{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":67,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":67,"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":"1da4e2f7a9b5","filters":{"venue":"Neotoma Paleoecological Database"}},"results":[{"id":"W6963807274","doi":"10.21233/107z-5606","title":"Lake Ontario/South Bay water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02719083757317186,"gpt":0.269282397587167,"spread":0.2420915600139951,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001724257,0.002364735,0.002172253,0.0003042237,0.0006635774,0.0004723287,0.004708751,0.00242909,0.3748726],"category_scores_gemma":[0.001101353,0.001648566,0.0005141211,0.0004273225,0.001588966,0.0005746108,0.004908719,0.003926989,0.1796512],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008356023,"about_ca_system_score_gemma":0.0007037486,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001935672,"about_ca_topic_score_gemma":0.482585,"domain_scores_codex":[0.9886046,0.0006003718,0.001803901,0.003978991,0.001893928,0.003118209],"domain_scores_gemma":[0.9893698,0.0004175398,0.0007800277,0.007606403,0.0003011733,0.001525052],"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":[0.0008269427,0.001198619,0.0009724835,0.0004036709,0.0002870446,0.005155888,0.00003311578,0.000001220823,0.0005452579,4.526173e-7,0.9905606,0.00001473218],"study_design_scores_gemma":[0.002040183,0.0003955007,0.002012826,0.0001507667,0.0007612371,0.0003167842,0.00001447832,0.000004061302,0.0007983363,0.00001517607,0.991073,0.00241765],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01277294,0.00003363864,0.000002257877,0.00005426454,0.001080362,0.001682458,0.9833072,0.0005451132,0.0005217016],"genre_scores_gemma":[0.000101262,0.00002619999,0.0004597607,0.001416734,0.001878851,0.00074658,0.9937628,0.0002083292,0.001399471],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4806493,"threshold_uncertainty_score":0.9989091,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963483783","doi":"10.21233/n3s45f","title":"Point Escuminac pollen dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Pollen; Point (geometry); Raw data; Vegetation (pathology)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04605927199852503,"gpt":0.3542656623959299,"spread":0.3082063903974049,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002387744,0.002281454,0.002656505,0.0007626966,0.001438624,0.0008253356,0.008367137,0.002070065,0.03996599],"category_scores_gemma":[0.00721423,0.001883717,0.0006315003,0.0004029702,0.001767615,0.001193503,0.006312891,0.003670398,0.2360224],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005936867,"about_ca_system_score_gemma":0.0006696231,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008942425,"about_ca_topic_score_gemma":0.006602822,"domain_scores_codex":[0.9886877,0.001151715,0.001723937,0.003743968,0.001998692,0.002694001],"domain_scores_gemma":[0.9813801,0.001139279,0.002072418,0.01357122,0.0002424973,0.001594499],"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":[0.0007334176,0.001523571,0.0001347205,0.0003453773,0.0002551837,0.01178704,0.000003138148,9.715585e-7,0.0000845012,0.00002525785,0.984728,0.0003788238],"study_design_scores_gemma":[0.002327163,0.0006911451,0.003564869,0.0002565732,0.0007878107,0.0004654179,0.00002211971,0.0000159237,0.00003243147,0.00007858791,0.9894381,0.002319824],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0004337065,0.0004694367,0.000004924047,0.0002781319,0.001741116,0.002779141,0.9930574,0.0005460361,0.0006900811],"genre_scores_gemma":[0.00007216639,0.0005636889,0.0007554234,0.00161681,0.001522566,0.0009433305,0.9938745,0.0002137759,0.0004376788],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1960564,"threshold_uncertainty_score":0.9998614,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888548473","doi":"10.21233/n3sp7q","title":"Edward Lake pollen dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Royal Ontario Museum","funders":"","keywords":"Pollen; Vegetation (pathology); Paleoecology; Taxon","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04735042579327303,"gpt":0.3511017711469031,"spread":0.3037513453536301,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002018978,0.002177425,0.002564887,0.0006987868,0.001402737,0.0008579714,0.008560527,0.002154123,0.05796479],"category_scores_gemma":[0.006408257,0.001815134,0.0005851588,0.0003972008,0.001617363,0.0009944306,0.005737532,0.003651462,0.2604429],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003060314,"about_ca_system_score_gemma":0.0006938559,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003608048,"about_ca_topic_score_gemma":0.04128424,"domain_scores_codex":[0.9891032,0.001085072,0.00159684,0.003565141,0.00203567,0.002614108],"domain_scores_gemma":[0.9826053,0.001103103,0.001940095,0.01258467,0.0002348868,0.001531941],"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":[0.0006500292,0.001294327,0.0003298938,0.000305575,0.0002735393,0.009990248,0.000002323301,0.000001324903,0.00003310603,0.00001262753,0.9866907,0.0004163074],"study_design_scores_gemma":[0.002137701,0.0005297887,0.005695471,0.0002087286,0.0007429614,0.0003220185,0.000009412188,0.00001015072,0.00001076972,0.00004715916,0.9880834,0.002202505],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0002253092,0.00035262,0.000003470045,0.0002238627,0.002058644,0.002462955,0.9932089,0.0005993127,0.0008648827],"genre_scores_gemma":[0.00004012356,0.0005961725,0.0006080713,0.00176066,0.001947477,0.0008740361,0.993557,0.000205129,0.000411266],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.2024781,"threshold_uncertainty_score":0.9998973,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6925831778","doi":"10.21233/sne3-1m81","title":"Ottawa River ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Scientific Computing and Data Management","field":"Decision 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":"Sample (material); Paleoecology; Hydrology (agriculture); Surface (topology); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.1328001093702441,"gpt":0.3985061333075233,"spread":0.2657060239372792,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.01179217,0.0009449407,0.0012821,0.0005697092,0.0008517064,0.001641801,0.009165246,0.0006186506,0.03817596],"category_scores_gemma":[0.02724075,0.0006710078,0.0003420901,0.001760745,0.001401314,0.000710668,0.007130299,0.001081578,0.03175405],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001320569,"about_ca_system_score_gemma":0.0002407464,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001111334,"about_ca_topic_score_gemma":0.004432375,"domain_scores_codex":[0.9864129,0.00126458,0.002193425,0.004414112,0.004296172,0.00141883],"domain_scores_gemma":[0.9779808,0.009306355,0.001216969,0.0102071,0.000476448,0.0008123369],"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":[0.00007762996,0.0004644563,0.0003930387,0.00003181451,0.00005014392,0.0005170711,0.000007569263,0.00004159787,0.000005090372,0.00002785939,0.99673,0.001653731],"study_design_scores_gemma":[0.0005512821,0.0002851828,0.0033046,0.0000443642,0.0001210959,0.00002798302,0.00003903461,0.0005766785,0.000008527851,0.0009136006,0.9932765,0.0008511567],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002076406,0.00007204244,0.001073083,0.0003166482,0.004608777,0.0009357928,0.990507,0.000176323,0.0002338944],"genre_scores_gemma":[0.0001146033,0.00006308632,0.008654604,0.001861907,0.0007945149,0.00003702918,0.9878815,0.00002373346,0.0005689852],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.01544857,"threshold_uncertainty_score":0.9995741,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888442610","doi":"10.21233/q2vf-rm65","title":"Edmonton District vertebrate fauna dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Fauna; Paleoecology; Vertebrate; Biodiversity; JSON","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03114831338250275,"gpt":0.3062672350064258,"spread":0.275118921623923,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001671473,0.002348236,0.002304317,0.0007162517,0.0007736587,0.000536187,0.005237333,0.001967254,0.04629781],"category_scores_gemma":[0.003724789,0.001930785,0.0005099145,0.001491143,0.001613699,0.0009161243,0.003875373,0.002854793,0.1680946],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007001295,"about_ca_system_score_gemma":0.0005107718,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005593001,"about_ca_topic_score_gemma":0.006954991,"domain_scores_codex":[0.987905,0.001415174,0.001921049,0.003910129,0.002089852,0.002758773],"domain_scores_gemma":[0.9879386,0.001140662,0.001357987,0.007637545,0.0003199996,0.00160519],"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":[0.0008243755,0.001744197,0.00021523,0.000285789,0.0002540595,0.004937793,0.000002497274,0.000001134423,0.0001158415,0.00001383706,0.9913003,0.0003049723],"study_design_scores_gemma":[0.002018393,0.001178742,0.004175974,0.0001638611,0.0007879474,0.0002117524,0.000009287269,0.00009551543,0.0000642,0.00004273284,0.9889095,0.002342106],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.001054451,0.000342325,0.00001081099,0.00007318785,0.002354731,0.002404757,0.9927003,0.0007237634,0.0003357198],"genre_scores_gemma":[0.00008773249,0.0005135558,0.00048411,0.001772818,0.00268825,0.0008134108,0.993315,0.0002092885,0.0001158348],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1217968,"threshold_uncertainty_score":0.9994457,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888554482","doi":"10.21233/gt6w-t875","title":"Toronto Moose vertebrate fauna dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Fauna; Vertebrate; Paleoecology; Invertebrate","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02905637955873201,"gpt":0.3173664998443533,"spread":0.2883101202856213,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","insufficient_payload"],"category_scores_codex":[0.001838687,0.00236567,0.002334256,0.0003181972,0.0007093677,0.0004840694,0.005337104,0.00206356,0.1260065],"category_scores_gemma":[0.002752746,0.001983398,0.0004815607,0.000657075,0.00144884,0.001552133,0.004285459,0.002053555,0.1722639],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001276969,"about_ca_system_score_gemma":0.0005324591,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002232152,"about_ca_topic_score_gemma":0.06659827,"domain_scores_codex":[0.9879164,0.001446525,0.001890515,0.003937514,0.002018333,0.002790696],"domain_scores_gemma":[0.9874677,0.0009238233,0.001192974,0.008353155,0.0003503417,0.001711981],"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":[0.000916807,0.001762502,0.00008470698,0.0002714931,0.0002908847,0.003913342,0.000005536085,8.989227e-7,0.0001371538,0.00002050403,0.9923462,0.0002499961],"study_design_scores_gemma":[0.002135902,0.001262992,0.001079174,0.0001618393,0.0007041416,0.0002145544,0.00002487838,0.00008993915,0.00006051509,0.00004933747,0.9918022,0.002414514],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0006447618,0.0008204249,0.000005337688,0.00005095618,0.002954113,0.002512543,0.9917014,0.0007615031,0.0005489643],"genre_scores_gemma":[0.0000414064,0.0008529029,0.0008712673,0.002843899,0.002896091,0.0009335381,0.9911748,0.0002347316,0.0001513902],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.06436612,"threshold_uncertainty_score":0.999232,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888610478","doi":"10.21233/vdm6-bp79","title":"Lake Ontario/South Bay