{"meta":{"query_hash":"b94c9a481f28","filters":{"venue":"Environmental Protection Agency (EPA) Repository"},"cohort_total":11,"direct_labels_cover":0,"predictions_cover":11,"exported":11,"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/b94c9a481f28","api":"https://metacan.xera.ac/api/v1/cohort?venue=Environmental+Protection+Agency+%28EPA%29+Repository"},"results":[{"id":"W6888797921","doi":"10.23719/1522454","title":"LaKind et al. PFAS and Breast Milk","year":2022,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","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":"Breast milk; Cohort; National Health and Nutrition Examination Survey; Cohort study; Breast feeding; Human breast milk","score_opus":0.009202197852914195,"score_gpt":0.2185709343446224,"score_spread":0.2093687364917082,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888797921","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008053461,0.0016628111,0.00023383772,0.0013735106,0.00028791127,0.00005367108,0.98478913,0.00031302258,0.010480801],"genre_scores_gemma":[0.010346312,0.0046156584,0.0014099858,0.0011191203,0.00036291513,0.00030942482,0.9568475,0.0004314087,0.024557726],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99882966,0.0001726402,0.00014471421,0.0002167079,0.0005073084,0.00012892301],"domain_scores_gemma":[0.99585634,0.0016191433,0.00047221198,0.0005753553,0.0011436689,0.000333149],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018661363,0.00061152526,0.0006236548,0.002516816,0.0004954694,0.0022503382,0.0011875677,0.0009594273,0.18270041],"category_scores_gemma":[0.015696017,0.0004447919,0.00095968647,0.004763993,0.00015890675,0.0011260343,0.0015215469,0.0008599324,0.07150263],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021230648,0.000019306397,0.009184954,0.00082095864,0.000086156455,0.00003415263,0.000042860407,0.00010387231,0.000055211956,0.00083114766,0.95463437,0.033974677],"study_design_scores_gemma":[0.00010526621,0.000023628798,0.01983934,0.0010602917,0.00008242528,0.000117144686,0.00007650776,0.00018308293,0.0001578841,0.0012258841,0.97710663,0.000021790675],"about_ca_topic_score_codex":0.03824142,"about_ca_topic_score_gemma":0.046913583,"teacher_disagreement_score":0.18270041,"about_ca_system_score_codex":0.001023235,"about_ca_system_score_gemma":0.0027968253,"threshold_uncertainty_score":0.6111938},"labels":[],"label_agreement":null},{"id":"W6888862654","doi":"10.23719/1530585","title":"Data Links for Monitoring Chemical Contaminants in the Gulf of Maine, using Sediments and Mussels: an evaluation","year":2024,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","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":"Sediment; Mussel; Ecosystem; Documentation; Environmental monitoring; General partnership; Water quality","score_opus":0.08042101175835435,"score_gpt":0.3315820852807193,"score_spread":0.25116107352236494,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888862654","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13798942,0.0029539436,0.021876829,0.008718977,0.00037885318,0.006213325,0.7351735,0.01842312,0.06827192],"genre_scores_gemma":[0.07888027,0.003564266,0.09415133,0.0017904316,0.00013754521,0.004407168,0.7920313,0.001978385,0.02305931],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9895532,0.0021029937,0.0011724407,0.0009787667,0.005895748,0.0002968751],"domain_scores_gemma":[0.9242592,0.02171952,0.0055479375,0.0101505015,0.034039777,0.004283072],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02277316,0.0010159096,0.0004656556,0.01311221,0.0018154858,0.003987486,0.002803204,0.001894613,0.010579959],"category_scores_gemma":[0.055307273,0.00061271054,0.0006739538,0.013161455,0.0008283349,0.007985645,0.0039654877,0.001074596,0.006931975],"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":[0.0017159604,0.002009535,