{"meta":{"query_hash":"49a0e8cecebf","filters":{"venue":"FAIRsharing.org"},"cohort_total":27,"direct_labels_cover":2,"predictions_cover":27,"exported":27,"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/49a0e8cecebf","api":"https://metacan.xera.ac/api/v1/cohort?venue=FAIRsharing.org"},"results":[{"id":"W6889176240","doi":"10.25504/fairsharing.41de90","title":"FAIRsharing record for: WHO Guidelines for good clinical practice (GCP) for trials on pharmaceutical products","year":2022,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"Advanced MIMO Systems Optimization","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":"Good clinical practice; Clinical Practice; Clinical trial; Good manufacturing practice; Good practice; Set (abstract data type); Pharmaceutical industry; Good laboratory practice; Alternative medicine","score_opus":0.34771768535315467,"score_gpt":0.4952829109576249,"score_spread":0.14756522560447022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6889176240","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":[0.00009458013,0.00024175293,0.0015990159,0.0017796188,0.00015610857,0.00020955161,0.98438245,0.0017070555,0.009829977],"genre_scores_gemma":[0.002392304,0.0007419469,0.008516091,0.0020230457,0.0001443195,0.0022865466,0.97188437,0.0016032626,0.010408109],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98688567,0.004744506,0.0025152226,0.0014057064,0.0036967518,0.0007520183],"domain_scores_gemma":[0.91330063,0.049168505,0.007530377,0.018524269,0.008888828,0.002587433],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.019382272,0.0011300315,0.0018871956,0.005274019,0.0013384911,0.009645292,0.0041119033,0.0051762126,0.27725738],"category_scores_gemma":[0.13642272,0.0012132614,0.0014711994,0.010905115,0.0014750642,0.00563085,0.0057772137,0.0044768555,0.20474546],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048880185,0.000009769373,0.0002358012,0.00066502724,0.000016281192,0.000011942617,0.00002295436,0.00019491083,0.000013180072,0.0027077522,0.9898046,0.006268833],"study_design_scores_gemma":[0.00031049465,0.0000139205,0.0008667043,0.0011235722,0.000015314494,0.00004360518,0.000029687722,0.00030995265,0.000103823106,0.011218914,0.9859313,0.000032751435],"about_ca_topic_score_codex":0.0095885955,"about_ca_topic_score_gemma":0.017023504,"teacher_disagreement_score":0.9806177,"about_ca_system_score_codex":0.0041537085,"about_ca_system_score_gemma":0.0118420785,"threshold_uncertainty_score":0.92751837},"labels":[],"label_agreement":null},{"id":"W6908006414","doi":"10.25504/fairsharing.8b7a2f","title":"FAIRsharing record for: Fisheries and Oceans Canada Pacific Region Data Archive","year":2022,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Arctic; Beaufort sea; Pacific ocean; Continental shelf; The arctic; Marine research; Data archive","score_opus":0.07765737325568293,"score_gpt":0.27037017371460376,"score_spread":0.1927128004589208,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6908006414","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000025246074,0.000013565646,0.000016017428,0.000040044433,0.000013170565,0.000004210626,0.99894184,0.0001010847,0.0008447888],"genre_scores_gemma":[0.00024954957,0.00005008743,0.00010854694,0.000033547796,0.0000057499797,0.000032484953,0.9977315,0.00010166125,0.0016868941],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99881315,0.00007651842,0.00011890871,0.00024372066,0.00047034197,0.00027731553],"domain_scores_gemma":[0.99506444,0.00059374806,0.0003134489,0.0009826103,0.0024777236,0.0005680267],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0010820898,0.0014784716,0.0012018712,0.006797609,0.001840842,0.004004072,0.003224145,0.0013595904,0.20337124],"category_scores_gemma":[0.009107262,0.0008706536,0.0009312388,0.018864047,0.00058909593,0.0026748092,0.0026479668,0.0019583437,0.13700832],"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":[0.000009101776,0.0000029358528,0.00022883511,0.00006658364,0.000004839558,0.00000516236,0.000007193262,0.00005517909,0.000007129426,0.00026029022,0.9984634,0.00088938046],"study_design_scores_gemma":[0.00005772439,0.0000026110552,0.0035810168,0.00019276324,0.000008623269,0.000014530155,0.00007183821,0.00015911866,0.000064560205,0.0008680408,0.99495333,0.00002569728],"about_ca_topic_score_codex":0.60598224,"about_ca_topic_score_gemma":0.5749052,"teacher_disagreement_score":0.99677587,"about_ca_system_score_codex":0.005613586,"about_ca_system_score_gemma":0.014317644,"threshold_uncertainty_score":0.7926764},"labels":[],"label_agreement":null},{"id":"W6908040303","doi":"10.25504/fairsharing.b3df6f","title":"FAIRsharing record for: Lunaris","year":2025,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Multidisciplinary approach; Service (business); Interface (matter); Documentation; Data collection","score_opus":0.044017425877515205,"score_gpt":0.3272946767283657,"score_spread":0.2832772508508505,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6908040303","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00013271361,0.000058592886,0.00012516217,0.0005512724,0.00020118107,0.000029854084,0.9845553,0.002591701,0.011754266],"genre_scores_gemma":[0.0005251449,0.00007665073,0.0003521407,0.000324604,0.00004151969,0.00007528001,0.980291,0.001173441,0.017140279],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980572,0.00017900193,0.00012715696,0.00047066866,0.0008120978,0.0003537368],"domain_scores_gemma":[0.9924137,0.0008465849,0.00024805713,0.0027317517,0.0024707061,0.0012892417],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0019999763,0.0014864716,0.001656582,0.0051067206,0.0033755247,0.007331796,0.0043235607,0.0018864314,0.47625333],"category_scores_gemma":[0.0125682615,0.0008391508,0.0009841893,0.009600235,0.0010218073,0.005142265,0.0051760655,0.0021971078,0.48239362],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010649636,0.0000029911837,0.0000507994,0.000024825247,0.0000014059054,0.0000033752688,0.0000064745327,0.0000102335225,0.000013759825,0.00020910885,0.9988103,0.0008560407],"study_design_scores_gemma":[0.00003226007,0.0000031387951,0.00059171964,0.000050866085,0.0000023986925,0.0000122008305,0.000041273343,0.00006817607,0.00009136044,0.0007332464,0.99836046,0.000012812606],"about_ca_topic_score_codex":0.13759507,"about_ca_topic_score_gemma":0.21169724,"teacher_disagreement_score":0.47625333,"about_ca_system_score_codex":0.0044604456,"about_ca_system_score_gemma":0.0084459465,"threshold_uncertainty_score":0.747061},"labels":[],"label_agreement":null},{"id":"W6908112345","doi":"10.25504/fairsharing.mtkkwk","title":"FAIRsharing record for: Coronavirus and MS Reporting Database","year":2018,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Multiple sclerosis; Coronavirus disease 2019 (COVID-19); Coronavirus; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); MEDLINE; Pandemic","score_opus":0.15813814023634293,"score_gpt":0.37969005331484035,"score_spread":0.22155191307849742,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6908112345","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000055840675,0.00002809343,0.000027639817,0.00010813658,0.000018357208,0.000008614394,0.99896574,0.0000951543,0.00069251604],"genre_scores_gemma":[0.00050005055,0.00007345116,0.00013962196,0.00012558184,0.000017223567,0.00006544181,0.99806947,0.00008027042,0.0009288919],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984847,0.00020386456,0.000397196,0.0003700885,0.0003429835,0.0002011623],"domain_scores_gemma":[0.9929939,0.0015887384,0.0012896179,0.0013198585,0.0020729632,0.00073501235],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014682813,0.0010484938,0.0011500919,0.0042830394,0.00084607746,0.0029975562,0.0024411576,0.0017105589,0.16800115],"category_scores_gemma":[0.015649274,0.00064588035,0.0009536631,0.009383951,0.0003707167,0.0024469877,0.0024350712,0.0014696129,0.09559349],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004192555,0.0000061277897,0.00072464836,0.00027442913,0.000014308226,0.000012988797,0.000011302956,0.0000571518,0.000022525011,0.00045853946,0.99699247,0.0013835031],"study_design_scores_gemma":[0.00033263327,0.0000141803175,0.0066350955,0.0005286575,0.000027235961,0.00006572458,0.00007216533,0.00020060422,0.00015144738,0.0015061,0.9904337,0.000032448363],"about_ca_topic_score_codex":0.036071174,"about_ca_topic_score_gemma":0.034144707,"teacher_disagreement_score":0.16800115,"about_ca_system_score_codex":0.0020282306,"about_ca_system_score_gemma":0.0037918882,"threshold_uncertainty_score":0.5620198},"labels":[],"label_agreement":null},{"id":"W6926639550","doi":"10.25504/fairsharing.424cb8","title":"FAIRsharing record for: Better Outcomes Registry & Network","year":2022,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"Bacillus and Francisella bacterial research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Government (linguistics); Quality (philosophy); Medical record; Data quality; Data collection; Patient registry","score_opus":0.03308845319192238,"score_gpt":0.30332947116585973,"score_spread":0.27024101797393735,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926639550","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00007241598,0.000027904689,0.00004142655,0.000108173364,0.00001391038,0.0000140174225,0.99858344,0.00007506998,0.0010635537],"genre_scores_gemma":[0.0010430824,0.00009011433,0.00028933218,0.00018248108,0.00002188753,0.0001635831,0.99532545,0.00010594959,0.0027781678],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99807906,0.00023545767,0.00037098338,0.00045706797,0.00051225396,0.00034517545],"domain_scores_gemma":[0.98547274,0.0025325834,0.0023022364,0.0025216322,0.0051501426,0.0020208005],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0020609088,0.0008121096,0.0013039719,0.0031193825,0.0011854357,0.0029298977,0.0023484174,0.0015713456,0.16125391],"category_scores_gemma":[0.020269396,0.00078469317,0.00094497774,0.009185067,0.00039699808,0.0016842225,0.0023032783,0.0015215606,0.065679386],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006013369,0.0000062019258,0.0017225725,0.00030680103,0.000018376959,0.000008943891,0.000014610062,0.000063699576,0.000015768344,0.0005158396,0.9955302,0.0017368736],"study_design_scores_gemma":[0.00059714844,0.000018193916,0.018136114,0.00084876677,0.000056086603,0.000046150453,0.000102532795,0.00025085025,0.000120421486,0.0015076773,0.9782649,0.000051138584],"about_ca_topic_score_codex":0.32492855,"about_ca_topic_score_gemma":0.44869635,"teacher_disagreement_score":0.9976516,"about_ca_system_score_codex":0.0052817273,"about_ca_system_score_gemma":0.011022559,"threshold_uncertainty_score":0.64607453},"labels":[],"label_agreement":null},{"id":"W6926640784","doi":"10.25504/fairsharing.e7e609","title":"FAIRsharing record for: FAIR* Reviews Ontology","year":2022,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"Streptococcal Infections and 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":"Sumitomo Precision Products (Canada)","funders":"","keywords":"Ontology; Scientific literature; Open Biomedical Ontologies; Subject (documents); Information system","score_opus":0.10726523789159373,"score_gpt":0.3781229969239672,"score_spread":0.2708577590323735,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926640784","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":[0.00012347037,0.0003303564,0.00035871717,0.00069033063,0.00012377961,0.000042578187,0.9894196,0.0012492692,0.0076620276],"genre_scores_gemma":[0.0015818395,0.0006514341,0.002061194,0.001052632,0.00008965773,0.00022300058,0.98685443,0.00079485134,0.0066910153],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971726,0.0006568684,0.00043877427,0.0005232134,0.0009570185,0.00025152677],"domain_scores_gemma":[0.98879015,0.004457089,0.0013464591,0.002037957,0.0024857905,0.0008824873],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.003333786,0.0014945981,0.0010166251,0.008781467,0.0015271439,0.0057329615,0.0021308295,0.0029340184,0.18538897],"category_scores_gemma":[0.020378305,0.0007873241,0.0014451343,0.009026057,0.00059730915,0.0034960783,0.0033310952,0.0016963433,0.11861124],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003075515,0.000007849669,0.00031165904,0.00064540986,0.00001597231,0.000014418321,0.000021298556,0.000059975187,0.00006091437,0.0015569916,0.99468887,0.0025858998],"study_design_scores_gemma":[0.000056956073,0.0000033865108,0.00069057883,0.00021582213,0.00001168952,0.000026441228,0.000025056885,0.00008183744,0.00010445528,0.0014760186,0.9972938,0.000013881345],"about_ca_topic_score_codex":0.024463413,"about_ca_topic_score_gemma":0.050764836,"teacher_disagreement_score":0.9978692,"about_ca_system_score_codex":0.004019797,"about_ca_system_score_gemma":0.006246949,"threshold_uncertainty_score":0.6201879},"labels":[],"label_agreement":null},{"id":"W6926654570","doi":"10.25504/fairsharing.027f17","title":"FAIRsharing record for: Alberta Geological Survey Open Data Portal","year":2022,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"Diatoms and Algae Research","field":"Materials Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Geological survey; Bedrock; Geologic map; Resource (disambiguation); Geographic information system; Regional