water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02719083757317186,"gpt":0.269282397587167,"spread":0.2420915600139951,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001724257,0.002364735,0.002172253,0.0003042237,0.0006635774,0.0004723287,0.004708751,0.00242909,0.3748726],"category_scores_gemma":[0.001101353,0.001648566,0.0005141211,0.0004273225,0.001588966,0.0005746108,0.004908719,0.003926989,0.1796512],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008356023,"about_ca_system_score_gemma":0.0007037486,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001935672,"about_ca_topic_score_gemma":0.482585,"domain_scores_codex":[0.9886046,0.0006003718,0.001803901,0.003978991,0.001893928,0.003118209],"domain_scores_gemma":[0.9893698,0.0004175398,0.0007800277,0.007606403,0.0003011733,0.001525052],"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":[0.0008269427,0.001198619,0.0009724835,0.0004036709,0.0002870446,0.005155888,0.00003311578,0.000001220823,0.0005452579,4.526173e-7,0.9905606,0.00001473218],"study_design_scores_gemma":[0.002040183,0.0003955007,0.002012826,0.0001507667,0.0007612371,0.0003167842,0.00001447832,0.000004061302,0.0007983363,0.00001517607,0.991073,0.00241765],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01277294,0.00003363864,0.000002257877,0.00005426454,0.001080362,0.001682458,0.9833072,0.0005451132,0.0005217016],"genre_scores_gemma":[0.000101262,0.00002619999,0.0004597607,0.001416734,0.001878851,0.00074658,0.9937628,0.0002083292,0.001399471],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4806493,"threshold_uncertainty_score":0.9989091,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6944791880","doi":"10.21233/n3t990","title":"Banks Island 74MS15 pollen dataset","year":2005,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University","funders":"","keywords":"Pollen; Apoidea; Vegetation (pathology); Aerial survey","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02883348087179219,"gpt":0.3106010849059356,"spread":0.2817676040341434,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.00171974,0.002330497,0.002459807,0.0009271151,0.0006291303,0.0004463168,0.004776833,0.002312268,0.1088429],"category_scores_gemma":[0.002437233,0.001939618,0.0005279366,0.001074378,0.0009990945,0.0008732552,0.003487477,0.004016298,0.1951878],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006183565,"about_ca_system_score_gemma":0.0005603363,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004853959,"about_ca_topic_score_gemma":0.01920855,"domain_scores_codex":[0.9882407,0.001237764,0.001991934,0.003595985,0.002118031,0.002815549],"domain_scores_gemma":[0.9886849,0.00145835,0.001191711,0.006960162,0.0002006012,0.001504256],"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":[0.0007479442,0.001871626,0.0001407923,0.0002301225,0.0002625267,0.004870047,0.000004253803,0.000008931312,0.0000689866,0.00001567081,0.9907694,0.001009732],"study_design_scores_gemma":[0.002846071,0.0006004063,0.002990338,0.0001440796,0.0007345463,0.0004312312,0.00001296652,0.00004397381,0.00001888084,0.00001988482,0.9897819,0.002375701],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0004602699,0.0009673234,0.000007527604,0.0002628027,0.001068924,0.002537239,0.9932478,0.0006143866,0.0008336907],"genre_scores_gemma":[0.00006215012,0.0009310746,0.0009281207,0.003660614,0.002844518,0.0007690847,0.9901858,0.0002256383,0.0003930275],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.08634489,"threshold_uncertainty_score":0.998983,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907246276","doi":"10.21233/7gnz-f732","title":"Site 2, Georgian Bay, Canadian Hydrographic Service pollen surface sample dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03591422543043152,"gpt":0.2856226842979104,"spread":0.2497084588674789,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002354617,0.002760481,0.002782017,0.001667922,0.001021188,0.0006694228,0.006068808,0.002730087,0.04860902],"category_scores_gemma":[0.002083483,0.002598034,0.0005719613,0.003645174,0.000656125,0.001008938,0.002827033,0.004229032,0.2017946],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00117203,"about_ca_system_score_gemma":0.002409409,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.5492137,"about_ca_topic_score_gemma":0.9737651,"domain_scores_codex":[0.9849961,0.00174804,0.001945266,0.004584509,0.002084731,0.004641292],"domain_scores_gemma":[0.9817774,0.002635971,0.001154721,0.009756253,0.0004524707,0.00422324],"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":[0.000367992,0.0007075169,0.005892213,0.0005826285,0.0003531835,0.003314515,0.0000117634,0.0001315093,0.00009949061,0.00001044483,0.988489,0.00003971352],"study_design_scores_gemma":[0.00219579,0.0004678212,0.01031736,0.0002113464,0.0007985915,0.0002217922,0.00004278158,0.0003173467,0.000008256957,0.00002337943,0.9823251,0.003070384],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01657976,0.0004286156,0.000002247975,0.0007160159,0.001348447,0.004324051,0.9759352,0.0004687962,0.0001969299],"genre_scores_gemma":[0.0006718453,0.0004407716,0.001033029,0.01023743,0.000575426,0.000289589,0.9863204,0.0003421777,0.00008932511],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4245514,"threshold_uncertainty_score":0.9993088,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907243641","doi":"10.21233/n3mm2k","title":"Pink Lake pollen dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Vegetation (pathology); Assemblage (archaeology); Table (database)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.0534204583264859,"gpt":0.3535709180005279,"spread":0.300150459674042,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001940045,0.002131413,0.002468525,0.0006811112,0.001440511,0.0008536829,0.008454865,0.002135241,0.05273413],"category_scores_gemma":[0.006206044,0.001770014,0.0005451332,0.0003998438,0.001621077,0.0009956912,0.00558116,0.003636868,0.2526061],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000293925,"about_ca_system_score_gemma":0.0006600461,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003620101,"about_ca_topic_score_gemma":0.04006511,"domain_scores_codex":[0.9893985,0.001039894,0.001569931,0.003511284,0.001928431,0.002552001],"domain_scores_gemma":[0.9826692,0.001063464,0.001937181,0.01263918,0.0002244659,0.001466553],"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":[0.0005852526,0.001279698,0.0002715309,0.000302975,0.0002550741,0.00969914,0.000002101122,0.000001176546,0.00003753417,0.00001884382,0.9871575,0.0003892192],"study_design_scores_gemma":[0.002227027,0.000532485,0.00634397,0.0002153034,0.0006853653,0.0003240746,0.000007932181,0.00001428071,0.000008614137,0.00004950446,0.9874504,0.00214102],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0002407648,0.0003660758,0.000003316634,0.0002276951,0.00181137,0.002447116,0.9935488,0.0005884214,0.0007665175],"genre_scores_gemma":[0.00003755994,0.000596023,0.0005526904,0.001871533,0.001789253,0.0008441226,0.9936869,0.0001979736,0.0004239446],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.199872,"threshold_uncertainty_score":0.9998595,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907265724","doi":"10.21233/n3m95m","title":"LD Lake pollen dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Vegetation (pathology); Apoidea; Phenology","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.05385351384362163,"gpt":0.3582293441080741,"spread":0.3043758302644525,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001950541,0.002110968,0.002445071,0.0006796831,0.001402574,0.000840272,0.008399122,0.002112932,0.06371807],"category_scores_gemma":[0.006278864,0.001754846,0.0005410489,0.0003977767,0.00156934,0.0009902732,0.005532574,0.003640445,0.2528305],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002849928,"about_ca_system_score_gemma":0.0006209025,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003689047,"about_ca_topic_score_gemma":0.04617896,"domain_scores_codex":[0.9894624,0.001021793,0.001552267,0.003482109,0.001941428,0.002539947],"domain_scores_gemma":[0.9827181,0.001125781,0.001906813,0.0125614,0.000215554,0.001472322],"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":[0.0005954781,0.001319283,0.000321309,0.0002898507,0.0002504713,0.0102103,0.00000196848,0.000001200724,0.00003795906,0.00001289173,0.9866209,0.0003383769],"study_design_scores_gemma":[0.002195213,0.0005054639,0.005969732,0.0002043096,0.0006979346,0.0003124294,0.000008476867,0.0000110386,0.00001107531,0.00004571485,0.9879063,0.002132305],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0002436832,0.000353079,0.000002614079,0.0002047204,0.001773186,0.002393469,0.9935572,0.000567616,0.0009044316],"genre_scores_gemma":[0.00004058521,0.0005367972,0.0005126838,0.001825308,0.001710501,0.0008350576,0.9938547,0.0001972582,0.0004871323],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1891124,"threshold_uncertainty_score":0.9998975,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907269138","doi":"10.21233/nx8e-mc81","title":"Site 11, Georgian Bay, Canadian Hydrographic Service pollen surface sample dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03522344810136093,"gpt":0.2844228873843981,"spread":0.2491994392830372,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002412623,0.002878941,0.002896513,0.001727699,0.001066905,0.0007080712,0.006294648,0.002790759,0.04659963],"category_scores_gemma":[0.002124326,0.002714027,0.0006002597,0.003763185,0.0006929599,0.001060374,0.002965697,0.003692305,0.1602581],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001194965,"about_ca_system_score_gemma":0.002431133,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.6122325,"about_ca_topic_score_gemma":0.9867958,"domain_scores_codex":[0.9844005,0.00183785,0.002029306,0.004757954,0.002154013,0.004820399],"domain_scores_gemma":[0.9811292,0.002758964,0.001213048,0.01004694,0.0004690526,0.004382776],"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":[0.000300087,0.0007191692,0.01118201,0.0006111806,0.0003029095,0.003536947,0.00001188801,0.0001381689,0.0000942265,0.00001283613,0.9830622,0.0000283623],"study_design_scores_gemma":[0.002260952,0.0004043671,0.01269969,0.0002218774,0.0008426771,0.0002369608,0.00004736638,0.0003238672,0.000008884973,0.00002052908,0.9797406,0.00319224],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0256576,0.0003716979,0.000002359935,0.0008372474,0.001388183,0.004420216,0.9666968,0.000487769,0.0001381417],"genre_scores_gemma":[0.0007800682,0.0003846172,0.001055758,0.009974183,0.0005963642,0.0003066299,0.9864451,0.0003606382,0.00009668597],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.3745634,"threshold_uncertainty_score":0.9990818,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888494408","doi":"10.21233/j63c-1q91","title":"Lake Ontario/Wellers Bay water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02553415180232026,"gpt":0.2710167980926327,"spread":0.2454826462903124,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001710075,0.002340355,0.002104473,0.0003104837,0.0006996017,0.0004286639,0.00470429,0.002415395,0.4114226],"category_scores_gemma":[0.0009190672,0.00166162,0.000526517,0.0004231102,0.001750352,0.000616813,0.004562222,0.0038223,0.1742907],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009716632,"about_ca_system_score_gemma":0.0007274256,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003076515,"about_ca_topic_score_gemma":0.6054748,"domain_scores_codex":[0.9887704,0.0005972779,0.001775828,0.003888645,0.001849998,0.003117879],"domain_scores_gemma":[0.989783,0.0005253483,0.0007045962,0.007177567,0.0002735409,0.001535905],"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":[0.0008580816,0.001359726,0.0005010254,0.0003947122,0.000328011,0.005472155,0.00001613864,0.000002715572,0.0007523994,4.289921e-7,0.990292,0.00002260258],"study_design_scores_gemma":[0.002035437,0.0003901764,0.00111029,0.0001447337,0.0006893504,0.0003496916,0.0000117212,0.000006239116,0.0009924967,0.00001773899,0.9918303,0.002421827],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.008331822,0.00003902829,0.000001759627,0.00009425348,0.001228382,0.001692915,0.9873214,0.0005086742,0.0007817112],"genre_scores_gemma":[0.00004405357,0.00007223406,0.0004456986,0.001610182,0.001608096,0.0007012636,0.9936412,0.0002050668,0.00167223],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.6023983,"threshold_uncertainty_score":0.9989335,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907444215","doi":"10.21233/n3qn03","title":"Site 