0.20242433,0.007707754,0.000431219,0.0008523346,0.00428267,0.005180035,0.0077153863,0.014042223,0.35381067,0.39982793],"study_design_scores_gemma":[0.000466823,0.00071417954,0.18361934,0.004026558,0.00034084587,0.0004613098,0.0032289322,0.0067340126,0.012046035,0.0023091736,0.7858701,0.00018269861],"about_ca_topic_score_codex":0.098443635,"about_ca_topic_score_gemma":0.0853882,"teacher_disagreement_score":0.098443635,"about_ca_system_score_codex":0.0035077496,"about_ca_system_score_gemma":0.0086609265,"threshold_uncertainty_score":0.19574124},"labels":[],"label_agreement":null},{"id":"W6888923347","doi":"10.23719/1524702","title":"Density data for Lake Ontario benthic invertebrate assemblages from 1964 to 2018","year":2022,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","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":"Benthic zone; Invertebrate; Trophic level; Taxon; Benthos; Dreissena; Community structure","score_opus":0.038260860208532185,"score_gpt":0.24523752839100246,"score_spread":0.2069766681824703,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888923347","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33005154,0.0011616802,0.0011595325,0.00029100492,0.00008778708,0.00030800578,0.6205862,0.00040703086,0.04594712],"genre_scores_gemma":[0.48049507,0.0018507551,0.0035227865,0.000196799,0.00008549427,0.0011587453,0.45687103,0.000117193624,0.05570205],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993333,0.000020548901,0.000072653056,0.00008973431,0.00036017684,0.00012353528],"domain_scores_gemma":[0.9964418,0.000105944986,0.00093900744,0.00011135632,0.0020959433,0.00030593082],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00024246967,0.00066351047,0.00035205463,0.004732554,0.0013036546,0.00075796584,0.00081514806,0.00021875936,0.0110929385],"category_scores_gemma":[0.001629035,0.00044113916,0.00039415134,0.005543179,0.00026808126,0.00049157516,0.0009201032,0.0003441756,0.0026841098],"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":[0.00025033468,0.000044655186,0.8536231,0.00072038145,0.00017361474,0.0002540585,0.0019480465,0.00058720954,0.0024814864,0.0003447301,0.10126564,0.038306665],"study_design_scores_gemma":[0.00001596833,0.000016063503,0.9370443,0.0000596881,0.000024488401,0.0000627819,0.00033176615,0.000121292585,0.00025866795,0.00002680864,0.062025677,0.000012517582],"about_ca_topic_score_codex":0.8393556,"about_ca_topic_score_gemma":0.9537597,"teacher_disagreement_score":0.16064441,"about_ca_system_score_codex":0.00617648,"about_ca_system_score_gemma":0.0051524835,"threshold_uncertainty_score":0.3231809},"labels":[],"label_agreement":null},{"id":"W6889013070","doi":"10.23719/1527744","title":"Locations and numbers of past producing metal and coal mining projects","year":2022,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","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":"Coal mining; Coal; Underground mining (soft rock); Mining industry","score_opus":0.015002904158296183,"score_gpt":0.21946800403267772,"score_spread":0.20446509987438152,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6889013070","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00013135992,0.000021709506,0.000007782377,0.00001548843,0.00000451612,0.0000025123366,0.9996008,0.00002390819,0.00019187984],"genre_scores_gemma":[0.0005952403,0.00005807508,0.00008055117,0.000020425578,0.0000029665339,0.000022176218,0.998497,0.000016662289,0.00070686697],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99865735,0.00006326728,0.0001478733,0.00032240245,0.00042374377,0.00038538492],"domain_scores_gemma":[0.9949372,0.0008091539,0.0005705789,0.0005700404,0.00244684,0.0006661761],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007763851,0.0023818903,0.0017966955,0.008712941,0.0013354707,0.0027042339,0.0031266138,0.0021470105,0.045261025],"category