{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":11,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":11,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"b94c9a481f28","filters":{"venue":"Environmental Protection Agency (EPA) Repository"}},"results":[{"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Shapefile; Metadata; Table (database); Population; Field (mathematics); Data source","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.01733650523964245,"gpt":0.2362543660386148,"spread":0.2189178607989724,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001022426,0.001397227,0.0008543716,0.003182509,0.0008797169,0.002077129,0.002678253,0.0009606516,0.05471324],"category_scores_gemma":[0.004810397,0.0005384684,0.0007978969,0.006179754,0.0003298013,0.001489791,0.002626545,0.001218023,0.06265643],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001551504,"about_ca_system_score_gemma":0.00239886,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.06111864,"about_ca_topic_score_gemma":0.1084845,"domain_scores_codex":[0.9987723,0.000261987,0.0001304952,0.0003555339,0.0003321024,0.000147608],"domain_scores_gemma":[0.9981889,0.0003474743,0.0001732764,0.0005248642,0.0005496016,0.0002159081],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"observational","study_design_scores_codex":[0.00002883244,0.000007592582,0.0008886497,0.0001406494,0.00001574872,0.00001452324,0.00002711742,0.0001135532,0.00003640452,0.0005703731,0.9962828,0.001873908],"study_design_scores_gemma":[0.00005998677,0.000006117267,0.004514853,0.0001375052,0.00001375852,0.0000274739,0.0001074386,0.0002950952,0.0001375164,0.001068532,0.9936132,0.00001853624],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0002642397,0.00004384336,0.00007401302,0.0001145438,0.00001886192,0.000009221725,0.9976206,0.0003657989,0.001488922],"genre_scores_gemma":[0.0004139995,0.00004160677,0.0002519429,0.00004789837,0.000006299657,0.00006293739,0.9981081,0.00009128894,0.0009759455],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.06111864,"threshold_uncertainty_score":0.183034,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Breast milk; Cohort; National Health and Nutrition Examination Survey; Cohort study; Breast feeding; Human breast milk","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.009202197852914195,"gpt":0.2185709343446224,"spread":0.2093687364917082,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001866136,0.0006115253,0.0006236548,0.002516816,0.0004954694,0.002250338,0.001187568,0.0009594273,0.1827004],"category_scores_gemma":[0.01569602,0.0004447919,0.0009596865,0.004763993,0.0001589068,0.001126034,0.001521547,0.0008599324,0.07150263],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001023235,"about_ca_system_score_gemma":0.002796825,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03824142,"about_ca_topic_score_gemma":0.04691358,"domain_scores_codex":[0.9988297,0.0001726402,0.0001447142,0.0002167079,0.0005073084,0.000128923],"domain_scores_gemma":[0.9958563,0.001619143,0.000472212,0.0005753553,0.001143669,0.000333149],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002123065,0.0000193064,0.009184954,0.0008209586,0.00008615645,0.00003415263,0.00004286041,0.0001038723,0.00005521196,0.0008311477,0.9546344,0.03397468],"study_design_scores_gemma":[0.0001052662,0.0000236288,0.01983934,0.001060292,0.00008242528,0.0001171447,0.00007650776,0.0001830829,0.0001578841,0.001225884,0.9771066,0.00002179067],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0008053461,0.001662811,0.0002338377,0.001373511,0.0002879113,0.00005367108,0.9847891,0.0003130226,0.0104808],"genre_scores_gemma":[0.01034631,0.004615658,0.001409986,0.00111912,0.0003629151,0.0003094248,0.9568475,0.0004314087,0.02455773],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1827004,"threshold_uncertainty_score":0.6111938,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"CMAQ; Atmospheric research; Air quality index; Product (mathematics); Ozone","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.003305197431199523,"gpt":0.1880840258122505,"spread":0.1847788283810509,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008924107,0.001512746,0.001012524,0.001789173,0.001428154,0.002247422,0.003131799,0.001402671,0.0366642],"category_scores_gemma":[0.002242955,0.0008955854,0.001792387,0.006134178,0.0004364713,0.001555192,0.001083047,0.001841505,0.02009768],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.009153122,"about_ca_system_score_gemma":0.02225132,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.948139,"about_ca_topic_score_gemma":0.9047407,"domain_scores_codex":[0.9991398,0.00006466396,0.00003623112,0.0001434999,0.0004542652,0.0001615691],"domain_scores_gemma":[0.9980095,0.00003450469,0.0000498875,0.00009009833,0.001702845,0.0001132233],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00005437837,0.00002150754,0.004082302,0.0002448713,0.0001030164,0.00002844169,0.00002991442,0.01021152,0.0002179265,0.001523176,0.9749102,0.00857291],"study_design_scores_gemma":[0.0002994458,0.00001207281,0.02122689,0.0003336134,0.0001084866,0.00003274708,0.0002676561,0.02397546,0.001263914,0.002577719,0.9497397,0.0001622937],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.001925295,0.0003219254,0.001709154,0.0006047671,0.0002538802,0.00008118677,0.9805033,0.002321413,0.01227901],"genre_scores_gemma":[0.01749237,0.0005836071,0.01066604,0.0004168574,0.00005440825,0.0002144699,0.9591123,0.001414912,0.01004502],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.05186099,"threshold_uncertainty_score":0.1226539,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Population; Phosphorus; Per