geology","score_opus":0.20963967290243912,"score_gpt":0.3946889083938543,"score_spread":0.18504923549141516,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926654570","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000054067026,0.000028674058,0.000046558176,0.00008076109,0.000022545819,0.00000995658,0.99606895,0.00046389227,0.0032245724],"genre_scores_gemma":[0.0003745553,0.00006878052,0.00023506935,0.000050320712,0.0000064234578,0.00003477844,0.9959656,0.00023240739,0.0030319563],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987338,0.000082545,0.00009393568,0.00021581123,0.00060401147,0.00026988884],"domain_scores_gemma":[0.994978,0.00064688554,0.00025533276,0.0011389416,0.0021212036,0.00085949566],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.001449168,0.0016858854,0.0012237511,0.0073169856,0.0019383756,0.004737538,0.0033372622,0.0014538767,0.226514],"category_scores_gemma":[0.008004781,0.0010082624,0.0007313239,0.018853998,0.00072835543,0.002703794,0.002864281,0.0018795468,0.16705768],"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":[0.000014010195,0.000004178579,0.0001933144,0.00007963026,0.000003099667,0.000006357211,0.00001360703,0.00006715344,0.0000197107,0.00043090776,0.9979819,0.0011861599],"study_design_scores_gemma":[0.000051616884,0.0000022424047,0.0018064263,0.00010704788,0.000005133743,0.00000991653,0.00006694095,0.000116458425,0.00008380398,0.0009886662,0.9967424,0.000019376963],"about_ca_topic_score_codex":0.6273442,"about_ca_topic_score_gemma":0.7108856,"teacher_disagreement_score":0.99666274,"about_ca_system_score_codex":0.008299573,"about_ca_system_score_gemma":0.014817122,"threshold_uncertainty_score":0.7577648},"labels":[],"label_agreement":null},{"id":"W6945367327","doi":"10.25504/fairsharing.bac679","title":"FAIRsharing record for: Canadian Integrated Ocean Observing System Data Catalogue","year":2022,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Ocean observations; Earth system science; Data sharing; Earth observation satellite; Earth observation; Selection (genetic algorithm); Data archive; Data access","score_opus":0.12017987762631747,"score_gpt":0.3020811123113687,"score_spread":0.1819012346850512,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6945367327","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000029718436,0.00001530321,0.000028297825,0.00004464667,0.000012238717,0.0000075324438,0.99849665,0.00010043048,0.0012650758],"genre_scores_gemma":[0.00027191945,0.000045211164,0.00016516974,0.00003352649,0.0000041003123,0.000039339935,0.9978206,0.0000830075,0.0015370564],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982147,0.00010612104,0.0001719229,0.00029316588,0.00081130315,0.00040276357],"domain_scores_gemma":[0.9921204,0.0008749988,0.00041139065,0.001426783,0.0044482513,0.0007182442],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0013546248,0.0015052438,0.0012785553,0.0075543053,0.0024131818,0.0040864064,0.004005198,0.001581135,0.16418096],"category_scores_gemma":[0.01314965,0.00087185524,0.0009115535,0.023857605,0.00071459304,0.0028355024,0.0026514262,0.0019983933,0.110880435],"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":[0.000009045705,0.0000030097435,0.00028902388,0.00008676489,0.0000049753303,0.0000061337896,0.000011049965,0.00009034348,0.00001092548,0.0004412216,0.99787736,0.0011701337],"study_design_scores_gemma":[0.00004276939,0.0000020559537,0.002982495,0.00016522573,0.0000072140388,0.0000107763735,0.00008237501,0.00018224836,0.00006918463,0.0008197815,0.9956073,0.00002864025],"about_ca_topic_score_codex":0.8321477,"about_ca_topic_score_gemma":0.8463408,"teacher_disagreement_score":0.9959948,"about_ca_system_score_codex":0.010539773,"about_ca_system_score_gemma":0.025719175,"threshold_uncertainty_score":0.54924005},"labels":[],"label_agreement":null},{"id":"W6945621737","doi":"10.25504/fairsharing.802704","title":"FAIRsharing record for: Canadian VirusSeq Data Portal","year":2022,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Sequence (biology); Data collection; Audit","score_opus":0.12299400258664209,"score_gpt":0.33334479280525514,"score_spread":0.21035079021861305,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6945621737","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00005054084,0.00002471498,0.00005062265,0.00009763077,0.000029979028,0.000012807701,0.99716586,0.00052900217,0.0020388674],"genre_scores_gemma":[0.00032999704,0.00004016297,0.00027099135,0.00007184811,0.00000769099,0.000039776838,0.9971091,0.0002605137,0.0018698656],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980242,0.00013271239,0.00010928854,0.00040271974,0.00083811977,0.000492837],"domain_scores_gemma":[0.99336064,0.0006963187,0.00023095858,0.0015655686,0.0031420374,0.0010045717],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017349106,0.0020343175,0.0014392376,0.005208133,0.0035465606,0.0050605885,0.0045338003,0.0018313262,0.22610798],"category_scores_gemma":[0.011734378,0.0010056626,0.0010298103,0.010881019,0.00088808505,0.0027759073,0.0031983943,0.0025600123,0.18192695],"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":[0.000013357767,0.0000029042722,0.00012313941,0.000049856884,0.0000033257372,0.0000035366688,0.000009224813,0.000039020273,0.000023512552,0.00025367827,0.9987525,0.00072598405],"study_design_scores_gemma":[0.000069788126,0.000003388114,0.0015304881,0.00010269456,0.000007294086,0.00001545416,0.00005880068,0.0001887243,0.00016366002,0.00076750177,0.9970601,0.000032062864],"about_ca_topic_score_codex":0.78895307,"about_ca_topic_score_gemma":0.8545738,"teacher_disagreement_score":0.22610798,"about_ca_system_score_codex":0.011695534,"about_ca_system_score_gemma":0.026586741,"threshold_uncertainty_score":0.75640655},"labels":[],"label_agreement":null},{"id":"W6964100610","doi":"10.25504/fairsharing.ea287c","title":"FAIRsharing record for: Canadian Disaster Database","year":2022,"lang":"en","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Disaster research; Information system; Emergency management; Disaster recovery","score_opus":0.0642383614762154,"score_gpt":0.30425015088355456,"score_spread":0.24001178940733917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6964100610","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000030164581,0.000022504233,0.000020701247,0.00007817399,0.000015520702,0.0000091547345,0.99832386,0.000114321134,0.0013856222],"genre_scores_gemma":[0.00035476603,0.00007235797,0.00014428385,0.00005948963,0.0000074207637,0.000046030123,0.9973345,0.00008391903,0.0018972476],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982393,0.00011156677,0.00019102477,0.00033262864,0.00072111527,0.00040440986],"domain_scores_gemma":[0.9918969,0.0008091226,0.00035326226,0.0012813288,0.004831487,0.00082773983],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012399779,0.0015383601,0.0012916162,0.007290699,0.0028030307,0.0045271143,0.0036237717,0.0016491197,0.20940551],"category_scores_gemma":[0.013427226,0.0008001751,0.00092924235,0.01948854,0.0006005012,0.0026491284,0.002547647,0.0017964679,0.11846335],"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":[0.000010407502,0.000002458677,0.00015622476,0.000078473946,0.00000348106,0.0000048962847,0.000008945863,0.000044630506,0.0000069160587,0.00032955976,0.998497,0.0008570336],"study_design_scores_gemma":[0.000055153127,0.0000028429124,0.0027289256,0.00023509916,0.00000862146,0.000015477217,0.000097901335,0.0001817085,0.000064849875,0.0007574748,0.9958211,0.00003074081],"about_ca_topic_score_codex":0.86564857,"about_ca_topic_score_gemma":0.8866239,"teacher_disagreement_score":0.20940551,"about_ca_system_score_codex":0.013020969,"about_ca_system_score_gemma":0.029070495,"threshold_uncertainty_score":0.70053124},"labels":[],"label_agreement":null},{"id":"W7108205655","doi":"10.25504/fairsharing.404b6c","title":"FAIRsharing