_scores_gemma":[0.0069446536,0.0011949178,0.0018321284,0.022755329,0.00056665204,0.001326093,0.001755775,0.0020971987,0.035019733],"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":[0.00007641324,0.000026692838,0.004982548,0.00070378167,0.000070490634,0.000029618444,0.000053258977,0.0009279841,0.00006347567,0.0004735021,0.9905911,0.0020010537],"study_design_scores_gemma":[0.0004495651,0.000022983972,0.070935406,0.0009114449,0.000113264185,0.00010073838,0.0005124203,0.0012076191,0.00035040476,0.0009176206,0.92438,0.00009856126],"about_ca_topic_score_codex":0.8065682,"about_ca_topic_score_gemma":0.87436694,"teacher_disagreement_score":0.8065682,"about_ca_system_score_codex":0.006308739,"about_ca_system_score_gemma":0.013486123,"threshold_uncertainty_score":0.38914186},"labels":[],"label_agreement":null},{"id":"W6907421595","doi":"10.23719/1529455","title":"MMWR Metadata","year":2024,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","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; MEDLINE; Identification (biology); Feature (linguistics)","score_opus":0.014151652292667342,"score_gpt":0.2253731245354094,"score_spread":0.21122147224274204,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907421595","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000038950897,0.000021125375,0.00004389947,0.000039105846,0.000010944044,0.000009906185,0.99901736,0.0001328161,0.0006858841],"genre_scores_gemma":[0.00023878092,0.00004302582,0.00024673142,0.00002819881,0.000003811398,0.000040816056,0.99859804,0.000058587535,0.00074201246],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99650526,0.00025978152,0.00044052312,0.0006677671,0.0013052169,0.0008215618],"domain_scores_gemma":[0.9843501,0.0017582644,0.0007777913,0.0020845872,0.009871999,0.00115734],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020631873,0.002294175,0.0016213676,0.01130896,0.0031661657,0.003916302,0.0045580645,0.00175053,0.100664765],"category_scores_gemma":[0.015600385,0.0010935867,0.0016128524,0.026115324,0.0007827138,0.0024891933,0.002893158,0.0025537359,0.06621422],"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":[0.000026928434,0.000007553062,0.00075010693,0.00028524254,0.000013301549,0.000011337343,0.000032871485,0.00016102346,0.00006343552,0.0006590082,0.99593014,0.0020590413],"study_design_scores_gemma":[0.000049783448,0.0000051346624,0.006356995,0.00042223616,0.000026433867,0.000024982583,0.0001993352,0.00029646498,0.00031130682,0.000934039,0.99133307,0.000040182476],"about_ca_topic_score_codex":0.9075991,"about_ca_topic_score_gemma":0.9394021,"teacher_disagreement_score":0.9075991,"about_ca_system_score_codex":0.015779072,"about_ca_system_score_gemma":0.038167294,"threshold_uncertainty_score":0.33675718},"labels":[],"label_agreement":null},{"id":"W6907566281","doi":"10.23719/1529748","title":"Autonomous underwater glider observations in southern Lake Ontario and Niagara River plume","year":2023,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","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":"Underwater glider; Glider; Underwater; Plume; Hydrology (agriculture); Ecosystem","score_opus":0.024405839757807794,"score_gpt":0.20941267840715427,"score_spread":0.18500683864934647,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907566281","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003745081,0.000031021274,0.000030123565,0.000042197906,0.000008447255,0.0000067107217,0.9986803,0.00014144284,0.0006852408],"genre_scores_gemma":[0.0009138005,0.000043381187,0.00014588314,0.000018104884,0.00000254649,0.00003692111,0.9976928,0.00003414745,0.001112363],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99943095,0.000032487296,0.000042909287,0.00010701354,0.0002499615,0.00013671037],"domain_scores_gemma":[0.9982185,0.00015298082,0.00013289822,0.00026041732,0.0010255832,0.