score_gemma":0.010446633,"threshold_uncertainty_score":0.080264986},"labels":[],"label_agreement":null},{"id":"W4416729431","doi":"10.1038/s41597-025-06322-9","title":"Balanced X-ray Security Dataset and Enhanced YOLO for Contraband Detection","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Department of Industrial and Systems Engineering, Hong Kong Polytechnic University; Hong Kong Polytechnic University","keywords":"Undersampling; Class (philosophy); Software deployment; Adaptability; Enhanced Data Rates for GSM Evolution; A priori and a posteriori; Economic shortage; Downgrade","score_opus":0.024705469303074025,"score_gpt":0.31039716708903015,"score_spread":0.2856916977859561,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416729431","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.49355367,0.008683392,0.33992112,0.002057306,0.0016569888,0.0013464198,0.09859444,0.02574894,0.028437661],"genre_scores_gemma":[0.5705946,0.0016504773,0.18080528,0.0009144011,0.00019202821,0.0006194577,0.23401874,0.0006581328,0.010546853],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992586,0.00007737622,0.000045877045,0.00026346996,0.000239069,0.00011557149],"domain_scores_gemma":[0.99929714,0.00013071629,0.000078277895,0.00024218079,0.00021417023,0.000037475926],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000831928,0.0017877691,0.0009601909,0.0018093315,0.00053254847,0.0012691366,0.001826595,0.0015219583,0.002664604],"category_scores_gemma":[0.0020731222,0.00031245075,0.0010029519,0.0012674901,0.0005474258,0.0013078345,0.00140187,0.0010792058,0.0026945577],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00236024,0.001299432,0.04243828,0.0017216993,0.00046436585,0.0012243532,0.00023312835,0.12892887,0.055650182,0.0039328425,0.1402843,0.6214623],"study_design_scores_gemma":[0.00020240393,0.0006646602,0.040860057,0.00027194453,0.00022685876,0.0018310827,0.0004975669,0.7729856,0.06739983,0.00475794,0.11014313,0.00015892943],"about_ca_topic_score_codex":0.0077370643,"about_ca_topic_score_gemma":0.013832488,"teacher_disagreement_score":0.0077370643,"about_ca_system_score_codex":0.0007349419,"about_ca_system_score_gemma":0.0010565354,"threshold_uncertainty_score":0.015384078},"labels":[],"label_agreement":null},{"id":"W4416878460","doi":"10.1038/s41597-025-06332-7","title":"Bulk RNA sequencing dataset of Claudin-low breast cancer cell lines with Neuropilin-1 knockdown","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Barrier Structure and Function Studies","field":"Neuroscience","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Transcriptome; Breast cancer; Gene knockdown; RNA; Paclitaxel; Cell culture; Cancer; Messenger RNA","score_opus":0.048178467584468414,"score_gpt":0.29740879415923926,"score_spread":0.24923032657477084,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416878460","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.28933832,0.0057389247,0.016388154,0.00041571364,0.0002727986,0.0003122999,0.679332,0.0023342906,0.005867477],"genre_scores_gemma":[0.14620304,0.0028546627,0.029817428,0.00060716766,0.000060247963,0.0010896801,0.81491184,0.00063138397,0.003824594],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.99948573,0.000037791393,0.000046236,0.0002056887,0.00016480035,0.000059705802],"domain_scores_gemma":[0.9994118,0.0002112454,0.00007745099,0.00007021037,0.0001844493,0.000044849483],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00040363485,0.00060963986,0.00086561183,0.0010266149,0.0007780548,0.00064765743,0.00039530228,0.0006634596,0.0035943196],"category_scores_gemma":[0.00095338956,0.00023581294,0.00061055506,0.001549728,0.0003317214,0.00019599884,0.00044465388,0.0008471685,0.0022748953],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010905109,0.00012749364,0.011159134,0.0026556274,0.00024285007,0.00086247385,0.00033295978,0.0036706894,