record for: Canadian Watershed Information Network","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Watershed; Interoperability; Geographic information system; Information sharing; Information system; Spatial analysis","score_opus":0.025258937776557966,"score_gpt":0.2643701847435108,"score_spread":0.23911124696695285,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108205655","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00005012929,0.000021356576,0.000045997447,0.00025843663,0.000060078775,0.000019391919,0.993186,0.00033879103,0.006019804],"genre_scores_gemma":[0.0005292048,0.00007967271,0.00029198764,0.0001326892,0.000019041852,0.000075993936,0.9901635,0.00026011706,0.008447625],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9966948,0.00015318261,0.00023283056,0.00038329908,0.0018837291,0.0006521231],"domain_scores_gemma":[0.9862299,0.0013561486,0.00050305604,0.0027042702,0.007272879,0.0019338217],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0023736763,0.001658452,0.0014946908,0.008023527,0.0040299776,0.007882865,0.004140188,0.0020442223,0.29506895],"category_scores_gemma":[0.021001356,0.0009768556,0.0009573318,0.021159256,0.0010085179,0.004166349,0.0037404206,0.0027258652,0.1756588],"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":[0.000008869364,0.0000025835454,0.000097727316,0.000027363107,0.0000015855101,0.0000027887727,0.0000060221005,0.000036068497,0.000006438861,0.00046106233,0.99838436,0.00096511014],"study_design_scores_gemma":[0.00004767112,0.0000024214658,0.0017446929,0.0001087352,0.00000331417,0.000007694859,0.000059360784,0.00020193764,0.000071112256,0.0011376388,0.99658847,0.000026864642],"about_ca_topic_score_codex":0.78608865,"about_ca_topic_score_gemma":0.8131601,"teacher_disagreement_score":0.9958598,"about_ca_system_score_codex":0.01722777,"about_ca_system_score_gemma":0.036498033,"threshold_uncertainty_score":0.987104},"labels":[],"label_agreement":null},{"id":"W7108206453","doi":"10.25504/fairsharing.c12e1f","title":"FAIRsharing record for: CANUE Data Portal","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Environmental data; Human health; Socioeconomic status; Health data; Data collection; Urban area","score_opus":0.10930819793845793,"score_gpt":0.36049401177852947,"score_spread":0.25118581384007155,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108206453","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00002414222,0.000017528197,0.000030585932,0.00009929841,0.000023909564,0.0000101772175,0.9979723,0.00035253566,0.0014694681],"genre_scores_gemma":[0.00028774492,0.000051646486,0.000203511,0.00009889884,0.000010701854,0.000068521425,0.9966408,0.00033504108,0.0023031037],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99792373,0.0002309821,0.00021031642,0.00040488603,0.0007359451,0.0004941764],"domain_scores_gemma":[0.9897512,0.0016596619,0.00048693354,0.0027042527,0.0040521687,0.0013457437],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0022992033,0.0015792224,0.0015309178,0.006217891,0.002110466,0.0054291957,0.0043205013,0.002469216,0.3393021],"category_scores_gemma":[0.0188286,0.0010713281,0.0011873546,0.014178171,0.00074542384,0.0035729238,0.0043048435,0.00225436,0.24955805],"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":[0.000012958389,0.00000315146,0.0001265731,0.000086500266,0.0000032930495,0.000003257998,0.00000841936,0.00003418166,0.000009162491,0.0003057572,0.9987914,0.0006154419],"study_design_scores_gemma":[0.000070018075,0.000003376384,0.0010618516,0.00016174011,0.000005383553,0.000009066643,0.000053268668,0.00011605159,0.00007369622,0.0008734878,0.9975503,0.000021762049],"about_ca_topic_score_codex":0.39779797,"about_ca_topic_score_gemma":0.47434795,"teacher_disagreement_score":0.9956795,"about_ca_system_score_codex":0.00707187,"about_ca_system_score_gemma":0.012598235,"threshold_uncertainty_score":0.9424054},"labels":[],"label_agreement":null},{"id":"W7108211530","doi":"10.25504/fairsharing.315ed5","title":"FAIRsharing record for: National Forestry Database","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"National forest; Principal (computer security); Forest management; National database; Information system; Data management","score_opus":0.0697100946501172,"score_gpt":0.34802508573918156,"score_spread":0.2783149910890644,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108211530","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000034632172,0.000018588462,0.000029333694,0.00008838179,0.000020546766,0.00000981416,0.99755365,0.00013387737,0.0021111178],"genre_scores_gemma":[0.00033645652,0.000055995846,0.000172186,0.000094912466,0.000011425031,0.000075837415,0.9964042,0.00012219088,0.0027267316],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99807966,0.00018526983,0.0003573834,0.00037339865,0.00069996767,0.00030435016],"domain_scores_gemma":[0.99261606,0.0014257719,0.0006963039,0.0016182467,0.0029715186,0.0006722165],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0018158107,0.0009893775,0.0010298918,0.0058601582,0.001363132,0.0038765802,0.0023814167,0.0017615895,0.31236777],"category_scores_gemma":[0.012962605,0.00079879235,0.000654044,0.0125374505,0.0004135983,0.0033632752,0.0025011816,0.0019505138,0.20010018],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014015634,0.0000049537734,0.00016168994,0.00010865354,0.0000032927449,0.000004157048,0.0000073603233,0.000033463548,0.000011469607,0.00054654345,0.99783546,0.0012690214],"study_design_scores_gemma":[0.00006372576,0.0000040777236,0.0021500322,0.00020245383,0.0000052755267,0.000014608028,0.00004336074,0.00008847269,0.000064000444,0.0012632855,0.99608386,0.000016827067],"about_ca_topic_score_codex":0.07054701,"about_ca_topic_score_gemma":0.08007074,"teacher_disagreement_score":0.99761856,"about_ca_system_score_codex":0.0033682561,"about_ca_system_score_gemma":0.006394171,"threshold_uncertainty_score":0.98082393},"labels":[],"label_agreement":null},{"id":"W7108212262","doi":"10.25504/fairsharing.f711e5","title":"FAIRsharing record for: CanPath Portal","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"General partnership; Epidemiology; Public health; Data collection; Health