capita; Wastewater; Sewage treatment; Water quality; Resource recovery; Water supply","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.008159062733175269,"gpt":0.2020862714666807,"spread":0.1939272087335054,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001107677,0.001198745,0.001182052,0.004892667,0.001710814,0.002827516,0.002781007,0.001342147,0.04658011],"category_scores_gemma":[0.01298891,0.0005227203,0.001233128,0.01478418,0.0004052901,0.001040839,0.001536824,0.00150895,0.01308049],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01635784,"about_ca_system_score_gemma":0.0320186,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.9393687,"about_ca_topic_score_gemma":0.9589766,"domain_scores_codex":[0.9987053,0.00009671098,0.0001566686,0.0002003824,0.0005695089,0.0002713535],"domain_scores_gemma":[0.9880459,0.001271221,0.0007845431,0.0007614004,0.008402978,0.0007340477],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000606149,0.0000352024,0.007896518,0.0004591261,0.00007481016,0.00002768528,0.0000513111,0.001186272,0.0000315256,0.0009809983,0.9840329,0.005163074],"study_design_scores_gemma":[0.0003035587,0.00001996661,0.06113817,0.001166699,0.000104977,0.00005464966,0.00059562,0.002919105,0.0002983438,0.001706407,0.9315782,0.0001143273],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0002229972,0.00003329356,0.00003164894,0.00006902218,0.000008359489,0.00001792358,0.9990922,0.00005961489,0.0004649498],"genre_scores_gemma":[0.002481662,0.0001607736,0.0005545387,0.00007263089,0.000008911521,0.0001884039,0.9954526,0.0000404098,0.001040173],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.06063128,"threshold_uncertainty_score":0.155826,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"CMAQ; Biogeochemical cycle; Greenhouse gas; Air quality index; Land cover; Land use; Climate model; Soil water","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02077058064040962,"gpt":0.2442739479854295,"spread":0.2235033673450199,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008256996,0.001588836,0.0007136063,0.001417089,0.0003964817,0.001081849,0.002374317,0.001223506,0.05816754],"category_scores_gemma":[0.002032569,0.0007202066,0.0009605367,0.002901393,0.000196538,0.001300923,0.0008315302,0.00142972,0.04322607],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001553004,"about_ca_system_score_gemma":0.001567487,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.06402852,"about_ca_topic_score_gemma":0.07140283,"domain_scores_codex":[0.9996574,0.00005674318,0.00003474488,0.0001014123,0.00009677872,0.0000528836],"domain_scores_gemma":[0.9992617,0.0001504239,0.00007702491,0.0001507956,0.0003028362,0.00005724021],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00003887137,0.00002882162,0.001883655,0.0002255218,0.00004022176,0.00002079492,0.00001722221,0.004367867,0.0001374281,0.0008348012,0.9897843,0.002620529],"study_design_scores_gemma":[0.0004492038,0.00002374038,0.008702103,0.0002305437,0.00003999938,0.00005628378,0.00008439677,0.01406871,0.001032256,0.003934392,0.971314,0.00006429501],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0002667973,0.00001612692,0.0002249362,0.00004082537,0.00001459698,0.000009542578,0.9982334,0.0004364646,0.0007572574],"genre_scores_gemma":[0.001035384,0.00002130297,0.0008854414,0.00003633926,0.000003940736,0.00007624531,0.9972576,0.0001414674,0.0005422711],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.06402852,"threshold_uncertainty_score":0.1945899,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Poison control center; Algae; Poison control; Cruelty; Center (category theory); Animal