0.9271364,0.0006337902,0.019412434,0.03267567],"study_design_scores_gemma":[0.00029561622,0.0010138417,0.21814573,0.0005191481,0.0008965504,0.0032086286,0.0010818469,0.018653836,0.4450023,0.0036750638,0.30722994,0.00027748643],"about_ca_topic_score_codex":0.0031428507,"about_ca_topic_score_gemma":0.007870978,"teacher_disagreement_score":0.0035943196,"about_ca_system_score_codex":0.00043673802,"about_ca_system_score_gemma":0.0009243063,"threshold_uncertainty_score":0.012024164},"labels":[],"label_agreement":null},{"id":"W4416947312","doi":"10.1038/s41597-025-06191-2","title":"Reconstruction of 1,979 prokaryotic metagenome-assembled genomes from 37 global cave environments","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Subterranean biodiversity and taxonomy","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Trinity Western University; Western University","funders":"Natural Science Foundation of Guangdong Province; National Natural Science Foundation of China; Li Ka Shing Foundation","keywords":"Cave; Phylum; Genome; Metagenomics; Archaea; Phylogenetics; Extremophile; Gene; Biodiversity","score_opus":0.05960361819547653,"score_gpt":0.22325499889342929,"score_spread":0.16365138069795276,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416947312","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8764923,0.0014132624,0.015082666,0.00011813121,0.0000334476,0.00009564477,0.103619225,0.0013403898,0.0018049788],"genre_scores_gemma":[0.5381096,0.0011503554,0.07205517,0.00013414885,0.000015798287,0.0001736007,0.38700488,0.00055674923,0.0007996717],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9995134,0.0000595438,0.00004862002,0.00017890905,0.00010124348,0.00009826911],"domain_scores_gemma":[0.9994567,0.00014858703,0.00007567049,0.00009468017,0.0001335086,0.00009075541],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008622565,0.0008192366,0.0006389892,0.002448554,0.0010153169,0.0012652442,0.0005841586,0.0007150948,0.0015319911],"category_scores_gemma":[0.0020976819,0.00049347396,0.0015495125,0.0035088095,0.000309882,0.0006165225,0.001223577,0.0007080363,0.0008443169],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003174415,0.00032677207,0.26827145,0.003763899,0.0013148923,0.0017758134,0.0030138888,0.026556365,0.54802084,0.0023034392,0.0087927375,0.13268548],"study_design_scores_gemma":[0.00025436794,0.0006491922,0.7561531,0.00069329987,0.0012259086,0.0021343688,0.003215986,0.048019297,0.08726554,0.0030574047,0.09704913,0.00028246213],"about_ca_topic_score_codex":0.0075726747,"about_ca_topic_score_gemma":0.014472485,"teacher_disagreement_score":0.0075726747,"about_ca_system_score_codex":0.0006111469,"about_ca_system_score_gemma":0.0020060828,"threshold_uncertainty_score":0.015057206},"labels":[],"label_agreement":null},{"id":"W4416947530","doi":"10.1038/s41597-025-06194-z","title":"A Unified Cellular Toxic Potency Dataset of PM2.5 across Chemicals, Emission Sources, and Regions in China","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Air Quality and Health Impacts","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"Fudan University; National Natural Science Foundation of China","keywords":"Human health; Risk assessment; Health risk; Resource (disambiguation); China; Particulates; Health risk assessment; Bioassay; Air quality