7 (Mott unpublished) pollen surface sample dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.0463923393343938,"gpt":0.3404525676084768,"spread":0.294060228274083,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.003309916,0.0027597,0.0031899,0.0006810166,0.001858802,0.002270153,0.009615053,0.002567462,0.04657585],"category_scores_gemma":[0.01754207,0.002350182,0.0006714321,0.0007807267,0.001782176,0.002503791,0.007822813,0.004447097,0.1487957],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006722146,"about_ca_system_score_gemma":0.0009740852,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005300616,"about_ca_topic_score_gemma":0.024865,"domain_scores_codex":[0.985445,0.001694412,0.002062726,0.004726199,0.002647468,0.003424193],"domain_scores_gemma":[0.9759213,0.003845461,0.002584387,0.01515007,0.0004325288,0.002066268],"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":[0.0007473205,0.001564111,0.00175391,0.0003812072,0.0003063875,0.004695396,0.000005814979,0.00001641711,0.0001347636,0.0000154496,0.9902342,0.0001450458],"study_design_scores_gemma":[0.002707428,0.0005275838,0.006244079,0.0002076465,0.000784477,0.0001982241,0.00001744993,0.00006628798,0.00002454997,0.00007531382,0.9863684,0.002778525],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002600062,0.0004644838,0.00001496075,0.0003391454,0.002017987,0.003009901,0.9904022,0.00079491,0.0003563649],"genre_scores_gemma":[0.0001077832,0.000538618,0.002868598,0.001815313,0.001330698,0.0003501748,0.9923687,0.0002855715,0.0003345699],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1022199,"threshold_uncertainty_score":0.9994407,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888560324","doi":"10.21233/ex6e-s069","title":"Government of Canada Pit [EgPm-VP] vertebrate fauna dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Traffic control and management","field":"Engineering","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":"Paleoecology; Fauna; Vertebrate; Government (linguistics); JSON","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.008006067275777419,"gpt":0.1967994027410867,"spread":0.1887933354653093,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002904785,0.0006365525,0.0008231353,0.0000636664,0.00009453299,0.00003979216,0.001134979,0.0002840821,0.004696754],"category_scores_gemma":[0.0001652051,0.0005479745,0.0001079665,0.0001783896,0.0001280047,0.0001105939,0.0006161422,0.000519068,0.000127751],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004063312,"about_ca_system_score_gemma":0.0001984649,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.009493503,"about_ca_topic_score_gemma":0.4888395,"domain_scores_codex":[0.9965633,0.00008442667,0.0007440927,0.0006723645,0.001154767,0.0007810901],"domain_scores_gemma":[0.9976031,0.0002078433,0.0001932736,0.00161089,0.00004171263,0.0003432432],"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":[0.00007219153,0.0001477752,0.00001464006,0.0005055621,0.0001985412,0.0005380645,0.000001565559,0.0004462549,0.00002627708,0.00001561557,0.9969299,0.001103574],"study_design_scores_gemma":[0.0007810639,0.0001751442,0.001715632,0.00007850733,0.0002058622,0.000007365568,0.00001263955,0.001014373,0.00003962345,0.000005104748,0.9953799,0.0005848455],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.001025134,0.0002344925,0.00002951837,0.00005667472,0.001868193,0.0007140862,0.9955369,0.0001247061,0.0004102721],"genre_scores_gemma":[0.003093009,0.0004729721,0.0001465199,0.0005120429,0.0003601905,0.0001646158,0.9951531,0.00003139342,0.00006618563],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.479346,"threshold_uncertainty_score":0.9996971,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963860407","doi":"10.21233/n3dt01","title":"Site 28 (Mott unpublished) pollen surface sample dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Photovoltaic System Optimization Techniques","field":"Energy","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03245494995724853,"gpt":0.3113474159590036,"spread":0.2788924660017551,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","insufficient_payload"],"category_scores_codex":[0.001516838,0.001456992,0.001954542,0.0004156844,0.0008905151,0.001072039,0.004853275,0.001883345,0.02085331],"category_scores_gemma":[0.006828501,0.001259183,0.0003726072,0.0004380599,0.000578206,0.001274743,0.003042085,0.001648201,0.005939594],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003831593,"about_ca_system_score_gemma":0.0004407933,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01695614,"about_ca_topic_score_gemma":0.01694875,"domain_scores_codex":[0.992164,0.0009689045,0.001574364,0.002434082,0.001341115,0.001517509],"domain_scores_gemma":[0.9874433,0.001801496,0.001619963,0.007959288,0.0003392602,0.0008367019],"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":[0.0001652059,0.0004542577,0.0003490218,0.0002424109,0.0001430003,0.0008802775,0.000005720224,0.0001001723,0.0001090141,0.00007723045,0.9972646,0.0002091232],"study_design_scores_gemma":[0.001041241,0.0002648879,0.0004490498,0.0001855468,0.0002313496,0.00008586063,0.00001295505,0.0002366401,0.0002804562,0.0001529722,0.9956131,0.001445951],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0002906912,0.0002685307,0.0003000248,0.0001745255,0.001198301,0.001829642,0.9938928,0.0009230712,0.001122446],"genre_scores_gemma":[0.0001490481,0.0009037299,0.009459741,0.001567683,0.0006389275,0.0004668476,0.9862279,0.000111756,0.0004743364],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.01491372,"threshold_uncertainty_score":0.999965,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6944602275","doi":"10.21233/5md3-gd58","title":"Lake Ottawa diatom bottom sample dataset","year":2022,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Diatom; Sample (material); Paleoecology; Hydrology (agriculture); Paleolimnology","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03092682306687908,"gpt":0.2992857042299124,"spread":0.2683588811630333,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.003037444,0.002388795,0.002606857,0.00103039,0.001384572,0.0004718762,0.006567789,0.001445239,0.7491173],"category_scores_gemma":[0.008105936,0.00220262,0.00068718,0.001917728,0.0009480524,0.0007805815,0.007417601,0.006532073,0.04416835],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007057111,"about_ca_system_score_gemma":0.0008191229,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001540559,"about_ca_topic_score_gemma":0.04722358,"domain_scores_codex":[0.9845001,0.002669784,0.002325533,0.00433217,0.003116657,0.003055815],"domain_scores_gemma":[0.9827129,0.005493677,0.001615491,0.00840564,0.0001693223,0.001603021],"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":[0.0005160591,0.002544674,0.001712732,0.0002995689,0.0003009298,0.007294011,0.000004688661,0.00001565408,0.00002239354,0.00005526437,0.9869891,0.0002449387],"study_design_scores_gemma":[0.00217808,0.0008092409,0.003868663,0.00004595864,0.0007525996,0.000259011,0.00003506347,0.00002584674,0.00000619113,0.00008200486,0.9893209,0.002616413],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0006052791,0.0004475579,0.0000160342,0.0001491488,0.002056456,0.003028762,0.9925665,0.0009973111,0.0001329356],"genre_scores_gemma":[0.00001749536,0.0004530669,0.001686822,0.003914275,0.0009545555,0.002998313,0.9895857,0.0002993596,0.00009043219],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.704949,"threshold_uncertainty_score":0.9999155,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888564111","doi":"10.21233/7z97-5w19","title":"Ottawa River ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Sample (material); Paleoecology; Hydrology (agriculture); Surface (topology); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03754192152436427,"gpt":0.3141483203499923,"spread":0.276606398825628,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002436608,0.002349275,0.002302114,0.000469428,0.00080404,0.0004030138,0.005326725,0.00215596,0.07144327],"category_scores_gemma":[0.006829774,0.002024404,0.000500358,0.001182279,0.002838545,0.0009341078,0.003703931,0.003123488,0.1526158],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005542106,"about_ca_system_score_gemma":0.0005606349,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003146739,"about_ca_topic_score_gemma":0.01836045,"domain_scores_codex":[0.9871016,0.001613347,0.002045417,0.004077402,0.002229341,0.002932876],"domain_scores_gemma":[0.9849657,0.003815508,0.001456375,0.007760592,0.0004069082,0.001594858],"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":[0.0005756941,0.001704229,0.0006737135,0.0002795234,0.0002698131,0.002978766,0.000006316521,0.00001453653,0.0002138214,0.00001630642,0.9932057,0.00006157742],"study_design_scores_gemma":[0.001902292,0.0009011429,0.003983527,0.0001407851,0.0007095366,0.0001890903,0.00001199533,0.0000640176,0.0001014198,0.0001043414,0.98945,0.002441828],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.004063886,0.0001864203,0.00003528061,0.00008270928,0.001613896,0.002516475,0.9906364,0.0007422769,0.0001226474],"genre_scores_gemma":[0.00007014217,0.0002661444,0.009234086,0.001858985,0.001545507,0.0002749574,0.9864289,0.0002366411,0.00008463776],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.08117247,"threshold_uncertainty_score":0.9998751,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888513827","doi":"10.21233/dwty-p603","title":"Lake Ottawa diatom surface sample dataset","year":2021,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Diatom; Sample (material); Hydrology (agriculture); Paleolimnology; JSON","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03292175507514862,"gpt":0.3033355232338611,"spread":0.2704137681587125,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002183574,0.002319184,0.002810746,0.00040748,0.0006884543,0.0006449349,0.004334734,0.002034178,0.1041356],"category_scores_gemma":[0.01020523,0.002084061,0.0006383726,0.001750051,0.0009258665,0.0007410739,0.004640636,0.003856203,0.07315301],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004031999,"about_ca_system_score_gemma":0.0009438633,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00106987,"about_ca_topic_score_gemma":0.08888464,"domain_scores_codex":[0.9861661,0.002258023,0.002152738,0.00427474,0.002312258,0.002836195],"domain_scores_gemma":[0.9827687,0.005477879,0.001275294,0.008450619,0.0003795741,0.001647958],"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":[0.0002429088,0.002137811,0.00183209,0.0004268619,0.0003130406,0.01014328,0.000002853629,0.00004106206,0.0001331034,0.00002006114,0.9845936,0.0001133001],"study_design_scores_gemma":[0.001823479,0.0003255742,0.003479586,0.0001741827,0.0007048601,0.0002492047,0.00002957061,0.00004289286,0.00006359607,0.0000302886,0.990623,0.002453773],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002368994,0.000871982,0.00002742834,0.0001376651,0.001596852,0.002016923,0.9921942,0.0006888494,0.0000970958],"genre_scores_gemma":[0.00002718008,0.0007681018,0.004975697,0.002530983,0.0009840514,0.0004212005,0.989952,0.0002478678,0.00009297852],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.08781477,"threshold_uncertainty_score":0.9992614,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6926447289","doi":"10.21233/34j3-b541","title":"Lake