data","score_opus":0.04065157103187518,"score_gpt":0.31624415401300365,"score_spread":0.2755925829811285,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108212262","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000025123656,0.000018491892,0.000028357814,0.00022517768,0.0000423702,0.000017899862,0.996545,0.00027423244,0.0028233605],"genre_scores_gemma":[0.0006199855,0.00009859834,0.00031610168,0.0004335312,0.000053060976,0.00022500227,0.98924434,0.00046826014,0.008541215],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977412,0.0003210352,0.00029165985,0.0003988008,0.0007276002,0.00051976036],"domain_scores_gemma":[0.98458993,0.003651363,0.0010557864,0.0037406273,0.004712495,0.0022498087],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0034700048,0.0012053457,0.0014073753,0.0059941043,0.0021732883,0.005676063,0.003958444,0.0025488797,0.57371306],"category_scores_gemma":[0.02725614,0.0011888773,0.0012099365,0.013354249,0.00076790754,0.0040144627,0.004677982,0.002183212,0.32103783],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001480718,0.0000026005844,0.00009674722,0.0000832995,0.0000027965275,0.0000028733705,0.000006283097,0.0000150281585,0.0000044719804,0.00024396631,0.99879175,0.00073541136],"study_design_scores_gemma":[0.00018611357,0.0000060321786,0.0015558224,0.00026808967,0.000008364258,0.000014408038,0.000057632096,0.000066268876,0.000052370782,0.0013771039,0.9963825,0.000025350204],"about_ca_topic_score_codex":0.18482634,"about_ca_topic_score_gemma":0.24466991,"teacher_disagreement_score":0.57371306,"about_ca_system_score_codex":0.0055685067,"about_ca_system_score_gemma":0.012509994,"threshold_uncertainty_score":0.60804653},"labels":[],"label_agreement":null},{"id":"W7108213479","doi":"10.25504/fairsharing.e8f037","title":"FAIRsharing record for: CITF DATA PORTAL","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Task (project management); Task force; Data access; Public access; Track (disk drive); Public health; Data sharing","score_opus":0.11756818024426903,"score_gpt":0.36332216462741335,"score_spread":0.24575398438314433,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108213479","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000024810006,0.00001564449,0.000025407466,0.00015473976,0.00003277564,0.000015086941,0.99832183,0.00023566166,0.0011740897],"genre_scores_gemma":[0.00043321282,0.000050183018,0.00025236217,0.0002129239,0.000029630635,0.00015376286,0.9960433,0.0002484172,0.002576281],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.997154,0.00040696794,0.00035808858,0.0006103718,0.0008525518,0.0006181094],"domain_scores_gemma":[0.98152083,0.004548114,0.0013501233,0.0054392274,0.005204251,0.0019375383],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.004187953,0.0013668844,0.0015417374,0.0068575493,0.0020657326,0.0056885956,0.0036349627,0.003504671,0.41128486],"category_scores_gemma":[0.033095263,0.0010951526,0.0012313378,0.013087909,0.00085845013,0.0037687353,0.003969058,0.002341969,0.26939118],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015484882,0.0000030445829,0.0001406551,0.00008511059,0.00000340416,0.000004296307,0.000005955197,0.00002552867,0.000007993321,0.00021346497,0.9988944,0.00060068956],"study_design_scores_gemma":[0.00020038075,0.00000729269,0.0013932827,0.0002492528,0.000007503175,0.000018704197,0.00005636233,0.00009615584,0.0000718024,0.001162894,0.9967096,0.000026831467],"about_ca_topic_score_codex":0.11593762,"about_ca_topic_score_gemma":0.14458904,"teacher_disagreement_score":0.58871514,"about_ca_system_score_codex":0.0055737747,"about_ca_system_score_gemma":0.010336101,"threshold_uncertainty_score":0.8397307},"labels":[],"label_agreement":null},{"id":"W7108213549","doi":"10.25504/fairsharing.e1e2d1","title":"FAIRsharing record for: BioChem database","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Marine fisheries; Data collection; Research data; Data source","score_opus":0.05784035114619293,"score_gpt":0.3336562620587494,"score_spread":0.2758159109125564,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108213549","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00003367893,0.000023615617,0.000022607559,0.00015352128,0.000030985146,0.000006671294,0.997384,0.0003237307,0.002021107],"genre_scores_gemma":[0.00031182496,0.000063676904,0.0001353396,0.00012756986,0.000019950221,0.000042710995,0.99593437,0.00026036616,0.0031041563],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984469,0.00014663023,0.00019565516,0.00040483242,0.00053210155,0.00027381125],"domain_scores_gemma":[0.9941803,0.00096633803,0.00053851894,0.0017367945,0.0017497637,0.0008282398],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.001360803,0.0013970155,0.001268609,0.005200803,0.0018221888,0.005190299,0.0033876286,0.0021249794,0.39771703],"category_scores_gemma":[0.011314501,0.0009683546,0.0008752888,0.011624927,0.00054718426,0.00394745,0.0037865853,0.002122668,0.34489846],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015285807,0.00000349803,0.00010502201,0.000073174466,0.0000026384153,0.0000040105115,0.0000069991333,0.000017354867,0.000016602227,0.0002667661,0.99881446,0.0006742058],"study_design_scores_gemma":[0.000066999506,0.000004104131,0.001076082,0.000091921254,0.0000046558616,0.000012830248,0.000039935378,0.00005225481,0.0000659714,0.00079503923,0.9977769,0.000013274413],"about_ca_topic_score_codex":0.06027026,"about_ca_topic_score_gemma":0.07045976,"teacher_disagreement_score":0.602283,"about_ca_system_score_codex":0.0034621665,"about_ca_system_score_gemma":0.0051376782,"threshold_uncertainty_score":0.85908353},"labels":[],"label_agreement":null},{"id":"W7108213874","doi":"10.25504/fairsharing.dc1f07","title":"FAIRsharing record for: Good Machine Learning Practice for Medical Device Development","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Agency (philosophy); Medical device; Medical practice; Health care; Food and drug administration; Regulatory agency; Medical record","score_opus":0.04618645610084761,"score_gpt":0.352321359727693,"score_spread":0.3061349036268454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108213874","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006335094,0.0003799315,0.00049569947,0.0028840185,0.00060682785,0.00008523065,0.9768092,0.0017420455,0.016363544],"genre_scores_gemma":[0.0025649897,0.0002223708,0.0013473708,0.0009862168,0.00017121177,0.0001684057,0.984304,0.00032665482,0.009908789],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9943151,0.0012834428,0.0005610493,0.00076789537,0.0023893926,0.00068319595],"domain_scores_gemma":[0.9785862,0.0059289355,0.0018579776,0.0061834618,0.0053172996,0.002126085],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.005475646,0.0017810947,0.0012809358,0.004049152,0.0027212196,0.0072236173,0.0038835541,0.0048096045,0.1503247],"category_scores_gemma":[0.030890863,0.00065057585,0.0014544624,0.005715933,0.0010157886,0.00634946,0.0060633156,0.003977994,0.16430727],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024384275,0.000014260205,0.00034904853,0.00006438735,0.0000036230408,0.0000053634453,0.000006758727,0.000056411824,0.00000789142,0.0004007057,0.99669576,0.0023714325],"study_design_scores_gemma":[0.0001919124,0.000030926218,0.0027361668,0.00028232817,0.000010066073,0.00005325577,0.00010076721,0.00053177844,0.000167336,0.0027973666,0.99306446,0.000033594177],"about_ca_topic_score_codex":0.03000222,"about_ca_topic_score_gemma":0.07233665,"teacher_disagreement_score":0.8496753,"about_ca_system_score_codex":0.00328618,"about_ca_system_score_gemma":0.0067896843,"threshold_uncertainty_score":0.50288624},"labels":[],"label_agreement":null},{"id":"W7108215335","doi":"10.25504/fairsharing.e1eb50","title":"FAIRsharing