index","score_opus":0.05077983788302821,"score_gpt":0.3439604614852171,"score_spread":0.2931806236021889,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416947530","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.36494097,0.002160229,0.0046873665,0.00035853533,0.000063991065,0.00028437912,0.6214863,0.0007528354,0.0052654706],"genre_scores_gemma":[0.3135745,0.0009032812,0.0026763259,0.00013176541,0.00002985233,0.0003672477,0.6811095,0.000034802193,0.0011727145],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9990606,0.000112135574,0.000118130585,0.00030371634,0.00026678474,0.00013870659],"domain_scores_gemma":[0.9991364,0.0001052217,0.00009257417,0.0002056572,0.00036500173,0.0000951133],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00085432193,0.0012020121,0.0012016561,0.0033660736,0.00074847613,0.00086524733,0.0013286505,0.0008187851,0.0013887177],"category_scores_gemma":[0.0011943022,0.000283092,0.001160435,0.0056412425,0.0004374278,0.00052099506,0.0014133042,0.00038225946,0.0006281946],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010193248,0.00055750966,0.6612575,0.0040814853,0.0012673109,0.0016677451,0.0005974837,0.071942195,0.027737753,0.0035666546,0.11135643,0.11494857],"study_design_scores_gemma":[0.00017207373,0.00024281426,0.8368837,0.00018417348,0.0005406597,0.00044977485,0.00046016485,0.03624973,0.010366184,0.0019600606,0.11232826,0.00016236241],"about_ca_topic_score_codex":0.109637685,"about_ca_topic_score_gemma":0.08337714,"teacher_disagreement_score":0.109637685,"about_ca_system_score_codex":0.002574464,"about_ca_system_score_gemma":0.0037851953,"threshold_uncertainty_score":0.21799904},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4417263449","doi":"10.1038/s41597-025-06289-7","title":"The Ouranos CRCM5-CMIP6 ensemble: A dynamically downscaled ensemble of CMIP6 simulations over North America","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Climate variability and models","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ouranos","funders":"Alliance de recherche numérique du Canada; Environment and Climate Change Canada; Innovation, Science and Economic Development Canada; Université du Québec à Montréal","keywords":"Downscaling; Coupled model intercomparison project; Climate model; Grid; General Circulation Model; Climate change; Component (thermodynamics); Climate simulation","score_opus":0.02206600415558369,"score_gpt":0.27750294221030913,"score_spread":0.25543693805472545,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417263449","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.63711923,0.0016389559,0.049024645,0.0019638757,0.0009534961,0.00056002487,0.2477419,0.011402488,0.049595404],"genre_scores_gemma":[0.7378261,0.0008662683,0.07126271,0.00044071197,0.0003145538,0.0005335138,0.18122916,0.0017673937,0.0057595866],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99977726,0.000034620043,0.000008363147,0.000063944244,0.00007538009,0.000040555144],"domain_scores_gemma":[0.99947125,0.000049582173,0.00003517368,0.00011926357,0.00022452525,0.000100282225],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006567436,0.0006794054,0.0007082222,0.0006337581,0.0008131864,0.00075393816,0.0016564819,0.0005166474,0.003024815],"category_scores_gemma":[0.0016696039,0.00033222145,0.0005950386,0.0014195944,0.0002495219,0.00091266766,0.000780691,0.000985211,0.00075428456],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00052906194,0.00032898673,0.050738215,0.000279915,0.0009171002,0.00018795813,0.00025367897,0.73444337,0.009562416,0.0063494425,0.108468354,0.08794145],"study_design_scores_gemma":[0.00054998294,0.00006146687,0.040734477,0.00005567242,0.00021783562,0.000044840697,0.00008465036,0.8724906,0.0046026832,0.0026902535,0.07832331,0.00014423975],"about_ca_topic_score_codex":0.3861906,"about_ca_topic_score_gemma":0.37984043,"teacher_disagreement_score":0.6138094,"about_ca_system_score_codex":0.0017437454,"about_ca_system_score_gemma":0.003615905,"threshold_uncertainty_score":0.76788545},"labels":[],"label_agreement":null},{"id":"W4417264405","doi":"10.1038/s41597-025-06403-9","title":"Near complete T2T genome assembly of the banded goonch (Bagarius rutilus)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Agriculture","funders":"","keywords":"Synteny; Sequence assembly; Genome; Clade; Gene; Reference genome; Nanopore sequencing; Genome size; Structural