Winnipeg ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); Hydrology (agriculture); JSON; Glacial period","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03069011364637703,"gpt":0.2983547078859254,"spread":0.2676645942395484,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.000747744,0.0009341011,0.001406464,0.0001674336,0.0003031138,0.000119979,0.0009859519,0.001146818,0.04165529],"category_scores_gemma":[0.003549683,0.0006462897,0.0002849303,0.0004447806,0.0006559871,0.0001935948,0.0009011559,0.001628756,0.003606881],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009971872,"about_ca_system_score_gemma":0.0002926952,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00033961,"about_ca_topic_score_gemma":0.004859606,"domain_scores_codex":[0.9951927,0.0002573365,0.000909363,0.001843731,0.0007162441,0.001080633],"domain_scores_gemma":[0.9949861,0.0008340834,0.0004203537,0.002846457,0.0002391283,0.0006738871],"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":[0.0008206558,0.0008214368,0.0001438667,0.0005341406,0.0001592896,0.001336127,0.000001573421,2.640211e-7,0.002738659,0.000001923259,0.9930681,0.000373975],"study_design_scores_gemma":[0.0009861734,0.0009611789,0.001464438,0.0002371597,0.000680287,0.0003134669,0.000005204133,0.000005338754,0.001696338,0.00001109509,0.9928474,0.0007919307],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.007241237,0.0003257367,0.000003698744,0.001161256,0.000921534,0.001547871,0.9885899,0.0001621048,0.00004662335],"genre_scores_gemma":[0.0002725964,0.0009231918,0.003498182,0.004704074,0.001950991,0.000009554215,0.9883519,0.00004405565,0.0002454976],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.03804841,"threshold_uncertainty_score":0.9995989,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907193336","doi":"10.21233/bp0d-hk47","title":"Lake Ontario/Athol Bay ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Bay; Sample (material); Biogeosciences; Hydrology (agriculture)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04239073029346103,"gpt":0.3058067063964637,"spread":0.2634159761030027,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.00275963,0.002691801,0.00269676,0.0004875897,0.0009263481,0.0006239884,0.005404953,0.002399811,0.1704559],"category_scores_gemma":[0.004777232,0.002344542,0.0005765591,0.001015659,0.001818246,0.001051394,0.004160221,0.004535501,0.1047642],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001029944,"about_ca_system_score_gemma":0.001492325,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.02059604,"about_ca_topic_score_gemma":0.9469351,"domain_scores_codex":[0.9860456,0.001478007,0.00226665,0.004533119,0.002414575,0.003262073],"domain_scores_gemma":[0.9846513,0.003235476,0.001501878,0.008429338,0.0004149956,0.001766985],"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":[0.0009568448,0.001995006,0.003668759,0.0002486408,0.0002661749,0.004436367,0.00001700083,0.00001704041,0.0001258607,0.000008071848,0.9882215,0.00003878948],"study_design_scores_gemma":[0.002115064,0.001089765,0.01340485,0.0001843092,0.000827422,0.000317786,0.0000144497,0.0000212678,0.00005465063,0.00006186475,0.9791749,0.002733677],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01052361,0.00008663104,0.00002703972,0.00006861312,0.001846246,0.003092222,0.9834568,0.0007164025,0.0001824691],"genre_scores_gemma":[0.00004918908,0.0001344604,0.007951424,0.001747578,0.001209581,0.0004235095,0.9878721,0.000250883,0.0003613349],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.926339,"threshold_uncertainty_score":0.9999763,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888452535","doi":"10.21233/wh7b-5598","title":"Alberta Lake diatom surface sample dataset","year":2023,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Diatom; Sample (material); Paleoecology; Hydrology (agriculture); Surface (topology)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04201300915475259,"gpt":0.3139840234316292,"spread":0.2719710142768766,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002019168,0.00229776,0.00252176,0.0005767797,0.0006920784,0.000485171,0.005134256,0.001969028,0.02852769],"category_scores_gemma":[0.01382414,0.002013661,0.0005589898,0.00215151,0.0009250292,0.0006939298,0.005235709,0.003336207,0.5254892],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003538727,"about_ca_system_score_gemma":0.000543775,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.01638153,"about_ca_topic_score_gemma":0.4807392,"domain_scores_codex":[0.9872794,0.001472073,0.001965327,0.004048331,0.002141437,0.003093428],"domain_scores_gemma":[0.976958,0.01247255,0.001144468,0.007637192,0.00020738,0.001580453],"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":[0.0004313779,0.001198036,0.001108218,0.0003762084,0.0003109134,0.003712834,0.000004715236,0.00005812106,0.00003756711,0.00002090433,0.9926798,0.00006135881],"study_design_scores_gemma":[0.001692409,0.0003999909,0.004462257,0.0001507872,0.0006369516,0.00009069397,0.00002127733,0.00009676293,0.00001243265,0.00006882785,0.9899853,0.002382282],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.001372589,0.0001347661,0.000007567516,0.0002019274,0.001929272,0.00266979,0.9925065,0.001098036,0.00007956869],"genre_scores_gemma":[0.00001170212,0.000621526,0.001090524,0.001552737,0.0008845984,0.0006236469,0.9943863,0.0003847122,0.000444225],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4969614,"threshold_uncertainty_score":0.9993266,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6944749696","doi":"10.21233/ve1q-9c92","title":"Adams-McGill Reservoir water chemistry dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Hydrology (agriculture); Water chemistry; JSON; Raw data; Biogeosciences","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04957984495907053,"gpt":0.3079917571458093,"spread":0.2584119121867388,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.0019045,0.002314577,0.002510639,0.0003460653,0.0006153161,0.000273692,0.005877872,0.002613963,0.1127927],"category_scores_gemma":[0.001972822,0.001624822,0.0005498785,0.0005833696,0.0008084661,0.0007834822,0.006850495,0.004375119,0.3020996],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000685739,"about_ca_system_score_gemma":0.0002595151,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002759038,"about_ca_topic_score_gemma":0.0008228155,"domain_scores_codex":[0.987653,0.0009003012,0.001889115,0.004148449,0.002219165,0.003189985],"domain_scores_gemma":[0.9868478,0.0009925403,0.0007722919,0.009798415,0.0002956,0.001293422],"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":[0.0009907131,0.001378168,0.00005196823,0.001216206,0.0002646162,0.005149559,0.000001868558,0.00002409455,0.004783485,0.000001921377,0.9860983,0.00003910875],"study_design_scores_gemma":[0.002177315,0.0003395686,0.0002249948,0.0001997655,0.0005662204,0.0003109021,0.00001624922,0.00003721625,0.002733585,0.00002418845,0.9909308,0.002439225],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.00539691,0.0001564276,6.977737e-7,0.0001883661,0.0009950562,0.002377048,0.9897807,0.0004857237,0.0006191061],"genre_scores_gemma":[0.0001281836,0.0002475717,0.0002607568,0.001867737,0.000885165,0.0007845395,0.9946339,0.0002596824,0.0009325158],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.189307,"threshold_uncertainty_score":0.9995008,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907177935","doi":"10.21233/b4q3-b180","title":"McGill water chemistry dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Cyberinfrastructure; Water chemistry; Raw data; JSON","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.0416792509301375,"gpt":0.3002539176726046,"spread":0.2585746667424671,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001241523,0.001926961,0.001997317,0.0002617525,0.0004532334,0.0001956166,0.004162596,0.002094728,0.1312434],"category_scores_gemma":[0.00109112,0.001326624,0.000452023,0.0003968886,0.0006239389,0.0005457596,0.004542046,0.003257376,0.3877276],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000491345,"about_ca_system_score_gemma":0.0001601702,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001221866,"about_ca_topic_score_gemma":0.0002641207,"domain_scores_codex":[0.9904306,0.0005568049,0.00147618,0.003369173,0.001606292,0.002560945],"domain_scores_gemma":[0.9902046,0.0006733736,0.0005935043,0.00734207,0.0001881084,0.0009982972],"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":[0.0004996118,0.001157125,0.00003007368,0.0008906517,0.0002017248,0.003799498,0.000001127893,0.000008496016,0.005387979,0.000001561697,0.9879597,0.00006249147],"study_design_scores_gemma":[0.001747609,0.0002321245,0.0001388308,0.0001192655,0.0005213968,0.0003189801,0.000008719408,0.00002044992,0.002799791,0.00001495719,0.9920219,0.002056047],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.001640577,0.00009886614,8.158983e-7,0.00009970088,0.0009638651,0.001873701,0.9943851,0.0004033001,0.0005341378],"genre_scores_gemma":[0.00007847917,0.0001730638,0.0001830808,0.002042164,0.0007217546,0.0005735553,0.9954567,0.0001962837,0.0005749341],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.2564843,"threshold_uncertainty_score":0.9993474,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907283470","doi":"10.21233/2we8-nt76","title":"Lake Winnipeg water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Water chemistry; Hydrology (agriculture); Paleolimnology; Biogeosciences","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02502836061534223,"gpt":0.2880269153572144,"spread":0.2629985547418721,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001673108,0.002117172,0.001945269,0.0002892242,0.0006094418,0.0004035259,0.004696735,0.002272051,0.2323453],"category_scores_gemma":[0.001604081,0.001463984,0.0004376617,0.0005302114,0.001843495,0.000598211,0.004726282,0.002913549,0.2125726],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002932111,"about_ca_system_score_gemma":0.0002812457,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009377758,"about_ca_topic_score_gemma":0.004281824,"domain_scores_codex":[0.9894684,0.0006455706,0.001662577,0.003563649,0.001760438,0.002899351],"domain_scores_gemma":[0.9898559,0.0005307375,0.0006495433,0.00734531,0.0003035654,0.00131487],"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":[0.0006194937,0.001338103,0.0000758426,0.0005752516,0.000226054,0.005192155,0.000003805108,9.242228e-7,0.002668363,7.297352e-7,0.9892516,0.00004775295],"study_design_scores_gemma":[0.001677701,0.0003479688,0.0002999995,0.0001413818,0.000581879,0.0004200244,0.000007652294,0.00001085554,0.00226511,0.00002582155,0.9920551,0.002166488],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002646328,0.0001036412,0.000002331983,0.0001194,0.0009963526,0.001516077,0.9935758,0.0006525193,0.000387519],"genre_scores_gemma":[0.00003688407,0.0001251286,0.000370714,0.001839716,0.002820464,0.0007188018,0.9934548,0.0002070334,0.0004265003],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.01977264,"threshold_uncertainty_score":0.9993868,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6926007326","doi":"10.21233/n3sb5h","title":"Lake Site South Burin Peninsula