record for: Burnaby's Open Data Portal","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Open data; Check-in; Open government; Range (aeronautics); Open source; Open standard","score_opus":0.16708865644286439,"score_gpt":0.40264336739972045,"score_spread":0.23555471095685607,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108215335","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00013337054,0.00003144324,0.00005475113,0.00027771926,0.000072728035,0.0000195356,0.9945193,0.0006228298,0.0042682705],"genre_scores_gemma":[0.0008150671,0.00006254994,0.00029773684,0.00015405146,0.000029475746,0.00013914019,0.9884979,0.00051740947,0.009486547],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981797,0.00019286636,0.0001409811,0.00030864455,0.0007309718,0.00044690142],"domain_scores_gemma":[0.9912349,0.0013361546,0.00057333393,0.002251027,0.0029755172,0.0016290344],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021677064,0.0012258388,0.0013159792,0.005834205,0.002937702,0.006678076,0.003291191,0.0018565401,0.26614237],"category_scores_gemma":[0.015261043,0.00088241923,0.00076567347,0.011673349,0.00086369796,0.0047597853,0.005227763,0.0027318604,0.21365768],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011590251,0.0000043744967,0.00014456533,0.000039594543,0.0000015383308,0.0000042303673,0.00001463038,0.000019666273,0.000007944415,0.00029586485,0.99860364,0.0008524252],"study_design_scores_gemma":[0.000043738208,0.000003848942,0.001940428,0.00012429508,0.000002506681,0.000009220214,0.00014281903,0.00010426497,0.00006451623,0.00080134283,0.99674207,0.000020896909],"about_ca_topic_score_codex":0.20312434,"about_ca_topic_score_gemma":0.24446446,"teacher_disagreement_score":0.79687566,"about_ca_system_score_codex":0.0038377538,"about_ca_system_score_gemma":0.0066192797,"threshold_uncertainty_score":0.8903349},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W7108218332","doi":"10.25504/fairsharing.bb2d92","title":"FAIRsharing record for: British Columbia Data Catalogue","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Government (linguistics); Publication; Public sector; Web site; Public access","score_opus":0.07781463831934764,"score_gpt":0.3219885897298507,"score_spread":0.24417395141050308,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108218332","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000608841,0.00003929062,0.000024576331,0.0001292734,0.000026495838,0.00001248779,0.99257845,0.00026890394,0.006859546],"genre_scores_gemma":[0.00053754204,0.00013335484,0.0001569432,0.00007623037,0.0000117737945,0.0000711825,0.98089015,0.00031735768,0.017805446],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99825174,0.00009122609,0.0001525423,0.00024044821,0.0008591112,0.00040488646],"domain_scores_gemma":[0.9894042,0.00073235907,0.00038218495,0.0015555427,0.0067463387,0.00117945],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0010961405,0.001499857,0.0016839748,0.0115660075,0.0035683094,0.0066612978,0.0030786316,0.0012779995,0.34901005],"category_scores_gemma":[0.010195466,0.0011251025,0.0006842835,0.03219622,0.0007836896,0.002717078,0.0025632153,0.0023387824,0.29194888],"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":[0.0000075807634,0.0000023727305,0.00009594812,0.000048159596,0.0000015692997,0.0000031191069,0.000007329004,0.00001800008,0.0000070757387,0.00024376427,0.99847764,0.0010874986],"study_design_scores_gemma":[0.000024191047,0.0000017571027,0.0025448266,0.00017606354,0.000004371971,0.000008839342,0.0000842233,0.00006634137,0.00007733896,0.00042197923,0.99657,0.000020075844],"about_ca_topic_score_codex":0.84979993,"about_ca_topic_score_gemma":0.87882453,"teacher_disagreement_score":0.99692136,"about_ca_system_score_codex":0.016873097,"about_ca_system_score_gemma":0.03578735,"threshold_uncertainty_score":0.9285582},"labels":[],"label_agreement":null},{"id":"W7108218962","doi":"10.25504/fairsharing.7041ab","title":"FAIRsharing record for: Pacific Climate Impacts Consortium Data Portal","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Climate change; National weather service; Climate model; Pacific ocean; Climate system; Download","score_opus":0.08113175826975973,"score_gpt":0.34990326200363664,"score_spread":0.2687715037338769,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108218962","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00004744771,0.000011972989,0.000048030684,0.00014280633,0.00003843862,0.000017095092,0.99528176,0.0004802329,0.003932192],"genre_scores_gemma":[0.00053212297,0.00005983293,0.00026319586,0.00011011364,0.000022942946,0.0001240653,0.9923834,0.0005619079,0.0059424005],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986644,0.00012580502,0.00014886157,0.00021094191,0.00057981844,0.00027029868],"domain_scores_gemma":[0.993291,0.0009465857,0.00041714468,0.0018002174,0.00249341,0.0010515698],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0018190182,0.0012155963,0.001096,0.0047252756,0.0014169667,0.0046838294,0.0029964861,0.0014940206,0.38279772],"category_scores_gemma":[0.01204194,0.00091713783,0.00076299306,0.012740503,0.0005436609,0.0045270994,0.00393366,0.0022969192,0.2522837],"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":[0.000012860505,0.0000047248295,0.000121863355,0.00006248024,0.0000025807426,0.0000049943287,0.0000075184976,0.000045198143,0.00001125279,0.0002978552,0.99825746,0.0011712528],"study_design_scores_gemma":[0.00007613116,0.000003251965,0.0018387531,0.00010850872,0.0000044572403,0.000011830966,0.00004806645,0.00011937816,0.000117354895,0.0010995581,0.99655116,0.000021463562],"about_ca_topic_score_codex":0.11156372,"about_ca_topic_score_gemma":0.087296024,"teacher_disagreement_score":0.88843626,"about_ca_system_score_codex":0.0031959598,"about_ca_system_score_gemma":0.0072376514,"threshold_uncertainty_score":0.8803641},"labels":[],"label_agreement":null},{"id":"W7108221972","doi":"10.25504/fairsharing.4727a4","title":"FAIRsharing record for: Canadian Coast Guard - e-Navigation Portal","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Coast