variation","score_opus":0.03447199554767675,"score_gpt":0.2710889516429916,"score_spread":0.23661695609531488,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417264405","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8678009,0.0023801925,0.0693682,0.0002972803,0.00015570632,0.00026302758,0.051715866,0.0018726316,0.006146133],"genre_scores_gemma":[0.6230421,0.0019197607,0.1285483,0.00020066807,0.000046406953,0.00032213118,0.23719719,0.0011426349,0.0075807357],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9998658,0.000011427859,0.000011072713,0.000058961865,0.000035705438,0.000017022705],"domain_scores_gemma":[0.99980336,0.00003231665,0.000045918783,0.000023866647,0.000051213403,0.000043324377],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00018950173,0.00068285904,0.0004950105,0.0010477154,0.00058348867,0.0005201815,0.00030289456,0.0004567073,0.0025438743],"category_scores_gemma":[0.00046146507,0.00026157062,0.00088314555,0.00083986437,0.00017115683,0.00032178732,0.000540226,0.00051571673,0.0014843569],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002341509,0.000025636773,0.003690094,0.00034467893,0.000068354755,0.00038168422,0.00027415194,0.0008909251,0.9774845,0.0002707113,0.00085040845,0.015484587],"study_design_scores_gemma":[0.00015553263,0.0010890403,0.38599515,0.000342046,0.0009179796,0.004147851,0.0009789373,0.01902354,0.45741346,0.0011776191,0.12853085,0.00022802329],"about_ca_topic_score_codex":0.00444342,"about_ca_topic_score_gemma":0.008709594,"teacher_disagreement_score":0.00444342,"about_ca_system_score_codex":0.00037074534,"about_ca_system_score_gemma":0.0007006971,"threshold_uncertainty_score":0.008835077},"labels":[],"label_agreement":null},{"id":"W4417423520","doi":"10.1038/s41597-025-06375-w","title":"A global dataset of helium isotopic data in tectonically and hydrothermally active geological environments","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Geological and Geochemical Analysis","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Chinese Academy of Sciences Key Project; National Natural Science Foundation of China","keywords":"Helium; Hydrothermal circulation; Mantle (geology); Isotope; Resource (disambiguation); Range (aeronautics)","score_opus":0.025556245052968756,"score_gpt":0.2548514132141965,"score_spread":0.22929516816122777,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417423520","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018427396,0.0058547235,0.0012071787,0.00014605268,0.00006192138,0.000039344537,0.9692377,0.00017429296,0.0048513464],"genre_scores_gemma":[0.035314802,0.006946274,0.0052276696,0.00013267872,0.00009248221,0.00020965461,0.95020187,0.000096223615,0.0017783708],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998941,0.00008958189,0.00033681747,0.0002968779,0.00025141262,0.0000844071],"domain_scores_gemma":[0.9953211,0.0014401916,0.001316325,0.00047344423,0.0011834442,0.00026545988],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010578235,0.0008512123,0.0010675219,0.031692054,0.0005522549,0.0013265901,0.0007293584,0.00089501374,0.008861918],"category_scores_gemma":[0.005243593,0.00030865867,0.000681053,0.040169533,0.0004961719,0.001118116,0.0016706438,0.0006045322,0.0040680915],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011065983,0.00025682346,0.25611162,0.05835468,0.0018318383,0.003496793,0.0030284687,0.004352888,0.044414308,0.014437919,0.30744183,0.3051663],"study_design_scores_gemma":[0.000052311938,0.00004668182,0.167398,0.0020240778,0.00042144948,0.00060437707,0.00072649407,0.00024713745,0.005407869,0.0018224694,0.82118237,0.000066713146],"about_ca_topic_score_codex":0.008651315,"about_ca_topic_score_gemma":0.01276918,"teacher_disagreement_score":0.031692054,"about_ca_system_score_codex":0.00071162765,"about_ca_system_score_gemma":0.0030344822,"threshold_uncertainty_score":0.029646099},"labels":[],"label_agreement":null},{"id":"W4417455182","doi":"10.1038/s41597-025-06365-y","title":"An