pollen dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Education, sociology, and vocational training","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Peninsula; Pollen; Shandong peninsula; Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.2329127525977666,"gpt":0.4840578810416286,"spread":0.2511451284438619,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts","insufficient_payload"],"consensus_categories":["sts","insufficient_payload"],"category_scores_codex":[0.004219602,0.000545297,0.0007406,0.0001978072,0.002780059,0.0003586316,0.002362044,0.001060588,0.009543329],"category_scores_gemma":[0.009839995,0.0005005521,0.0002088434,0.0001908505,0.002786041,0.0004197293,0.0005602214,0.001218727,0.004068219],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001154275,"about_ca_system_score_gemma":0.001024658,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001099595,"about_ca_topic_score_gemma":0.0219857,"domain_scores_codex":[0.994673,0.001644571,0.0006233411,0.001202985,0.0008158226,0.001040329],"domain_scores_gemma":[0.9941497,0.002406268,0.0007650554,0.001908586,0.0001981809,0.0005722046],"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":[0.0000233648,0.0002541039,0.002687251,0.0000544751,0.00005575043,0.0001052113,0.0007462413,4.075625e-7,0.000001176232,0.0005379695,0.9953808,0.0001533043],"study_design_scores_gemma":[0.0004734237,0.00009303101,0.009696925,0.00006014539,0.0001383255,0.000006772162,0.00268835,0.000001545117,3.0108e-7,0.0001476866,0.9860827,0.0006107359],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002286604,0.0001759908,0.000005814301,0.001901815,0.002614605,0.0008550088,0.9900619,0.0001255477,0.001972718],"genre_scores_gemma":[0.001507425,0.001150792,0.0006733631,0.001325501,0.006099411,0.0002842831,0.9880537,0.00002290744,0.0008826089],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.02088611,"threshold_uncertainty_score":0.9999278,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6926272728","doi":"10.21233/218z-yn28","title":"Site 31, Georgian Bay, Canadian Hydrographic Service pollen surface sample dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Prenatal Screening and Diagnostics","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":"Hydrography; Pollen; Georgian; Sample (material); Service (business)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03314748601294594,"gpt":0.2896120739771524,"spread":0.2564645879642065,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0006699139,0.00104798,0.0014852,0.0005016604,0.0003632826,0.0001562279,0.001283533,0.001390324,0.006574077],"category_scores_gemma":[0.002714445,0.0008950843,0.0002851294,0.001014641,0.0002437882,0.0002340135,0.0009676641,0.002134486,0.008717948],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003164225,"about_ca_system_score_gemma":0.001352576,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.2889909,"about_ca_topic_score_gemma":0.6719825,"domain_scores_codex":[0.9944456,0.000279805,0.0008594875,0.001750156,0.000893354,0.001771597],"domain_scores_gemma":[0.9911819,0.00211459,0.0003448142,0.003615768,0.0002078898,0.002535046],"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":[0.0003490804,0.0003996772,0.01418751,0.0005787141,0.0001818587,0.003751142,0.000005465837,0.00002836902,0.00001610705,0.000005792779,0.9803694,0.0001268803],"study_design_scores_gemma":[0.001996011,0.0007414583,0.02877162,0.0004079891,0.0007657689,0.0002787567,0.00002199694,0.0002299309,0.000008441916,0.00001019687,0.9657739,0.0009938654],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01810935,0.0005339261,0.000007044582,0.00161504,0.0006241372,0.001979082,0.9766499,0.0001598249,0.0003216975],"genre_scores_gemma":[0.001679096,0.0007693337,0.001440452,0.01425658,0.0004464446,0.00007997316,0.9811591,0.00006190355,0.0001071357],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.3829916,"threshold_uncertainty_score":0.9999061,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6926068258","doi":"10.21233/n3z47r","title":"Site 13 (Mott unpublished) pollen surface sample dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04517204369349025,"gpt":0.335582708193329,"spread":0.2904106644998388,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"category_scores_codex":[0.003427888,0.002900964,0.003341049,0.0007187828,0.00191652,0.002395671,0.009971499,0.002682645,0.05998688],"category_scores_gemma":[0.01803946,0.002474452,0.0007053041,0.000834204,0.001861243,0.002667946,0.008193883,0.004640752,0.1513339],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006861844,"about_ca_system_score_gemma":0.001004151,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.007620921,"about_ca_topic_score_gemma":0.04075733,"domain_scores_codex":[0.9847344,0.001797372,0.00217381,0.004942865,0.002764399,0.003587174],"domain_scores_gemma":[0.9751135,0.00402794,0.002694388,0.01552485,0.0004599565,0.002179364],"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":[0.000787045,0.001638972,0.002368012,0.0004005956,0.0003279376,0.004846319,0.000006742297,0.00002324084,0.0001337484,0.00001194155,0.9893059,0.000149545],"study_design_scores_gemma":[0.002910385,0.0005536006,0.00873885,0.0002186147,0.0008229564,0.0002072184,0.00002015389,0.00008384667,0.00002603588,0.00007704438,0.9834259,0.002915426],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.004024554,0.0004862513,0.00001544276,0.0003537895,0.002089719,0.003200679,0.988605,0.0008343369,0.0003901954],"genre_scores_gemma":[0.0001567789,0.000578264,0.002428309,0.001900626,0.001379309,0.0003849897,0.9925022,0.0003058638,0.0003636295],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.09134704,"threshold_uncertainty_score":0.9998277,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6926108429","doi":"10.21233/gqxk-pc97","title":"Lake Ontario water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Environmental and Social Impact Assessments","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":"Paleoecology; Water chemistry; Paleolimnology; Hydrology (agriculture); Biogeosciences; Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.01720340078734997,"gpt":0.268880043723337,"spread":0.251676642935987,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0003880705,0.0007324253,0.0006202699,0.00002506598,0.0004095449,0.0001211612,0.001415423,0.000743915,0.4876045],"category_scores_gemma":[0.00006561683,0.0005077306,0.0001736417,0.00009950036,0.0009759272,0.0003728907,0.003094931,0.001060463,0.02896168],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007995221,"about_ca_system_score_gemma":0.00004565266,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006233002,"about_ca_topic_score_gemma":0.3056589,"domain_scores_codex":[0.9960794,0.0001455339,0.0005593308,0.001239423,0.0008707545,0.001105546],"domain_scores_gemma":[0.9976014,0.00008030263,0.0001968256,0.001527958,0.000005744315,0.000587694],"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":[0.00006248237,0.0006506912,0.005962877,0.00003446415,0.00004303593,0.0005134689,0.0000194986,0.000001721167,0.0005677221,5.792485e-8,0.9920714,0.00007258205],"study_design_scores_gemma":[0.0004802245,0.0001974179,0.01839784,0.00001980781,0.0001212338,0.00003333728,0.000008277759,0.000001352001,0.0003997753,0.00002850272,0.9795726,0.0007396623],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.04697065,0.000004667616,0.000001504818,0.00006519107,0.0004087685,0.000520716,0.9501244,0.00005337783,0.001850703],"genre_scores_gemma":[0.0002775889,0.00005700205,0.0002162385,0.001583791,0.0003747955,0.0001419866,0.9949549,0.00002649409,0.002367227],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4586428,"threshold_uncertainty_score":0.9997374,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888486444","doi":"10.21233/wqms-b579","title":"Lake Ontario/Collins Bay water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Bay; Paleoecology; Water chemistry; Hydrology (agriculture); Water quality; Limnology","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02716741108711867,"gpt":0.2773472757355713,"spread":0.2501798646484527,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001774149,0.002389272,0.00225592,0.0003178083,0.0007529834,0.0004827278,0.004842192,0.002612309,0.3565253],"category_scores_gemma":[0.001059366,0.001691313,0.0005404982,0.0004923863,0.001598841,0.0006440721,0.0046555,0.003967498,0.1588527],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0010383,"about_ca_system_score_gemma":0.0008857877,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003013728,"about_ca_topic_score_gemma":0.7182617,"domain_scores_codex":[0.9884077,0.0006123685,0.001932133,0.003965771,0.001921485,0.003160545],"domain_scores_gemma":[0.9892988,0.0005587852,0.0007518832,0.007490535,0.0003395659,0.001560495],"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":[0.0008461223,0.001890155,0.0006363944,0.0004086929,0.0003118592,0.006026516,0.00001450466,0.000002023763,0.0008717977,6.780501e-7,0.9889727,0.00001854393],"study_design_scores_gemma":[0.002123138,0.000467709,0.001671661,0.0001547465,0.0007685074,0.0003805366,0.000008262827,0.000007377061,0.001104802,0.00002165438,0.9908636,0.002428027],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01031538,0.00003926535,0.000002762037,0.00008242682,0.00121883,0.002003109,0.9852142,0.0005428155,0.0005812175],"genre_scores_gemma":[0.00004774299,0.00007019773,0.0005870819,0.001514298,0.002028154,0.0009842969,0.9928476,0.0002125917,0.001708],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.7152479,"threshold_uncertainty_score":0.9988845,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907320565","doi":"10.21233/y0hg-f171","title":"Lake Manitoba ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); Surface (topology); Hydrology (agriculture); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.0464830546314058,"gpt":0.3116888561157952,"spread":0.2652058014843894,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002204526,0.002356358,0.002262261,0.0004535107,0.0008223701,0.0005482822,0.005342504,0.002093144,0.07458737],"category_scores_gemma":[0.005668518,0.002055153,0.0004696395,0.001128833,0.001753216,0.0008420132,0.003928144,0.003186389,0.1459519],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004404536,"about_ca_system_score_gemma":0.0005187733,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002051129,"about_ca_topic_score_gemma":0.2687262,"domain_scores_codex":[0.9875029,0.001441827,0.00197449,0.003949838,0.002180067,0.002950837],"domain_scores_gemma":[0.9856246,0.003409944,0.001344112,0.007726931,0.0003840852,0.001510328],"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":[0.0007473101,0.001699829,0.0007792742,0.0003315191,0.0002474663,0.003574606,0.000002843248,0.00001048188,0.0001560862,0.000009642805,0.9923638,0.00007719061],"study_design_scores_gemma":[0.00170278,0.0009074087,0.003705384,0.0001620409,0.0006939752,0.0002578828,0.00002904454,0.00004293726,0.00007686104,0.00007612228,0.9899365,0.002409104],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.005138516,0.000160962,0.0000227361,0.00007571299,0.001866566,0.002692595,0.9891171,0.0008295192,0.00009631092],"genre_scores_gemma":[0.000046065,0.0003541554,0.005134834,0.0018107,0.001982098,0.0003385695,0.9900364,0.0002509893,0.00004614474],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.2666751,"threshold_uncertainty_score":0.9992024,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6925886978","doi":"10.21233/7734-zx54","title":"Adams-McGill Reservoir diatom surface sample dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Diatom; Sample (material); Hydrology (agriculture); JSON; Paleolimnology","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.06992012292253903,"gpt":0.3300284588049191,"spread":0.2601083358823801,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002731771,0.002661316,0.003246842,0.0005913367,0.0008399381,0.0003616032,0.006785937,0.0024566,0.04360753],"category_scores_gemma":[0.008763567,0.002299352,0.0006473985,0.001606172,0.0008853258,0.001139007,0.006738238,0.004555831,0.2100442],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008934187,"about_ca_system_score_gemma":0.0005825898,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002727073,"about_ca_topic_score_gemma":0.01085055,"domain_scores_codex":[0.9844374,0.002261633,0.002274933,0.004806602,0.002843899,0.003375553],"domain_scores_gemma":[0.9779891,0.00715428,0.001611178,0.01122061,0.0003775585,0.001647303],"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":[0.001115893,0.001765631,0.0008511607,0.0006468429,0.0003107934,0.002661775,0.000002157265,0.0002196668,0.0001633276,0.00003076701,0.9921744,0.00005758521],"study_design_scores_gemma":[0.00250196,0.0008155757,0.002374887,0.0002092878,0.0006562066,0.0001190094,0.00002502241,0.0001617666,0.00006584574,0.00006741682,0.9901634,0.002839551],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.007723642,0.0004133719,0.000006464232,0.0001484114,0.001754298,0.00403765,0.9851124,0.0006696468,0.0001340753],"genre_scores_gemma":[0.0001321283,0.0007523595,0.002868509,0.002379692,0.0006218855,0.0004184949,0.9923286,0.0003309329,0.0001674277],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1664367,"threshold_uncertainty_score":0.999586,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888460439","doi":"10.21233/n3r94g","title":"Site 10 (Mott unpublished) pollen surface sample dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04802777879815841,"gpt":0.3404180817763692,"spread":0.2923903029782108,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"category_scores_codex":[0.003683774,0.002867584,0.003289405,0.0007095092,0.001872768,0.00235174,0.009880397,0.002691126,0.1185174],"category_scores_gemma":[0.02044566,0.002448847,0.0006991983,0.0008233125,0.001840383,0.002573653,0.008110473,0.004147846,0.2338226],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006876399,"about_ca_system_score_gemma":0.001012015,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004146745,"about_ca_topic_score_gemma":0.02203959,"domain_scores_codex":[0.9848114,0.00190829,0.002139383,0.004896061,0.002728234,0.003516665],"domain_scores_gemma":[0.9751291,0.004034352,0.00267326,0.01550741,0.0005135468,0.002142295],"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":[0.001045849,0.001654961,0.0009840863,0.000383777,0.0003192277,0.004487616,0.000005080477,0.00001602191,0.0001216252,0.00001220774,0.9908221,0.0001473851],"study_design_scores_gemma":[0.002902023,0.0006334732,0.004518834,0.0002016426,0.0008178702,0.0001808952,0.00001631432,0.00007174932,0.00002466973,0.00006321302,0.9876844,0.002884876],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002086275,0.0004494365,0.00001119117,0.0003189214,0.001693603,0.002991274,0.9910969,0.0008170289,0.0005354366],"genre_scores_gemma":[0.00006869272,0.000421203,0.002709482,0.00147027,0.001339371,0.0003685352,0.9923646,0.0002996864,0.000958112],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1153052,"threshold_uncertainty_score":0.9999117,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6944769255","doi":"10.21233/rycc-n949","title":"Lake Manitoba water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Wheat and Barley Genetics and Pathology","field":"Agricultural and Biological 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":"Water chemistry; Paleoecology; Hydrology (agriculture); Water quality; Paleolimnology; Limnology","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02211137752414439,"gpt":0.2359748199573171,"spread":0.2138634424331727,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0004269685,0.0006385545,0.0006474917,0.00001799285,0.000406075,0.0001560605,0.001397821,0.0009990871,0.05979388],"category_scores_gemma":[0.00009801766,0.0002064937,0.0002010854,0.0001556368,0.0003799584,0.0000811703,0.001083,0.0007043076,0.003832838],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002523245,"about_ca_system_score_gemma":0.00001848614,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001430491,"about_ca_topic_score_gemma":0.02468047,"domain_scores_codex":[0.996637,0.0001934246,0.0005509615,0.001203173,0.0004055584,0.001009854],"domain_scores_gemma":[0.9986049,0.0002256603,0.0001524511,0.0005402694,0.0000936366,0.0003830965],"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":[0.00008580833,0.0003089442,0.00009586568,0.0000543768,0.00002556621,0.0005806173,0.000001160441,8.048671e-8,0.02476181,4.209762e-7,0.9730371,0.001048266],"study_design_scores_gemma":[0.0002090937,0.0005167662,0.001650007,0.0000259449,0.00009351049,0.0001182726,0.00001177412,0.00000175411,0.002618281,0.0000456597,0.9940518,0.0006570954],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.09635794,0.00009834803,1.72769e-7,0.0002886385,0.0005570495,0.0003897639,0.9020884,0.00007417655,0.0001455],"genre_scores_gemma":[0.0005136346,0.0004199127,0.00004981262,0.001435249,0.00281631,0.0001181543,0.9945035,0.000002708926,0.0001407284],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.09584431,"threshold_uncertainty_score":0.9969428,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888578027","doi":"10.21233/nsvg-1q27","title":"Lake Winnipeg ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); Hydrology (agriculture); JSON; Glacial period","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04079354183707228,"gpt":0.3170129130044492,"spread":0.2762193711673769,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002462238,0.002446713,0.002445933,0.0004740998,0.0008278716,0.0005257917,0.005434324,0.002226673,0.1011416],"category_scores_gemma":[0.007109813,0.00212111,0.0004998562,0.001344759,0.002053076,0.0009103695,0.003913658,0.003348841,0.1288547],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003625348,"about_ca_system_score_gemma":0.0006432025,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001011206,"about_ca_topic_score_gemma":0.06806458,"domain_scores_codex":[0.9868788,0.001509661,0.002093046,0.004158668,0.002270813,0.003088985],"domain_scores_gemma":[0.9846615,0.003703109,0.001415207,0.008176653,0.0004068752,0.001636616],"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":[0.0008200576,0.001756059,0.0005990748,0.0003232281,0.0002543217,0.003049953,0.000003981957,0.00001494828,0.0002196258,0.0000102735,0.9928761,0.00007240666],"study_design_scores_gemma":[0.001871299,0.000980194,0.002826018,0.000171137,0.0006842348,0.0002316634,0.00001400546,0.0000533829,0.00008658423,0.000074225,0.9905,0.002507292],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002968557,0.0002152124,0.00002824399,0.0001095633,0.001707147,0.002633958,0.9913626,0.000862162,0.0001125178],"genre_scores_gemma":[0.00004412547,0.0003222406,0.005463487,0.00210918,0.001813972,0.0003126844,0.9895797,0.0002567673,0.00009786453],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.06705337,"threshold_uncertainty_score":0.9999468,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6907318632","doi":"10.21233/zvxf-w514","title":"Bear Flat, Peace River, NE British Columbia vertebrate fauna dataset","year":2015,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Fauna; Vertebrate; Biodiversity; Biogeography","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03290956435333087,"gpt":0.2828544687059061,"spread":0.2499449043525753,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002822925,0.001888409,0.00312216,0.0003805157,0.001024044,0.002555267,0.005781448,0.002432851,0.05004076],"category_scores_gemma":[0.004892888,0.002640439,0.0005480697,0.001588727,0.00222807,0.001496908,0.005340446,0.004450422,0.1211648],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00104929,"about_ca_system_score_gemma":0.001220904,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.03659616,"about_ca_topic_score_gemma":0.5245814,"domain_scores_codex":[0.9845534,0.002105405,0.002225394,0.004684722,0.003090933,0.003340142],"domain_scores_gemma":[0.9870492,0.0008975986,0.001492266,0.006925128,0.0006863848,0.002949375],"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":[0.0003780262,0.001755985,0.0008911358,0.0003523158,0.0003321743,0.02429434,0.000007351112,0.000004736872,0.0000462281,0.000001173573,0.9711854,0.0007510995],"study_design_scores_gemma":[0.004544117,0.0010653,0.006122411,0.0003599849,0.0009783496,0.00130484,0.00003438348,0.0001154091,0.000004308506,0.00007059825,0.9823143,0.003086004],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.006756969,0.001747397,0.000004434962,0.00004914627,0.002448689,0.003544497,0.9842234,0.0009618632,0.0002636439],"genre_scores_gemma":[0.00006191283,0.001480579,0.001077952,0.002035853,0.001589471,0.0008313399,0.9915754,0.0003365112,0.001010953],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4879852,"threshold_uncertainty_score":0.9995977,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888495679","doi":"10.21233/47cj-np11","title":"Site 3, Georgian Bay, Canadian Hydrographic Service pollen surface sample dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03618750031423847,"gpt":0.285837592773914,"spread":0.2496500924596755,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002351274,0.002760321,0.002781569,0.001667714,0.001021396,0.0006684096,0.006068914,0.002726228,0.04772128],"category_scores_gemma":[0.002082811,0.002597809,0.0005718899,0.003645361,0.0006559938,0.001008987,0.0028273,0.004222039,0.2037015],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001172364,"about_ca_system_score_gemma":0.00240191,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.5487334,"about_ca_topic_score_gemma":0.974295,"domain_scores_codex":[0.9849988,0.001747927,0.001945219,0.004584038,0.002082061,0.00464194],"domain_scores_gemma":[0.9817808,0.002634407,0.001154625,0.009754568,0.0004524028,0.004223181],"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":[0.0003707641,0.0006945214,0.005953635,0.0005792073,0.000350991,0.003315677,0.00001187047,0.0001342895,0.00009946971,0.00001036964,0.98844,0.00003918919],"study_design_scores_gemma":[0.00217189,0.0004717192,0.01028857,0.0002112263,0.0008065376,0.0002218496,0.00004284626,0.0003215078,0.000008150262,0.00002334669,0.9823621,0.003070258],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01640749,0.0004295506,0.000002279709,0.0007305644,0.001348675,0.004324001,0.9760959,0.0004688576,0.000192694],"genre_scores_gemma":[0.0006705632,0.0004481649,0.001041922,0.0103153,0.0005750632,0.0002836096,0.9862343,0.0003421443,0.00008891281],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4255615,"threshold_uncertainty_score":0.9993087,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6926310424","doi":"10.21233/aj7a-zd39","title":"Site 22, Georgian Bay, Canadian Hydrographic Service pollen surface sample dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"History of Computing Technologies","field":"Computer 