guard; Nautical mile; Marine safety; Guard (computer science); Maritime safety","score_opus":0.03419281077533253,"score_gpt":0.29938871541123485,"score_spread":0.2651959046359023,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108221972","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00003998506,0.000028084913,0.00002730713,0.00008464183,0.000027808746,0.000006701869,0.9973828,0.0003271792,0.0020754505],"genre_scores_gemma":[0.00031629542,0.00006447133,0.00017811057,0.000070225324,0.000008951577,0.0000340036,0.99643135,0.00016977884,0.0027268126],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998818,0.00008136734,0.00008648599,0.00025880765,0.00041863992,0.00033670067],"domain_scores_gemma":[0.9939898,0.0007019122,0.00023145536,0.001216797,0.0031216778,0.0007383809],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008693016,0.0022274454,0.001475853,0.0063160583,0.0028699096,0.0041277106,0.0039914176,0.0022949283,0.28232777],"category_scores_gemma":[0.009696472,0.0007456533,0.0010542515,0.015215026,0.0007514521,0.003427652,0.0037105673,0.0019142386,0.23055995],"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":[0.000009337651,0.000002685297,0.00013319377,0.000070232774,0.0000025034326,0.0000037471927,0.000007999098,0.000038846487,0.0000082724955,0.00016917862,0.99868816,0.00086586527],"study_design_scores_gemma":[0.00004287016,0.000003519195,0.0018473377,0.00018793184,0.000007261532,0.000010240518,0.00008948262,0.00018649104,0.00008756762,0.0007556447,0.9967518,0.000029916951],"about_ca_topic_score_codex":0.76822203,"about_ca_topic_score_gemma":0.84336835,"teacher_disagreement_score":0.28232777,"about_ca_system_score_codex":0.00772832,"about_ca_system_score_gemma":0.014706701,"threshold_uncertainty_score":0.9444805},"labels":[],"label_agreement":null},{"id":"W7108222160","doi":"10.25504/fairsharing.8fd23d","title":"FAIRsharing record for: Canadian Urban Data Catalogue","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Metadata; Urban district; Urban planning; Open data; Urban area; Data access","score_opus":0.09322639422575053,"score_gpt":0.32616811538589247,"score_spread":0.23294172116014195,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108222160","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00006662586,0.000037465965,0.00011984661,0.0002613986,0.00007126245,0.00003188532,0.9911418,0.0008456084,0.0074241296],"genre_scores_gemma":[0.00046898032,0.000110454435,0.0004834912,0.000126154,0.000015900789,0.00008547647,0.988792,0.00050523184,0.009412205],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99623495,0.00016901558,0.00026806685,0.00045476737,0.0021339476,0.00073928223],"domain_scores_gemma":[0.9833403,0.0012432048,0.00046363866,0.0032110272,0.009770552,0.0019712807],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024390405,0.0021714142,0.0017881452,0.012631388,0.0054291575,0.008832569,0.0050586015,0.0021053576,0.27202675],"category_scores_gemma":[0.020862805,0.0010524758,0.0011234136,0.029533725,0.00137879,0.0053591593,0.004773716,0.0026146055,0.23391275],"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":[0.0000070737133,0.0000023139232,0.00008049062,0.000040523588,0.0000015987773,0.0000034355714,0.000012455668,0.000029035458,0.000010576463,0.0006014549,0.9978187,0.0013921923],"study_design_scores_gemma":[0.000016946306,0.0000014399484,0.0010328036,0.00009614132,0.0000029659338,0.00000789264,0.00007911613,0.00010965368,0.00007332938,0.00060304435,0.99795425,0.000022261334],"about_ca_topic_score_codex":0.90507704,"about_ca_topic_score_gemma":0.9197637,"teacher_disagreement_score":0.27202675,"about_ca_system_score_codex":0.024118235,"about_ca_system_score_gemma":0.05602473,"threshold_uncertainty_score":0.9100201},"labels":[],"label_agreement":null},{"id":"W7108222311","doi":"10.25504/fairsharing.16ac95","title":"FAIRsharing record for: Data MB","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Transparency (behavior); Accountability; Government (linguistics); Open government; Public access; Data access; Information governance; Open data","score_opus":0.14196214217624067,"score_gpt":0.37526323473571566,"score_spread":0.23330109255947498,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108222311","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00008254034,0.000027528236,0.000095142255,0.0008700386,0.00020215032,0.00004841503,0.98244274,0.0010898636,0.015141652],"genre_scores_gemma":[0.0016870268,0.00011396422,0.0006325274,0.0010613545,0.00011737703,0.000371785,0.95170754,0.0013382268,0.0429703],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99741095,0.00033556245,0.00023860559,0.0004816574,0.0010125088,0.00052069355],"domain_scores_gemma":[0.9833376,0.0026184074,0.0007372977,0.0053279586,0.0058600167,0.0021187214],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0036618293,0.00088583276,0.000997732,0.003821932,0.0024994335,0.006959362,0.0029036729,0.0019675798,0.56219363],"category_scores_gemma":[0.028079487,0.00094767957,0.0007442151,0.009781534,0.0010683103,0.0041271853,0.004519726,0.002345596,0.42058685],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013422158,0.0000031793827,0.000087230896,0.00004247363,0.0000014310607,0.000003030694,0.000010637342,0.000013201194,0.000010453047,0.0005682002,0.9981207,0.0011260749],"study_design_scores_gemma":[0.00006056459,0.0000033282515,0.0011117326,0.00010594986,0.0000018796212,0.000007863368,0.000055339715,0.000040020594,0.00006678695,0.0011615278,0.9973718,0.000013328696],"about_ca_topic_score_codex":0.09059686,"about_ca_topic_score_gemma":0.10074474,"teacher_disagreement_score":0.56219363,"about_ca_system_score_codex":0.004812536,"about_ca_system_score_gemma":0.012982009,"threshold_uncertainty_score":0.6244776},"labels":[],"label_agreement":null},{"id":"W7108226430","doi":"10.25504/fairsharing.b8b8b8","title":"FAIRsharing record for: Saskatchewan Conservation Data Centre","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Biodiversity; Biodiversity conservation; Range (aeronautics); Wildlife