Egocentric Life-Saving Interventional Procedure Dataset of Actions, Medical Questions, Maneuvers and Tools","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"U.S. Army Research Institute of Environmental Medicine; U.S. Army Medical Research and Development Command; Purdue University; School of Medicine, Indiana University; U.S. Department of Defense; National Science Foundation","keywords":"CLIPS; Action (physics); Resource (disambiguation); Object (grammar); Field (mathematics); Metadata","score_opus":0.04955760178494154,"score_gpt":0.3772135807482431,"score_spread":0.32765597896330156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417455182","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027201299,0.0021875845,0.004446418,0.0006174945,0.00033431663,0.00036850662,0.9546243,0.0037434257,0.0064767697],"genre_scores_gemma":[0.017270865,0.00038472374,0.0055878726,0.0001873796,0.000049451708,0.00025597936,0.974502,0.000105376304,0.0016563917],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99917114,0.00014269477,0.00007715834,0.00025851207,0.00023844541,0.00011208971],"domain_scores_gemma":[0.99883884,0.00031697043,0.00012583467,0.00023563355,0.00029893132,0.00018370053],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00055339123,0.001918769,0.000883133,0.0024453844,0.00069087354,0.0009029025,0.0016631301,0.002200801,0.007632561],"category_scores_gemma":[0.0024886476,0.00029729874,0.0011292993,0.002290029,0.0005106703,0.0007519324,0.0013473514,0.0012199437,0.009349943],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009992669,0.00078173936,0.015544033,0.0029844167,0.00026563095,0.0010216661,0.00043058465,0.0061230944,0.0063045854,0.0014696091,0.87620044,0.08787504],"study_design_scores_gemma":[0.00039460356,0.00054525444,0.08773055,0.0011456672,0.0002299515,0.002137885,0.0014486997,0.026091695,0.009688511,0.0036000751,0.8667606,0.00022653304],"about_ca_topic_score_codex":0.0229802,"about_ca_topic_score_gemma":0.07182089,"teacher_disagreement_score":0.0229802,"about_ca_system_score_codex":0.0012325095,"about_ca_system_score_gemma":0.0014783744,"threshold_uncertainty_score":0.04569292},"labels":[],"label_agreement":null},{"id":"W4417456164","doi":"10.1038/s41597-025-06429-z","title":"Global Ensemble Fire Emission Product Version 1.0 (EnsemFire V1.0)","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Atmospheric chemistry and aerosols","field":"Earth and Planetary Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"NOAA Weather Program Office; National Oceanic and Atmospheric Administration; U.S. Department of Commerce; U.S. Environmental Protection Agency","keywords":"Greenhouse gas; Particulates; Methane; Biomass burning; Air quality index; Aerosol; Carbon dioxide; Carbon fibers; Radiative transfer","score_opus":0.020977967436498504,"score_gpt":0.25195727652020417,"score_spread":0.23097930908370568,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417456164","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.057141013,0.0011840423,0.09403831,0.000331335,0.0006199718,0.00030005275,0.7875859,0.046484184,0.012315222],"genre_scores_gemma":[0.067687094,0.00041619927,0.08549661,0.00022700844,0.00011436014,0.00068457256,0.8390284,0.0041327947,0.002212917],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99953985,0.00010861547,0.000035954166,0.00012837992,0.00013380869,0.00005337827],"domain_scores_gemma":[0.99929726,0.00018507442,0.00007987121,0.000198356,0.00017636699,0.00006311016],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017649813,0.0017096065,0.0012489627,0.0014576439,0.0005217579,0.000989042,0.0016773662,0.0010979249,0.006726582],"category_scores_gemma":[0.002565313,0.0007032332,0.0014663502,0.002059649,0.00023793001,0.0011736233,0.00086869334,0.0014609131,0.005