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":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02623565170144206,"gpt":0.2573196254817599,"spread":0.2310839737803178,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001239428,0.001279329,0.001380687,0.001094391,0.000796047,0.0005517708,0.01188461,0.001523597,0.001849878],"category_scores_gemma":[0.001110059,0.001226626,0.0002780616,0.002131064,0.0004275861,0.0007962277,0.004980453,0.002382102,0.009708287],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005982904,"about_ca_system_score_gemma":0.001352017,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.08761707,"about_ca_topic_score_gemma":0.4803408,"domain_scores_codex":[0.9920758,0.0005823598,0.001027268,0.002965021,0.001107446,0.002242118],"domain_scores_gemma":[0.987731,0.001545119,0.0006318863,0.008707509,0.0002292218,0.001155271],"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":[0.00001807101,0.0001866373,0.0005230536,0.0001811497,0.00006902843,0.001111166,0.00001280164,0.0001120689,0.00001268816,0.0001465719,0.9972017,0.0004250511],"study_design_scores_gemma":[0.0005810943,0.0003275541,0.001563047,0.000110481,0.00007788575,0.0001149632,0.000009693084,0.002700374,0.000007344052,0.0001773902,0.9928729,0.001457305],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002727732,0.0003827723,0.0007918848,0.0023985,0.001771757,0.001383639,0.9894311,0.0009959491,0.0001166417],"genre_scores_gemma":[0.0005042573,0.000199722,0.02442845,0.006140914,0.0001610762,0.00006440393,0.9684049,0.00004595686,0.00005034218],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.3927237,"threshold_uncertainty_score":0.9999959,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6888558042","doi":"10.21233/x603-js09","title":"South Saskatchewan River ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); JSON; Hydrology (agriculture); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.0360434516052837,"gpt":0.2991243810822944,"spread":0.2630809294770107,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002231441,0.002567492,0.002499979,0.0004771919,0.0008591067,0.0004510721,0.005255714,0.002295583,0.05048331],"category_scores_gemma":[0.004308914,0.00221877,0.0005667233,0.00119798,0.003017973,0.0008216717,0.004130383,0.003389698,0.1219881],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005596917,"about_ca_system_score_gemma":0.0009901876,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003325055,"about_ca_topic_score_gemma":0.02630267,"domain_scores_codex":[0.9865428,0.001588082,0.00201389,0.004389812,0.002314835,0.003150539],"domain_scores_gemma":[0.9854714,0.002712458,0.001601258,0.008035579,0.0003992179,0.001780049],"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":[0.0009188137,0.001378393,0.0008393781,0.0002889879,0.0003086825,0.002430201,0.00006757987,0.00001576387,0.000193285,0.000003216619,0.9934952,0.00006048795],"study_design_scores_gemma":[0.002164877,0.0008871913,0.003148411,0.000157323,0.0008807411,0.0001289877,0.0001791373,0.00005081252,0.0001167695,0.00007886059,0.989516,0.002690896],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01375737,0.0001400204,0.0000575347,0.00005092453,0.001373179,0.002618149,0.9811587,0.0008025101,0.00004166719],"genre_scores_gemma":[0.0003700114,0.00009037865,0.009791296,0.001655548,0.001563035,0.0002589081,0.9858914,0.0002718945,0.0001075504],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.07150475,"threshold_uncertainty_score":0.9996952,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6944604238","doi":"10.21233/hfcf-0j34","title":"Lake Manitoba ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); Surface (topology); Hydrology (agriculture); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.0464830546314058,"gpt":0.3116888561157952,"spread":0.2652058014843894,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002204526,0.002356358,0.002262261,0.0004535107,0.0008223701,0.0005482822,0.005342504,0.002093144,0.07458737],"category_scores_gemma":[0.005668518,0.002055153,0.0004696395,0.001128833,0.001753216,0.0008420132,0.003928144,0.003186389,0.1459519],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004404536,"about_ca_system_score_gemma":0.0005187733,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002051129,"about_ca_topic_score_gemma":0.2687262,"domain_scores_codex":[0.9875029,0.001441827,0.00197449,0.003949838,0.002180067,0.002950837],"domain_scores_gemma":[0.9856246,0.003409944,0.001344112,0.007726931,0.0003840852,0.001510328],"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":[0.0007473101,0.001699829,0.0007792742,0.0003315191,0.0002474663,0.003574606,0.000002843248,0.00001048188,0.0001560862,0.000009642805,0.9923638,0.00007719061],"study_design_scores_gemma":[0.00170278,0.0009074087,0.003705384,0.0001620409,0.0006939752,0.0002578828,0.00002904454,0.00004293726,0.00007686104,0.00007612228,0.9899365,0.002409104],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.005138516,0.000160962,0.0000227361,0.00007571299,0.001866566,0.002692595,0.9891171,0.0008295192,0.00009631092],"genre_scores_gemma":[0.000046065,0.0003541554,0.005134834,0.0018107,0.001982098,0.0003385695,0.9900364,0.0002509893,0.00004614474],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.2666751,"threshold_uncertainty_score":0.9992024,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6944816815","doi":"10.21233/yjpa-cw32","title":"Site 24, Georgian Bay, Canadian Hydrographic Service pollen surface sample dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04014400315304083,"gpt":0.2839229327623963,"spread":0.2437789296093555,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002363993,0.002880518,0.002899993,0.001727924,0.001073736,0.000704124,0.006314511,0.002830021,0.04703969],"category_scores_gemma":[0.002154709,0.002714996,0.0006004682,0.003758756,0.0006929649,0.00106264,0.002975482,0.004542292,0.1943747],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00118993,"about_ca_system_score_gemma":0.002459924,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.5262645,"about_ca_topic_score_gemma":0.973439,"domain_scores_codex":[0.9843769,0.001842764,0.002032926,0.004769331,0.002165681,0.00481238],"domain_scores_gemma":[0.9811884,0.002684681,0.001216127,0.01006591,0.0004579783,0.004386923],"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":[0.0004011815,0.0007234852,0.006124502,0.0006105117,0.0003729742,0.003580577,0.00001290485,0.000144322,0.0001134589,0.00001022525,0.9878674,0.0000384947],"study_design_scores_gemma":[0.002278952,0.0004967992,0.008927411,0.0002230849,0.0008499998,0.0002379041,0.00004692854,0.0003324164,0.000009514461,0.00002415353,0.9833651,0.003207696],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01602516,0.0004724054,0.00000241312,0.0008203032,0.001828576,0.004512675,0.9756613,0.0004933756,0.000183754],"genre_scores_gemma":[0.0005950096,0.0004973123,0.001162288,0.01073911,0.0006105824,0.0003097361,0.9856102,0.00036176,0.0001140187],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4471745,"threshold_uncertainty_score":0.9990618,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6926139745","doi":"10.21233/n3p94v","title":"Site 26 (Mott unpublished) pollen surface sample dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Smart Agriculture and AI","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02816435944357716,"gpt":0.2695232438418428,"spread":0.2413588843982657,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0007826911,0.00109593,0.001257961,0.00003797445,0.001325971,0.001051571,0.003497707,0.001273495,0.01803008],"category_scores_gemma":[0.001749248,0.0003862101,0.0004156201,0.0004062553,0.000382831,0.0007972736,0.002057095,0.001438344,0.006448648],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008880922,"about_ca_system_score_gemma":0.00006527078,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003719666,"about_ca_topic_score_gemma":0.03699938,"domain_scores_codex":[0.9944323,0.0004288058,0.0008252308,0.001961771,0.0009504182,0.001401472],"domain_scores_gemma":[0.9950932,0.001888236,0.0007869292,0.001270209,0.0001758924,0.0007855874],"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":[0.0001044698,0.0005130001,0.002103356,0.00004709456,0.00006330742,0.0004852063,0.000002580052,0.00000104175,0.001058474,0.000005581985,0.9943745,0.00124142],"study_design_scores_gemma":[0.0003825029,0.0004814366,0.04171622,0.00007220657,0.0001661028,0.00005120186,0.00002665219,0.000004273597,0.00004152066,0.00004678313,0.9559359,0.001075181],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.03129102,0.0003265582,5.417656e-7,0.001639679,0.001059577,0.001016082,0.9642144,0.0002915298,0.000160559],"genre_scores_gemma":[0.0006030306,0.000641122,0.0002110746,0.002480684,0.002280812,0.00009952488,0.9933543,0.000005626454,0.0003237832],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.03961287,"threshold_uncertainty_score":0.9999855,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963651761","doi":"10.21233/3yc9-q729","title":"Site 5, Georgian Bay, Canadian Hydrographic Service pollen surface sample dataset","year":2019,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03618938238550142,"gpt":0.2858668067461939,"spread":0.2496774243606924,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002350879,0.002760705,0.002789002,0.001667966,0.001021585,0.0006692254,0.006068929,0.002726604,0.04807707],"category_scores_gemma":[0.002084019,0.002598118,0.0005720089,0.003645648,0.0006559948,0.001008445,0.002825872,0.004223726,0.203769],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001174066,"about_ca_system_score_gemma":0.002408962,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.5461543,"about_ca_topic_score_gemma":0.9745979,"domain_scores_codex":[0.9849976,0.001748311,0.001945706,0.00458403,0.002082124,0.004642244],"domain_scores_gemma":[0.981777,0.002636441,0.001154861,0.009755532,0.0004525436,0.004223698],"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":[0.0003692514,0.0006948375,0.005970902,0.0005829787,0.0003529209,0.003315672,0.00001188886,0.0001329336,0.00009905775,0.00001045634,0.9884206,0.00003848825],"study_design_scores_gemma":[0.002167591,0.0004667808,0.0103243,0.0002113956,0.0008039325,0.0002218093,0.00004284464,0.0003172735,0.000008045902,0.00002330905,0.9823422,0.003070502],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01642503,0.0004285054,0.000002250946,0.0007403945,0.001348588,0.004325644,0.9760686,0.0004689218,0.0001920712],"genre_scores_gemma":[0.0006663273,0.0004445499,0.001032968,0.01031289,0.0005753121,0.0002894116,0.986246,0.0003421631,0.00009030198],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.4284436,"threshold_uncertainty_score":0.9993087,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963197343","doi":"10.21233/0afk-8p17","title":"Lake Ontario/Soup Harbour ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); Harbour; Biogeosciences; Hydrology (agriculture)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.04191524403828403,"gpt":0.3021973633052851,"spread":0.2602821192670011,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002551224,0.002724869,0.002727856,0.0004947074,0.0009363508,0.0006801892,0.00564825,0.002391887,0.1843992],"category_scores_gemma":[0.004604931,0.002408297,0.0005952317,0.001015041,0.001699608,0.001001278,0.004160018,0.004627363,0.1069775],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00108326,"about_ca_system_score_gemma":0.001578362,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.03710985,"about_ca_topic_score_gemma":0.9358524,"domain_scores_codex":[0.9859819,0.001318513,0.002288528,0.00454456,0.002437662,0.003428773],"domain_scores_gemma":[0.9848285,0.002973374,0.001598938,0.008285951,0.000408211,0.001905035],"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":[0.001052325,0.002077782,0.002644214,0.0002546212,0.000343897,0.00369029,0.00001279844,0.00001697392,0.0001560467,0.00001019251,0.9897091,0.00003177644],"study_design_scores_gemma":[0.002228657,0.001086773,0.01122729,0.0001906033,0.0008349021,0.000225084,0.00001137562,0.0000252832,0.00006855299,0.00007442148,0.9812329,0.002794132],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01111596,0.00009728256,0.00002215385,0.00008520068,0.00196764,0.00288128,0.982934,0.0007444393,0.0001520311],"genre_scores_gemma":[0.00006922929,0.0001386723,0.006068926,0.001970506,0.001367369,0.0003830141,0.989395,0.000264738,0.0003425089],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.8987426,"threshold_uncertainty_score":0.9997317,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6926064083","doi":"10.21233/n3zq80","title":"Lac