welfare","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.0084232857580886,"gpt":0.2098859442953439,"spread":0.2014626585372553,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008721864,0.0007322287,0.0006642985,0.003414013,0.0006067077,0.001295762,0.001599623,0.0007893668,0.01180398],"category_scores_gemma":[0.004933154,0.0004049234,0.0007644082,0.007598836,0.0002275036,0.0007421878,0.001191838,0.0009515787,0.008408015],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002569736,"about_ca_system_score_gemma":0.002808178,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.3192034,"about_ca_topic_score_gemma":0.3529046,"domain_scores_codex":[0.9992433,0.00008627151,0.0001024318,0.0001853366,0.000239572,0.0001431915],"domain_scores_gemma":[0.9971951,0.0004996266,0.0005513662,0.0003211463,0.001170617,0.0002621797],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"observational","study_design_scores_codex":[0.0001332893,0.00003754581,0.03839451,0.0003933299,0.0001037363,0.00004158452,0.00005387517,0.0007729168,0.00009043016,0.0004096567,0.9560242,0.003544844],"study_design_scores_gemma":[0.000425846,0.00004516526,0.3731774,0.000567422,0.0001423644,0.0001673854,0.0006256436,0.002792756,0.0005863799,0.000602958,0.6207839,0.00008284915],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.001868441,0.00004726214,0.00002327057,0.000096779,0.00001292063,0.000005616702,0.9972056,0.00006923703,0.0006708601],"genre_scores_gemma":[0.00435927,0.00005897884,0.0001344664,0.00002885355,0.000007616784,0.00004073794,0.994761,0.00001774712,0.0005912823],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.3192034,"threshold_uncertainty_score":0.6346909,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Underwater glider; Glider; Underwater; Plume; Hydrology (agriculture); Ecosystem","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02440583975780779,"gpt":0.2094126784071543,"spread":0.1850068386493465,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004746321,0.001309368,0.0008095935,0.001911511,0.00120839,0.001237451,0.002208496,0.001365694,0.02980832],"category_scores_gemma":[0.002527089,0.0007041563,0.0008656028,0.006247316,0.0003929044,0.0007389318,0.001315908,0.0008283755,0.02997715],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006175348,"about_ca_system_score_gemma":0.009481933,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.8748109,"about_ca_topic_score_gemma":0.9337181,"domain_scores_codex":[0.999431,0.0000324873,0.00004290929,0.0001070135,0.0002499615,0.0001367104],"domain_scores_gemma":[0.9982185,0.0001529808,0.0001328982,0.0002604173,0.001025583,0.0002095831],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"observational","study_design_scores_codex":[0.00004983878,0.00001507685,0.003018453,0.0002931317,0.00002687386,0.00002800172,0.00005400649,0.0004021882,0.0001125589,0.0002388765,0.9934112,0.002349757],"study_design_scores_gemma":[0.00033112,0.00001672639,0.06706676,0.0004330721,0.00004443948,0.00004398953,0.000325343,0.001799211,0.0005026403,0.0006316254,0.9287371,0.00006801449],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0003745081,0.00003102127,0.00003012357,0.00004219791,0.000008447255,0.000006710722,0.9986803,0.0001414428,0.0006852408],"genre_scores_gemma":[0.0009138005,0.00004338119,0.0001458831,0.00001810488,0.00000254649,0.00003692111,0.9976928,0.00003414745,0.001112363],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1251891,"threshold_uncertainty_score":0.2518528,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Metadata; MEDLINE; Identification (biology); Feature (linguistics)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.01415165229266734,"gpt":0.2253731245354094,"spread":0.211221472242742,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002063187,0.002294175,0.001621368,0.01130896,0.003166166,0.003916302,0.004558065,0.00175053,0.1006648],"category_scores_gemma":[0.01560039,0.001093587,0.001612852,0.02611532,0.0007827138,0.002489193,0.002893158,0.002553736,0.06621422],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01577907,"about_ca_system_score_gemma":0.03816729,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.9075991,"about_ca_topic_score_gemma":0.9394021,"domain_scores_codex":[0.9965053,0.0002597815,0.0004405231,0.0006677671,0.001305217,0.0008215618],"domain_scores_gemma":[0.9843501,0.001758264,0.0007777913,0.002084587,0.009871999,0.00115734],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00002692843,0.000007553062,0.0007501069,0.0002852425,0.00001330155,0.00001133734,0.00003287149,0.0001610235,0.00006343552,0.0006590082,0.9959301,0.002059041],"study_design_scores_gemma":[0.00004978345,0.000005134662,0.006356995,0.0004222362,0.00002643387,0.00002498258,0.0001993352,0.000296465,0.0003113068,0.000934039,0.9913331,0.00004018248],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0000389509,0.00002112538,0.00004389947,0.00003910585,0.00001094404,0.000009906185,0.9990174,0.0001328161,0.0006858841],"genre_scores_gemma":[0.0002387809,0.00004302582,0.0002467314,0.00002819881,0.000003811398,0.00004081606,0.998598,0.00005858754,0.0007420125],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.9075991,"threshold_uncertainty_score":0.3367572,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Sediment; Mussel; Ecosystem; Documentation; Environmental monitoring; General partnership; Water