477132],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00065828266,0.00030308435,0.046162874,0.001096823,0.0016171014,0.0004117975,0.0002685537,0.28820223,0.009592181,0.011348528,0.529182,0.111156605],"study_design_scores_gemma":[0.0006622312,0.00020551644,0.036045928,0.00023509862,0.00039071214,0.00040690703,0.00006405647,0.53454983,0.01174558,0.015316722,0.4000535,0.0003238041],"about_ca_topic_score_codex":0.014477839,"about_ca_topic_score_gemma":0.013369765,"teacher_disagreement_score":0.014477839,"about_ca_system_score_codex":0.0004508757,"about_ca_system_score_gemma":0.0012035404,"threshold_uncertainty_score":0.028787136},"labels":[],"label_agreement":null},{"id":"W4417502311","doi":"10.1038/s41597-025-06349-y","title":"A comprehensive dataset of riverine levee overtopping events for advancing risk assessment","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Dam Engineering and Safety","field":"Engineering","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"United Nations University Institute for Water, Environment, and Health","funders":"U.S. Army Corps of Engineers; U.S. Department of Defense","keywords":"Levee; Flood myth; Risk assessment; Consistency (knowledge bases); Hydrology (agriculture); Foundation (evidence); Flood control","score_opus":0.02816973448253714,"score_gpt":0.3182483642074622,"score_spread":0.2900786297249251,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417502311","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024861738,0.0002496278,0.0031900993,0.00023115505,0.000059560207,0.000092268274,0.9662978,0.0011156076,0.0039020288],"genre_scores_gemma":[0.023524836,0.00014404127,0.0049706707,0.000053441705,0.000020756852,0.00016570228,0.9701855,0.00008271476,0.00085247925],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994104,0.00008788329,0.000096963384,0.00016281793,0.00016754011,0.0000744661],"domain_scores_gemma":[0.9981488,0.00034325253,0.00035622902,0.0003729131,0.00063018815,0.00014868003],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006550789,0.0006968925,0.0004184284,0.0028067986,0.0004086568,0.0007707018,0.00083833997,0.0007063995,0.005266085],"category_scores_gemma":[0.0032793698,0.00019522464,0.0006599585,0.0039657173,0.0002119169,0.00078851,0.001482349,0.00081402756,0.0047082272],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019217088,0.00028537202,0.16727422,0.001456158,0.00025369413,0.0005439052,0.00074543944,0.012266402,0.004256977,0.0033033267,0.7386006,0.07082173],"study_design_scores_gemma":[0.000083853476,0.00010375686,0.30881375,0.0005299657,0.00009836964,0.00033456858,0.0013752283,0.015342538,0.0029804504,0.003050295,0.66717976,0.00010748828],"about_ca_topic_score_codex":0.022910427,"about_ca_topic_score_gemma":0.048095044,"teacher_disagreement_score":0.022910427,"about_ca_system_score_codex":0.00062821317,"about_ca_system_score_gemma":0.0012884603,"threshold_uncertainty_score":0.0455541},"labels":[],"label_agreement":null},{"id":"W4417524405","doi":"10.1038/s41597-025-06447-x","title":"De novo assembly of complete circular mitochondrial genomes from 2,695 fungal species","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"European Regional Development Fund; Ministère de l'Enseignement Supérieur, de la Recherche, de la Science et de la Technologie; Agence Nationale de la Recherche; Austrian Science Fund; European Commission","keywords":"Phylum; Genome; Phylogenetic tree; Mitochondrial DNA; Phylogenetics; Sequence assembly; Fungal Diversity; Whole genome sequencing; Taxonomic