Louis pollen dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Protist diversity and phylogeny","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Assemblage (archaeology); Table (database); Phenology","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02573800130060465,"gpt":0.3020086824278549,"spread":0.2762706811272502,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0003610588,0.0005732853,0.0005469356,0.00008169605,0.0005371666,0.0001670331,0.002361692,0.00105643,0.002315522],"category_scores_gemma":[0.0007308311,0.0005175194,0.0002211761,0.00005432071,0.0004435278,0.00001143381,0.002508638,0.0006364771,0.002498617],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002838498,"about_ca_system_score_gemma":0.0001966881,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001871056,"about_ca_topic_score_gemma":0.0009553141,"domain_scores_codex":[0.9972047,0.0002224991,0.000343271,0.001225617,0.0003391018,0.0006647899],"domain_scores_gemma":[0.9959812,0.00005430245,0.0003672414,0.003155293,0.00007388966,0.000368079],"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":[0.0003027492,0.0002965147,0.0009800461,0.0001054391,0.0001334325,0.0006703412,8.127793e-7,6.518485e-7,0.0006084774,0.000004828694,0.9966451,0.0002516234],"study_design_scores_gemma":[0.0007655267,0.0005578625,0.004806198,0.00002996426,0.0001466553,0.0000620914,0.000004029059,0.000001088153,0.0003400027,0.00001884015,0.9925987,0.0006690847],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002290227,0.0005449255,0.00001263224,0.0001416723,0.0006114258,0.0006562223,0.9954765,0.0000220692,0.0002443358],"genre_scores_gemma":[0.0005617944,0.001091615,0.0002282313,0.001591487,0.0009118535,0.00009079898,0.9950873,0.00002010345,0.0004168622],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.004046426,"threshold_uncertainty_score":0.9997277,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963627274","doi":"10.21233/3ydv-4176","title":"Lake Ontario/Wellington Bay water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02602915201791873,"gpt":0.2734806788983457,"spread":0.247451526880427,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001898253,0.002337405,0.002133868,0.0003014849,0.000688021,0.0004681926,0.004565104,0.002541095,0.3317007],"category_scores_gemma":[0.0008507377,0.001663212,0.0005085538,0.0004130896,0.001475863,0.0006025474,0.003897565,0.004043455,0.1587281],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008688176,"about_ca_system_score_gemma":0.0006324845,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003097564,"about_ca_topic_score_gemma":0.5628595,"domain_scores_codex":[0.9886824,0.0005732676,0.001885463,0.00393671,0.001864238,0.003057846],"domain_scores_gemma":[0.989911,0.0005474751,0.0007628839,0.00705455,0.000311792,0.001412275],"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":[0.0008530989,0.001503127,0.0004998305,0.000469116,0.0002652803,0.004468688,0.00001510968,0.000003118973,0.0010172,5.6609e-7,0.9908798,0.00002506974],"study_design_scores_gemma":[0.001748624,0.0004342374,0.0007145218,0.0001824151,0.0007250819,0.0003039341,0.000006691152,0.000009588585,0.002169667,0.00002841852,0.9912706,0.002406214],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01050311,0.00004919463,0.000004141031,0.00006931052,0.001239318,0.001651254,0.9854355,0.0005378902,0.0005102557],"genre_scores_gemma":[0.00004555122,0.00009611437,0.0005848178,0.001322079,0.002046798,0.0006386159,0.9936543,0.0002126987,0.001399016],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.5597619,"threshold_uncertainty_score":0.9989364,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963481671","doi":"10.21233/f7bb-av64","title":"FkPh-VP [Fort Saskatchewan] vertebrate fauna dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Fauna; Vertebrate; Taxon; Ectotherm; Biodiversity","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02853241011434994,"gpt":0.3035471006171193,"spread":0.2750146905027693,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002605762,0.00301249,0.00304518,0.0009413839,0.0009554272,0.000614198,0.006402064,0.002712372,0.0630869],"category_scores_gemma":[0.002675029,0.002528399,0.0006846647,0.001621877,0.002310551,0.001191524,0.005108901,0.003847187,0.2166023],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007498362,"about_ca_system_score_gemma":0.001272668,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008736821,"about_ca_topic_score_gemma":0.02625415,"domain_scores_codex":[0.9848059,0.001526852,0.002525366,0.004881372,0.002636811,0.003623734],"domain_scores_gemma":[0.9851039,0.001205458,0.001657741,0.009530379,0.0004427609,0.002059753],"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":[0.001002801,0.001883951,0.0003279734,0.0004112357,0.0004194286,0.006312057,0.00001178368,0.000002025353,0.0001945069,0.000008597909,0.9891849,0.000240774],"study_design_scores_gemma":[0.002613599,0.001341819,0.001528296,0.0002405847,0.001017845,0.0003894198,0.00007487778,0.00008330178,0.0001528419,0.00008285396,0.9893876,0.003087006],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.002404641,0.0002715914,0.00001476693,0.00009506792,0.003186012,0.003191778,0.9895862,0.0009873869,0.000262589],"genre_scores_gemma":[0.0001378031,0.0002858985,0.001378641,0.004082312,0.003129711,0.001126131,0.9892578,0.0003290389,0.0002727193],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1535154,"threshold_uncertainty_score":0.9989738,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963527136","doi":"10.21233/zh8q-tv24","title":"Montreal Lake ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); Raw data; Hydrology (agriculture); Biogeosciences","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03471035648026747,"gpt":0.3089511318896027,"spread":0.2742407754093352,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.002007257,0.002301614,0.002335645,0.0004185379,0.000741489,0.0004693509,0.004772669,0.002083223,0.07386746],"category_scores_gemma":[0.005512067,0.001971926,0.0004807615,0.001054759,0.001680277,0.0007932661,0.003412961,0.002934187,0.09494486],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003405809,"about_ca_system_score_gemma":0.0005422473,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004085558,"about_ca_topic_score_gemma":0.2643532,"domain_scores_codex":[0.9877773,0.001400009,0.001925718,0.003901292,0.00207157,0.002924147],"domain_scores_gemma":[0.9858878,0.003533995,0.001301933,0.007299553,0.0003548881,0.001621879],"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":[0.001016144,0.001856415,0.0005182271,0.0002451441,0.0002616101,0.003563776,0.000005208066,0.00001936545,0.0001049546,0.00000479472,0.9922835,0.0001208176],"study_design_scores_gemma":[0.002045082,0.0009556891,0.006442256,0.0001285243,0.0007095842,0.0002046728,0.00001770771,0.00008881101,0.00004531178,0.00007564059,0.9870483,0.002238409],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.003924004,0.0001752858,0.00002016849,0.00007940786,0.001347073,0.00246866,0.9910414,0.0007914493,0.0001526118],"genre_scores_gemma":[0.00007180862,0.0003856561,0.003224147,0.00134712,0.001654266,0.0003179682,0.9926844,0.0002289589,0.000085726],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.2602676,"threshold_uncertainty_score":0.9993661,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963240565","doi":"10.21233/4pkc-1652","title":"Montreal River ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); Hydrology (agriculture); Raw data; JSON","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03457881301172576,"gpt":0.3072172005494131,"spread":0.2726383875376874,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.001734901,0.002241591,0.002231115,0.0004164582,0.0007311074,0.000369363,0.004665034,0.002017841,0.04716757],"category_scores_gemma":[0.004640497,0.001913388,0.000478784,0.0009798449,0.00249022,0.0008583062,0.003588882,0.002779915,0.1081009],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005278498,"about_ca_system_score_gemma":0.0004990192,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.009952987,"about_ca_topic_score_gemma":0.04465941,"domain_scores_codex":[0.9881177,0.001401333,0.001776146,0.003860843,0.00204302,0.002800979],"domain_scores_gemma":[0.9864511,0.00334314,0.001254437,0.007024931,0.0003600644,0.001566252],"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":[0.001061568,0.001809168,0.000496218,0.0002122552,0.0002826395,0.003648184,0.000008838117,0.00001954647,0.0001241366,0.000006596938,0.9921954,0.0001354602],"study_design_scores_gemma":[0.002139732,0.0009049753,0.009177858,0.0001188231,0.0007697182,0.0001744894,0.00001378573,0.0001096121,0.00005606301,0.0001294445,0.984193,0.002212507],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.006188623,0.0002024228,0.00002521541,0.00005905922,0.001258554,0.002380364,0.9890752,0.0006885089,0.0001220435],"genre_scores_gemma":[0.000091116,0.0004160533,0.00538711,0.001279161,0.001479922,0.0002607187,0.9907933,0.0002138416,0.00007874606],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.06093335,"threshold_uncertainty_score":0.9995207,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6963501954","doi":"10.21233/9h5p-ev74","title":"North Saskatchewan River ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","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":"Paleoecology; Sample (material); Hydrology (agriculture); JSON; Raw data","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.03079982511111349,"gpt":0.2751116107433206,"spread":0.2443117856322072,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.000972242,0.000834233,0.001054444,0.0002016913,0.0007440927,0.0001121722,0.002333802,0.001240956,0.08389002],"category_scores_gemma":[0.001040091,0.0006535167,0.0001911533,0.0004951457,0.002290095,0.0003294244,0.0004036442,0.002072824,0.0224271],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000006748572,"about_ca_system_score_gemma":0.0004380294,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005484979,"about_ca_topic_score_gemma":0.5555419,"domain_scores_codex":[0.9940097,0.001021888,0.0007746911,0.001707415,0.0007615199,0.001724822],"domain_scores_gemma":[0.9934244,0.003202947,0.0003598083,0.002082677,0.0001346526,0.0007955586],"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":[0.0001983391,0.00009470061,0.4629148,0.00007510973,0.00005702618,0.0009768496,0.00000797672,0.00002191845,1.47097e-7,3.093289e-7,0.5355868,0.00006606843],"study_design_scores_gemma":[0.0004285441,0.0005166091,0.3396165,0.0000189469,0.0001037847,0.000149127,0.00002418211,0.0000858796,0.000002508226,0.0000615623,0.6584222,0.0005701321],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.1308852,0.000253192,0.000007840217,0.0001300192,0.0008301603,0.0007148543,0.8669899,0.00009439142,0.00009439298],"genre_scores_gemma":[0.005556377,0.0004935327,0.001963575,0.001326574,0.000457265,0.00001363273,0.9900958,0.000009568226,0.00008369364],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.5500569,"threshold_uncertainty_score":0.9995916,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}