quality","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.08042101175835435,"gpt":0.3315820852807193,"spread":0.2511610735223649,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02277316,0.00101591,0.0004656556,0.01311221,0.001815486,0.003987486,0.002803204,0.001894613,0.01057996],"category_scores_gemma":[0.05530727,0.0006127105,0.0006739538,0.01316146,0.0008283349,0.007985645,0.003965488,0.001074596,0.006931975],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00350775,"about_ca_system_score_gemma":0.008660926,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.09844363,"about_ca_topic_score_gemma":0.0853882,"domain_scores_codex":[0.9895532,0.002102994,0.001172441,0.0009787667,0.005895748,0.0002968751],"domain_scores_gemma":[0.9242592,0.02171952,0.005547937,0.0101505,0.03403978,0.004283072],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00171596,0.002009535,0.2024243,0.007707754,0.000431219,0.0008523346,0.00428267,0.005180035,0.007715386,0.01404222,0.3538107,0.3998279],"study_design_scores_gemma":[0.000466823,0.0007141795,0.1836193,0.004026558,0.0003408459,0.0004613098,0.003228932,0.006734013,0.01204604,0.002309174,0.7858701,0.0001826986],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.1379894,0.002953944,0.02187683,0.008718977,0.0003788532,0.006213325,0.7351735,0.01842312,0.06827192],"genre_scores_gemma":[0.07888027,0.003564266,0.09415133,0.001790432,0.0001375452,0.004407168,0.7920313,0.001978385,0.02305931],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.09844363,"threshold_uncertainty_score":0.1957412,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Benthic zone; Invertebrate; Trophic level; Taxon; Benthos; Dreissena; Community structure","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03826086020853219,"gpt":0.2452375283910025,"spread":0.2069766681824703,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002424697,0.0006635105,0.0003520546,0.004732554,0.001303655,0.0007579658,0.0008151481,0.0002187594,0.01109294],"category_scores_gemma":[0.001629035,0.0004411392,0.0003941513,0.005543179,0.0002680813,0.0004915752,0.0009201032,0.0003441756,0.00268411],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00617648,"about_ca_system_score_gemma":0.005152483,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.8393556,"about_ca_topic_score_gemma":0.9537597,"domain_scores_codex":[0.9993333,0.0000205489,0.00007265306,0.00008973431,0.0003601768,0.0001235353],"domain_scores_gemma":[0.9964418,0.000105945,0.0009390074,0.0001113563,0.002095943,0.0003059308],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002503347,0.00004465519,0.8536231,0.0007203814,0.0001736147,0.0002540585,0.001948046,0.0005872095,0.002481486,0.0003447301,0.1012656,0.03830666],"study_design_scores_gemma":[0.00001596833,0.0000160635,0.9370443,0.0000596881,0.0000244884,0.0000627819,0.0003317661,0.0001212926,0.0002586679,0.00002680864,0.06202568,0.00001251758],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.3300515,0.00116168,0.001159532,0.0002910049,0.00008778708,0.0003080058,0.6205862,0.0004070309,0.04594712],"genre_scores_gemma":[0.4804951,0.001850755,0.003522787,0.000196799,0.00008549427,0.001158745,0.456871,0.0001171936,0.05570205],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1606444,"threshold_uncertainty_score":0.3231809,"prediction_status":"machine_predicted_unvalidated"},"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Coal mining; Coal; Underground mining (soft rock); Mining industry","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.01500290415829618,"gpt":0.2194680040326777,"spread":0.2044650998743815,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007763851,0.00238189,0.001796696,0.008712941,0.001335471,0.002704234,0.003126614,0.002147011,0.04526103],"category_scores_gemma":[0.006944654,0.001194918,0.001832128,0.02275533,0.000566652,0.001326093,0.001755775,0.002097199,0.03501973],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006308739,"about_ca_system_score_gemma":0.01348612,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.8065682,"about_ca_topic_score_gemma":0.8743669,"domain_scores_codex":[0.9986573,0.00006326728,0.0001478733,0.0003224025,0.0004237438,0.0003853849],"domain_scores_gemma":[0.9949372,0.0008091539,0.0005705789,0.0005700404,0.00244684,0.0006661761],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007641324,0.00002669284,0.004982548,0.0007037817,0.00007049063,0.00002961844,0.00005325898,0.0009279841,0.00006347567,0.0004735021,0.9905911,0.002001054],"study_design_scores_gemma":[0.0004495651,0.00002298397,0.07093541,0.0009114449,0.0001132642,0.0001007384,0.0005124203,0.001207619,0.0003504048,0.0009176206,0.92438,0.00009856126],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0001313599,0.00002170951,0.000007782377,0.00001548843,0.00000451612,0.000002512337,0.9996008,0.00002390819,0.0001918798],"genre_scores_gemma":[0.0005952403,0.00005807508,0.00008055117,0.00002042558,0.000002966534,0.00002217622,0.998497,0.00001666229,0.000706867],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.8065682,"threshold_uncertainty_score":0.3891419,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}