rank","score_opus":0.03950466578097342,"score_gpt":0.26856863300109435,"score_spread":0.22906396722012093,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417524405","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6545824,0.0069297347,0.12892014,0.0007255373,0.00052136334,0.00059594284,0.19538268,0.004795525,0.0075467373],"genre_scores_gemma":[0.2455411,0.0024111415,0.29474488,0.00034152504,0.00009994798,0.00036601978,0.4522797,0.0016763913,0.0025393378],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99890554,0.000112243775,0.00012682778,0.00049152086,0.00025260338,0.00011133024],"domain_scores_gemma":[0.99804926,0.00049002963,0.0003248853,0.0003346106,0.00052718865,0.0002739997],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001661734,0.0009189169,0.000990867,0.0025598381,0.001857875,0.0017932098,0.00096332247,0.001034548,0.0029003697],"category_scores_gemma":[0.004076804,0.0009132692,0.0018967025,0.0027303344,0.0004400568,0.0014540725,0.0018525717,0.0016276228,0.0025137928],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025378377,0.0003678266,0.039904874,0.003798713,0.00072258426,0.0018208503,0.0022297655,0.0074279783,0.7860503,0.006149031,0.017180642,0.13180964],"study_design_scores_gemma":[0.0006443033,0.0014219257,0.20703861,0.0014311495,0.0016893711,0.0035926125,0.002113369,0.06949429,0.2799268,0.013806417,0.41832286,0.0005182727],"about_ca_topic_score_codex":0.0030495727,"about_ca_topic_score_gemma":0.013450808,"teacher_disagreement_score":0.0030495727,"about_ca_system_score_codex":0.0009258126,"about_ca_system_score_gemma":0.002281851,"threshold_uncertainty_score":0.0097026825},"labels":[],"label_agreement":null},{"id":"W7116384933","doi":"10.1038/s41597-025-06021-5","title":"Responsible AI measures dataset for ethics evaluation of AI systems","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Google (Canada); McGill University","funders":"","keywords":"Normative; Usability; Set (abstract data type); Interpretation (philosophy); Domain (mathematical analysis); Visualization; Corporate governance","score_opus":0.5162686606936119,"score_gpt":0.5809189398766879,"score_spread":0.06465027918307598,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7116384933","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0076415148,0.00042338338,0.0018171935,0.0008291737,0.00013246006,0.0007607373,0.9656383,0.0005981006,0.022159226],"genre_scores_gemma":[0.0194277,0.0003069815,0.007372459,0.00040744024,0.00009863262,0.0040366105,0.96197426,0.00036946603,0.006006493],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.98761785,0.0030347889,0.0028185542,0.00091986894,0.0050629238,0.00054603285],"domain_scores_gemma":[0.8916572,0.055597275,0.011893631,0.011807175,0.02657065,0.002474017],"candidate_categories":["research_integrity"],"consensus_categories":[],"category_scores_codex":[0.007680823,0.0008173162,0.0006569101,0.013440086,0.0016828296,0.0031111918,0.0015597674,0.0017415218,0.03653676],"category_scores_gemma":[0.070521764,0.0003513855,0.0008475198,0.014757239,0.0009434574,0.0022846567,0.0025688808,0.0019798034,0.015671782],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023047438,0.00025395572,0.014952726,0.003341391,0.00009296506,0.00012490718,0.0010227527,0.00096199114,0.00070123305,0.011649728,0.9247168,0.041951176],"study_design_scores_gemma":[0.00015224292,0.00009305802,0.04434413,0.0009798604,0.000036010337,0.000128972,0.0012597903,0.0006796632,0.0009043153,0.00442792,0.94692034,0.000073799405],"about_ca_topic_score_codex":0.009714425,"about_ca_topic_score_gemma":0.020767212,"teacher_disagreement_score":0.9982585,"about_ca_system_score_codex":0.0032487155,"about_ca_system_score_gemma":0.004370062,"threshold_uncertainty_score":0.12222761},"labels":[],"label_agreement":null},{"id":"W7117106493","doi":"10.1038/s41597-025-06301-0","title":"National human footprint maps for Peru and Ecuador","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Environmental Impact and Sustainability","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Northern British Columbia","funders":"","keywords":"Biodiversity; Land cover; Footprint; Ecological footprint; Population; Land use; Population pressure; Sustainable