conservation","score_opus":0.08931551712024237,"score_gpt":0.3335251852747637,"score_spread":0.24420966815452133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108226430","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000045566016,0.000011937762,0.00003022618,0.00008032029,0.000018444305,0.000010798269,0.9975103,0.00012032177,0.002171992],"genre_scores_gemma":[0.0006472167,0.00006231138,0.0002428697,0.00008958909,0.0000072339867,0.0001101889,0.99289316,0.00016501262,0.0057823877],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986547,0.00012967204,0.00018901784,0.00024617053,0.0004820217,0.00029835422],"domain_scores_gemma":[0.989293,0.0013207293,0.00056085776,0.0020580296,0.005577364,0.0011899266],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001638739,0.0011774416,0.0011364531,0.0065141185,0.0020225872,0.0044598165,0.0029972813,0.0013181594,0.3028929],"category_scores_gemma":[0.013333119,0.00091237156,0.00067611027,0.019686077,0.0005830609,0.00301277,0.0026374087,0.0020846466,0.17207824],"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":[0.000014550278,0.0000034045722,0.00033502173,0.0000769615,0.0000053612084,0.0000057597676,0.000013348508,0.000058033453,0.000009864242,0.00052701234,0.99764246,0.0013082862],"study_design_scores_gemma":[0.00006746672,0.0000028976076,0.0035534278,0.00028473735,0.000007402413,0.0000101022215,0.00011320022,0.00016788134,0.000082947816,0.0013340053,0.9943427,0.000033266082],"about_ca_topic_score_codex":0.66008276,"about_ca_topic_score_gemma":0.7291483,"teacher_disagreement_score":0.33991724,"about_ca_system_score_codex":0.010202773,"about_ca_system_score_gemma":0.024691118,"threshold_uncertainty_score":0.9943387},"labels":[],"label_agreement":null},{"id":"W7108227603","doi":"10.25504/fairsharing.5345f7","title":"FAIRsharing record for: Open Calgary","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Accountability; Key (lock); Work (physics); Best practice","score_opus":0.07284523779408726,"score_gpt":0.35480926246307515,"score_spread":0.2819640246689879,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108227603","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000113520844,0.000045094577,0.000041320975,0.0003790318,0.00009006079,0.00001791323,0.9949694,0.0004450936,0.0038986397],"genre_scores_gemma":[0.00066617207,0.00008744791,0.00018758331,0.00023012227,0.000039132814,0.00010244055,0.9907674,0.00030017836,0.0076195197],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978275,0.00023866243,0.00016647884,0.0003987031,0.0008221461,0.0005464867],"domain_scores_gemma":[0.9891533,0.0016248975,0.000786321,0.00312304,0.0038015298,0.0015108812],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020384174,0.001334917,0.0013970594,0.005500675,0.0027109655,0.007022102,0.0034490987,0.002253813,0.20037155],"category_scores_gemma":[0.016810203,0.0009115613,0.0009328955,0.013229755,0.00090713945,0.0037199936,0.004817695,0.0031208762,0.18516968],"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":[0.000011393331,0.0000030159076,0.0001439286,0.00003804037,0.0000019152096,0.0000037147418,0.000009722647,0.000021726924,0.000005186109,0.00029166302,0.9988041,0.000665621],"study_design_scores_gemma":[0.00007512638,0.000005307114,0.0029160194,0.00017851919,0.000003819962,0.000011606538,0.00013194601,0.00011384821,0.00006358794,0.0009297356,0.9955479,0.000022496235],"about_ca_topic_score_codex":0.23884568,"about_ca_topic_score_gemma":0.29188845,"teacher_disagreement_score":0.7611543,"about_ca_system_score_codex":0.004507032,"about_ca_system_score_gemma":0.010067318,"threshold_uncertainty_score":0.6703096},"labels":[],"label_agreement":null},{"id":"W7108230307","doi":"10.25504/fairsharing.a94677","title":"FAIRsharing record for: Hakai Data","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Margin (machine learning); Data collection; Information system; Historical record; Data system","score_opus":0.14064748246781356,"score_gpt":0.37435965435751095,"score_spread":0.2337121718896974,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108230307","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000035160752,0.000014282465,0.000023860077,0.00008740506,0.000023308537,0.0000078025805,0.9978956,0.0002599676,0.0016524557],"genre_scores_gemma":[0.000434663,0.000036297075,0.00015099686,0.00009316275,0.000013406731,0.000080421974,0.9963595,0.00025767303,0.0025739097],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99816906,0.00019937188,0.00024284657,0.00041643574,0.00057712256,0.000395097],"domain_scores_gemma":[0.99176544,0.0017589177,0.00072277215,0.0024633277,0.0023154705,0.00097409636],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0019842044,0.001137176,0.001134524,0.0055578994,0.0017616726,0.0050044777,0.002993277,0.0020959696,0.36806205],"category_scores_gemma":[0.016430728,0.0009559769,0.00090069044,0.011729612,0.00060494256,0.0041338736,0.0032835589,0.0022361192,0.27095392],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016033577,0.000003802451,0.00017920238,0.00008184423,0.0000034290026,0.0000036651047,0.000008729891,0.000038350874,0.000010090411,0.00036012082,0.99854714,0.00074771844],"study_design_scores_gemma":[0.00009418875,0.0000049504006,0.0022441132,0.00018772767,0.000006986765,0.000014236969,0.00007119333,0.00013081988,0.0000891933,0.0011302942,0.9960031,0.000023181325],"about_ca_topic_score_codex":0.08695283,"about_ca_topic_score_gemma":0.09195835,"teacher_disagreement_score":0.36806205,"about_ca_system_score_codex":0.0036965774,"about_ca_system_score_gemma":0.0057843802,"threshold_uncertainty_score":0.9013828},"labels":[],"label_agreement":null},{"id":"W7108238944","doi":"10.25504/fairsharing.6bacac","title":"FAIRsharing record for: Central Lake Ontario Conservation Open Data Portal","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Download; Natural (archaeology); Flood myth; Open data; Public access; Access to information; Natural resource","score_opus":0.13926967350016684,"score_gpt":0.35263213643828717,"score_spread":0.21336246293812033,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108238944","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00010835953,0.00002556974,0.000061797946,0.00034162652,0.000067551926,0.000030503465,0.9896046,0.0005271498,0.009232832],"genre_scores_gemma":[0.0010961657,0.00008496466,0.00042519826,0.00019909872,0.00003399204,0.00022433742,0.97361034,0.00063003984,0.023695828],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983115,0.00012504112,0.000107432155,0.0002515841,0.0008155338,0.0003888029],"domain_scores_gemma":[0.9888815,0.0013394679,0.0005451004,0.002548365,0.0045178775,0.0021675678],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.002020723,0.001074515,0.0011044844,0.0049156416,0.00328886,0.0061315503,0.0029447973,0.001625678,0.40926144],"category_scores_gemma":[0.014521848,0.0008766961,0.000679689,0.011306252,0.0009770697,0.0042892937,0.0045122807,0.001736621,0.22977774],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009250322,0.000002579796,0.00015092544,0.000033699824,0.0000014586236,0.000003532026,0.000017003322,0.000013718629,0.000007813219,0.00027667408,0.99861586,0.0008673298],"study_design_scores_gemma":[0.000038436676,0.0000020267523,0.0020896255,0.000086762986,0.0000024941799,0.00000707788,0.00010861282,0.00006566269,0.00004354773,0.0005793501,0.9969584,0.000017989532],"about_ca_topic_score_codex":0.600554,"about_ca_topic_score_gemma":0.74281555,"teacher_disagreement_score":0.600554,"about_ca_system_score_codex":0.010971275,"about_ca_system_score_gemma":0.017387228,"threshold_uncertainty_score":0.84261686},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"}]}