00020958312],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004746321,0.001309368,0.00080959353,0.001911511,0.0012083903,0.0012374505,0.002208496,0.0013656938,0.02980832],"category_scores_gemma":[0.0025270889,0.0007041563,0.0008656028,0.006247316,0.0003929044,0.0007389318,0.0013159077,0.0008283755,0.029977147],"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":[0.00004983878,0.000015076851,0.0030184533,0.00029313174,0.000026873857,0.000028001718,0.000054006494,0.00040218822,0.000112558875,0.00023887646,0.99341124,0.0023497574],"study_design_scores_gemma":[0.00033111995,0.000016726393,0.06706676,0.0004330721,0.000044439483,0.00004398953,0.00032534302,0.0017992106,0.00050264027,0.0006316254,0.9287371,0.00006801449],"about_ca_topic_score_codex":0.8748109,"about_ca_topic_score_gemma":0.9337181,"teacher_disagreement_score":0.12518913,"about_ca_system_score_codex":0.0061753476,"about_ca_system_score_gemma":0.009481933,"threshold_uncertainty_score":0.25185275},"labels":[],"label_agreement":null},{"id":"W6926303235","doi":"10.23719/1530881","title":"Datasets for manuscript: Phosphorus recovery in municipal wastewater and socioeconomic impacts in Canada and the United States","year":2024,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","topic":"Microbial Natural Products and Biosynthesis","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Population; Phosphorus; Per capita; Wastewater; Sewage treatment; Water quality; Resource recovery; Water supply","score_opus":0.008159062733175269,"score_gpt":0.20208627146668068,"score_spread":0.1939272087335054,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926303235","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0002229972,0.000033293556,0.000031648942,0.00006902218,0.000008359489,0.000017923583,0.9990922,0.00005961489,0.00046494984],"genre_scores_gemma":[0.002481662,0.00016077365,0.00055453874,0.00007263089,0.000008911521,0.00018840386,0.9954526,0.000040409803,0.0010401728],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99870527,0.00009671098,0.0001566686,0.00020038236,0.00056950893,0.0002713535],"domain_scores_gemma":[0.9880459,0.0012712209,0.00078454305,0.0007614004,0.008402978,0.0007340477],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011076765,0.0011987453,0.0011820524,0.004892667,0.0017108138,0.0028275156,0.002781007,0.0013421471,0.04658011],"category_scores_gemma":[0.012988909,0.00052272034,0.0012331282,0.014784178,0.0004052901,0.001040839,0.0015368244,0.00150895,0.013080488],"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":[0.0000606149,0.0000352024,0.007896518,0.00045912605,0.00007481016,0.00002768528,0.0000513111,0.0011862724,0.000031525597,0.0009809983,0.9840329,0.0051630735],"study_design_scores_gemma":[0.00030355869,0.000019966606,0.061138168,0.001166699,0.00010497702,0.000054649656,0.00059561996,0.0029191046,0.00029834383,0.0017064066,0.93157816,0.00011432735],"about_ca_topic_score_codex":0.9393687,"about_ca_topic_score_gemma":0.9589766,"teacher_disagreement_score":0.060631275,"about_ca_system_score_codex":0.016357835,"about_ca_system_score_gemma":0.0320186,"threshold_uncertainty_score":0.15582597},"labels":[],"label_agreement":null},{"id":"W6926571148","doi":"10.23719/1532366","title":"CMAQ v5.5 Annual 2023 Gridded Predictions Across the US and Canada (v1.0)","year":2025,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","topic":"Aquatic Ecosystems and Phytoplankton Dynamics","field":"Environmental 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":"CMAQ; Atmospheric research; Air quality index; Product (mathematics); Ozone","score_opus":0.0033051974311995233,"score_gpt":0.18808402581225048,"score_spread":0.18477882838105095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926571148","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019252946,0.00032192536,0.0017091539,0.00060476706,0.0002538802,0.00008118677,0.9805033,0.002321413,0.01227901],"genre_scores_gemma":[0.017492369,0.0005836071,0.010666044,0.00041685737,0.000054408254,0.00021446991,0.9591123,0.0014149116,0.010045022],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991398,0.00006466396,0.000036231122,0.00014349994,0.00045426525,0.00016156906],"domain_scores_gemma":[0.9980095,0.000034504686,0.000049887498,0.00009009833,0.0017028452,0.0001132233],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008924107,0.001512746,0.0010125244,0.0017891726,0.0014281541,0.0022474215,0.0031317987,0.0014026706,0.036664195],"category_scores_gemma":[0.0022429554,0.0008955854,0.001792387,0.006134178,0.00043647128,0.0015551917,0.0010830469,0.0018415047,0.02009768],"study_design_candidate":"simulation_or_modeling","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":[0.000054378368,0.000021507542,0.004082302,0.00024487133,0.00010301639,0.00002844169,0.00002991442,0.010211525,0.00021792646,0.0015231757,0.9749102,0.00857291],"study_design_scores_gemma":[0.00029944576,0.000012072809,0.021226889,0.00033361337,0.000108486616,0.00003274708,0.0002676561,0.02397546,0.0012639143,0.0025777186,0.9497397,0.00016229365],"about_ca_topic_score_codex":0.948139,"about_ca_topic_score_gemma":0.9047407,"teacher_disagreement_score":0.05186099,"about_ca_system_score_codex":0.009153122,"about_ca_system_score_gemma":0.022251317,"threshold_uncertainty_score":0.1226539},"labels":[],"label_agreement":null},{"id":"W6945241372","doi":"10.23719/1395247","title":"Great Lakes Proxies Project","year":2018,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","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":"Shapefile; Metadata; Table (database); Population; Field (mathematics); Data source","score_opus":0.017336505239642454,"score_gpt":0.23625436603861483,"score_spread":0.21891786079897238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6945241372","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00026423967,0.00004384336,0.00007401302,0.00011454378,0.00001886192,0.000009221725,0.99762064,0.0003657989,0.0014889217],"genre_scores_gemma":[0.00041399949,0.000041606774,0.00025194293,0.000047898375,0.000006299657,0.00006293739,0.99810815,0.00009128894,0.0009759455],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9987723,0.00026198704,0.00013049522,0.0003555339,0.00033210244,0.00014760796],"domain_scores_gemma":[0.9981889,0.00034747433,0.0001732764,0.00052486424,0.00054960157,0.00021590815],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010224262,0.001397227,0.00085437164,0.0031825092,0.0008797169,0.0020771292,0.0026782535,0.0009606516,0.054713238],"category_scores_gemma":[0.0048103966,0.00053846836,0.00079789694,0.006179754,0.00032980135,0.0014897905,0.002626545,0.0012180228,0.06265643],"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":[0.000028832445,0.000007592582,0.00088864967,0.00014064944,0.000015748723,0.000014523243,0.00002711742,0.000113553244,0.00003640452,0.0005703731,0.99628276,0.0018739082],"study_design_scores_gemma":[0.00005998677,0.000006117267,0.004514853,0.00013750515,0.000013758522,0.000027473898,0.000107438565,0.0002950952,0.00013751643,0.0010685321,0.9936132,0.000018536235],"about_ca_topic_score_codex":0.061118636,"about_ca_topic_score_gemma":0.108484484,"teacher_disagreement_score":0.061118636,"about_ca_system_score_codex":0.0015515035,"about_ca_system_score_gemma":0.0023988604,"threshold_uncertainty_score":0.183034},"labels":[],"label_agreement":null},{"id":"W6963939463","doi":"10.23719/1529114","title":"Geographic Variability, Seasonality, and Increase in ASPCA Animal Poison Control Center Harmful Blue-green Algae Calls 2010-2022","year":2024,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Poison control center; Algae; Poison control; Cruelty; Center (category theory); Animal