development","score_opus":0.035932136756059996,"score_gpt":0.32784292886499244,"score_spread":0.29191079210893245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7117106493","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.053017925,0.0015187955,0.002349322,0.0006157076,0.00007344828,0.00021615255,0.90888417,0.001085222,0.03223932],"genre_scores_gemma":[0.10536707,0.0014717556,0.011864477,0.00013864334,0.000058585367,0.0015372777,0.8705934,0.00023728612,0.008731493],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999754,0.000047561203,0.000034952904,0.000055544893,0.00006686978,0.00004107946],"domain_scores_gemma":[0.99935645,0.00010125781,0.00012057699,0.00007721852,0.0003026187,0.000041956482],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00050062995,0.0005883115,0.00026192347,0.004521719,0.00031576667,0.0007840923,0.0005452902,0.00024814965,0.019180791],"category_scores_gemma":[0.0023844785,0.00020820802,0.00033317396,0.0073753246,0.00011198064,0.0008094172,0.0010553318,0.00035621136,0.0033808867],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021751986,0.0001508732,0.10149858,0.0027679508,0.00027769353,0.0006910348,0.002718401,0.00304638,0.0015140297,0.005919797,0.7082341,0.17296371],"study_design_scores_gemma":[0.00008391442,0.00003353241,0.2707132,0.00047671213,0.00006455637,0.00047319298,0.0019252987,0.00236601,0.00063469546,0.0014201334,0.7217545,0.000054315904],"about_ca_topic_score_codex":0.034158073,"about_ca_topic_score_gemma":0.024158688,"teacher_disagreement_score":0.034158073,"about_ca_system_score_codex":0.0004632189,"about_ca_system_score_gemma":0.0006571483,"threshold_uncertainty_score":0.06791854},"labels":[],"label_agreement":null},{"id":"W7117660909","doi":"10.1038/s41597-025-06463-x","title":"Features of interest from a multi-season satellite survey of baleen whales on the West Antarctic Peninsula","year":2025,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine animal studies overview","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of New Brunswick","funders":"World Wildlife Fund","keywords":"Peninsula; Baleen; Satellite; Satellite imagery; Bay; Data processing","score_opus":0.15326359689348787,"score_gpt":0.31867899644264147,"score_spread":0.1654153995491536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7117660909","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9487607,0.00044713367,0.0010576304,0.00015901602,0.000045554858,0.000079092635,0.046555057,0.00016434625,0.002731418],"genre_scores_gemma":[0.85398895,0.0004349695,0.0062553324,0.00010793226,0.000039184233,0.0001506488,0.13593335,0.000044991444,0.0030445177],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.999694,0.000037625156,0.000026484146,0.000087312175,0.000091114,0.000063530584],"domain_scores_gemma":[0.9992582,0.00011520897,0.00013142466,0.00014060304,0.00025525634,0.00009931529],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00033865724,0.000370031,0.00019303271,0.0024858685,0.0004981366,0.000455248,0.00035755048,0.00036931914,0.0013012133],"category_scores_gemma":[0.00070565636,0.00014916653,0.00023960872,0.0021477328,0.00028654054,0.0002715476,0.0006422389,0.00023082561,0.00063271855],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021450142,0.00019031031,0.88336664,0.00036697707,0.0002231032,0.0015573354,0.0022915963,0.0037172113,0.016570674,0.0001818111,0.025165275,0.06615458],"study_design_scores_gemma":[0.0000051057655,0.000028782571,0.98245883,0.00005256721,0.000028187937,0.00019162112,0.0022200644,0.0012118754,0.0012056429,0.00005848898,0.012525189,0.000013513284],"about_ca_topic_score_codex":0.0722947,"about_ca_topic_score_gemma":0.20066436,"teacher_disagreement_score":0.0722947,"about_ca_system_score_codex":0.0003574699,"about_ca_system_score_gemma":0.0005563197,"threshold_uncertainty_score":0.1437478},"labels":[],"label_agreement":null}]}