welfare","score_opus":0.0084232857580886,"score_gpt":0.2098859442953439,"score_spread":0.2014626585372553,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963939463","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018684411,0.000047262136,0.000023270572,0.000096779,0.000012920626,0.000005616702,0.99720556,0.00006923703,0.00067086006],"genre_scores_gemma":[0.0043592704,0.000058978843,0.00013446643,0.000028853548,0.0000076167844,0.000040737938,0.99476105,0.000017747123,0.0005912823],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99924326,0.00008627151,0.00010243178,0.0001853366,0.00023957202,0.00014319147],"domain_scores_gemma":[0.9971951,0.0004996266,0.00055136625,0.0003211463,0.0011706174,0.0002621797],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008721864,0.00073222874,0.00066429854,0.0034140132,0.0006067077,0.0012957625,0.0015996225,0.0007893668,0.011803979],"category_scores_gemma":[0.004933154,0.00040492343,0.0007644082,0.007598836,0.00022750364,0.0007421878,0.0011918378,0.00095157867,0.008408015],"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":[0.00013328932,0.00003754581,0.038394507,0.00039332986,0.00010373634,0.00004158452,0.000053875166,0.00077291677,0.000090430156,0.0004096567,0.9560242,0.0035448438],"study_design_scores_gemma":[0.00042584602,0.00004516526,0.3731774,0.00056742196,0.0001423644,0.00016738541,0.0006256436,0.002792756,0.0005863799,0.000602958,0.62078387,0.000082849154],"about_ca_topic_score_codex":0.31920344,"about_ca_topic_score_gemma":0.35290462,"teacher_disagreement_score":0.31920344,"about_ca_system_score_codex":0.002569736,"about_ca_system_score_gemma":0.0028081776,"threshold_uncertainty_score":0.63469094},"labels":[],"label_agreement":null},{"id":"W6964231116","doi":"10.23719/1518370","title":"BDSNP Module for Improved Soil NO Emission Estimates for CMAQ Model, Conterminous USA","year":2020,"lang":"en","type":"dataset","venue":"Environmental Protection Agency (EPA) Repository","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":"CMAQ; Biogeochemical cycle; Greenhouse gas; Air quality index; Land cover; Land use; Climate model; Soil water","score_opus":0.020770580640409618,"score_gpt":0.2442739479854295,"score_spread":0.22350336734501988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6964231116","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00026679735,0.000016126916,0.00022493623,0.000040825365,0.000014596976,0.0000095425785,0.99823344,0.00043646464,0.0007572574],"genre_scores_gemma":[0.0010353844,0.000021302973,0.0008854414,0.000036339257,0.0000039407364,0.00007624531,0.99725765,0.00014146743,0.0005422711],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996574,0.000056743178,0.00003474488,0.0001014123,0.00009677872,0.000052883603],"domain_scores_gemma":[0.9992617,0.00015042392,0.00007702491,0.00015079565,0.00030283624,0.000057240213],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008256996,0.0015888361,0.0007136063,0.0014170895,0.00039648166,0.0010818492,0.0023743166,0.0012235058,0.05816754],"category_scores_gemma":[0.002032569,0.00072020665,0.0009605367,0.002901393,0.00019653799,0.0013009234,0.0008315302,0.0014297196,0.04322607],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003887137,0.000028821616,0.0018836554,0.00022552177,0.00004022176,0.000020794918,0.000017222206,0.0043678675,0.00013742813,0.0008348012,0.9897843,0.0026205294],"study_design_scores_gemma":[0.00044920377,0.000023740384,0.008702103,0.00023054372,0.000039999384,0.000056283778,0.00008439677,0.014068705,0.0010322557,0.003934392,0.971314,0.00006429501],"about_ca_topic_score_codex":0.06402852,"about_ca_topic_score_gemma":0.07140283,"teacher_disagreement_score":0.06402852,"about_ca_system_score_codex":0.0015530045,"about_ca_system_score_gemma":0.0015674871,"threshold_uncertainty_score":0.19458985},"labels":[],"label_agreement":null}]}