{"meta":{"query_hash":"1da4e2f7a9b5","filters":{"venue":"Neotoma Paleoecological Database"},"cohort_total":67,"direct_labels_cover":0,"predictions_cover":67,"exported":67,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/1da4e2f7a9b5","api":"https://metacan.xera.ac/api/v1/cohort?venue=Neotoma+Paleoecological+Database"},"results":[{"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Fauna; Paleoecology; Vertebrate; Biodiversity; JSON","score_opus":0.031148313382502745,"score_gpt":0.3062672350064258,"score_spread":0.27511892162392304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888442610","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00058607594,0.000053808617,0.00004787453,0.00006090365,0.000020756826,0.000010939351,0.9972385,0.00012256442,0.001858555],"genre_scores_gemma":[0.0005937422,0.00003638465,0.00026875208,0.00003570854,0.00000885443,0.00005374841,0.99717486,0.000038830745,0.0017890997],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99954575,0.00004700627,0.00003208461,0.00013928574,0.00015249712,0.000083360945],"domain_scores_gemma":[0.9987685,0.00017370802,0.00011350171,0.00023462206,0.00053799694,0.00017166577],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006748777,0.0007655556,0.0006908476,0.0033127405,0.00060429366,0.0011499302,0.0016763941,0.0006516851,0.057149272],"category_scores_gemma":[0.0028371667,0.0003469725,0.00044341266,0.004979916,0.00018395006,0.0006907607,0.0012644876,0.0011062814,0.04291982],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030426529,0.000014377739,0.0017933131,0.00019180414,0.000017336975,0.000015611355,0.000024823796,0.000097623364,0.00011463584,0.0004990943,0.9941772,0.0030237942],"study_design_scores_gemma":[0.000065673106,0.0000071975696,0.024956314,0.00021600391,0.000015356674,0.000029750021,0.00009423427,0.00022070788,0.00015241884,0.000513102,0.97371477,0.000014517903],"about_ca_topic_score_codex":0.11059492,"about_ca_topic_score_gemma":0.28166658,"teacher_disagreement_score":0.8894051,"about_ca_system_score_codex":0.0015357696,"about_ca_system_score_gemma":0.0024979264,"threshold_uncertainty_score":0.2199024},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Diatom; Sample (material); Paleoecology; Hydrology (agriculture); Surface (topology)","score_opus":0.04201300915475259,"score_gpt":0.3139840234316292,"score_spread":0.2719710142768766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888452535","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00040224538,0.00005814707,0.00006884522,0.000029634677,0.000012595194,0.000009880632,0.9982886,0.00025430828,0.00087565574],"genre_scores_gemma":[0.0005468149,0.000031913332,0.00028688493,0.000016622615,0.0000026253783,0.000022679287,0.99823064,0.000042165997,0.0008196284],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994649,0.000033838984,0.000028756649,0.00012750718,0.00023453502,0.00011053332],"domain_scores_gemma":[0.9985417,0.00014757093,0.00007261601,0.00017779194,0.00087650114,0.00018386371],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007142077,0.0016329451,0.0009476443,0.0034304042,0.0011229583,0.0017918581,0.001957861,0.0010237279,0.029737556],"category_scores_gemma":[0.0024776384,0.0005786664,0.0006366983,0.008995297,0.00037135408,0.00052871415,0.0011069312,0.0013545377,0.030290954],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000057922534,0.00002063138,0.0022324836,0.00024511784,0.000029404537,0.000024803381,0.00003302606,0.0004483633,0.000215821,0.0003882306,0.99254817,0.0037560354],"study_design_scores_gemma":[0.00017447422,0.00001151877,0.02276283,0.00019020686,0.00004104162,0.000035765748,0.00012514953,0.0009382542,0.0006212944,0.0009476022,0.9741175,0.000034354307],"about_ca_topic_score_codex":0.6176875,"about_ca_topic_score_gemma":0.8181609,"teacher_disagreement_score":0.38231248,"about_ca_system_score_codex":0.004358761,"about_ca_system_score_gemma":0.008147917,"threshold_uncertainty_score":0.76912796},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","score_opus":0.04802777879815841,"score_gpt":0.3404180817763692,"score_spread":0.2923903029782108,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888460439","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0001944608,0.000012996888,0.000056498993,0.000015422296,0.000007870789,0.000007040945,0.999073,0.00010880013,0.00052390626],"genre_scores_gemma":[0.000506249,0.0000128792835,0.00022586597,0.000013904553,0.0000037738307,0.000048743124,0.99841547,0.00007256253,0.00070056785],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9993104,0.000080946826,0.000077024175,0.00017111668,0.00026262063,0.00009779978],"domain_scores_gemma":[0.9974202,0.00042478307,0.0003107068,0.0005262773,0.0010813199,0.00023673332],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009152985,0.0010237874,0.0009512319,0.0027788444,0.00053154794,0.0016616384,0.0016739785,0.0009982407,0.0534874],"category_scores_gemma":[0.0044746483,0.0005563913,0.0006795838,0.005743844,0.00027775965,0.00084184494,0.001328741,0.0013398813,0.07375102],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004884117,0.000014835748,0.0015871243,0.00027932093,0.000025831252,0.000015513693,0.000027042355,0.00017609294,0.00015827,0.00034997877,0.99541795,0.0018992176],"study_design_scores_gemma":[0.0002335077,0.000014164248,0.013594819,0.00014775626,0.000028187727,0.00003414217,0.00010065277,0.00036977325,0.00041701415,0.0006844519,0.98435163,0.000023820307],"about_ca_topic_score_codex":0.029685078,"about_ca_topic_score_gemma":0.060709238,"teacher_disagreement_score":0.0534874,"about_ca_system_score_codex":0.0010742045,"about_ca_system_score_gemma":0.002508986,"threshold_uncertainty_score":0.1789332},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bay; Paleoecology; Water chemistry; Hydrology (agriculture); Water quality; Limnology","score_opus":0.027167411087118665,"score_gpt":0.27734727573557133,"score_spread":0.25017986464845265,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888486444","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00025399288,0.000036816215,0.000049458362,0.000039164504,0.000011686592,0.00000897353,0.99884653,0.00015032702,0.0006030557],"genre_scores_gemma":[0.0006424468,0.000038799353,0.00016238775,0.000018206894,0.000004132985,0.000036638834,0.99825865,0.00003498991,0.00080365577],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99933296,0.000042976462,0.000051923176,0.00017252765,0.00026440277,0.00013523111],"domain_scores_gemma":[0.9978854,0.00015440639,0.0001407743,0.00026278992,0.0013153519,0.00024139487],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00056431454,0.0015181748,0.0012063403,0.0025935955,0.001257679,0.0014378267,0.002339962,0.0010805983,0.030603914],"category_scores_gemma":[0.002813865,0.0005529496,0.00083770003,0.0071149343,0.0005156629,0.0007849763,0.001341662,0.0011937165,0.028802868],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004129278,0.0000129437185,0.0017914791,0.00026550493,0.000028737166,0.00001968125,0.000027296897,0.00026146244,0.00009186698,0.00026578436,0.99543864,0.001755407],"study_design_scores_gemma":[0.00017484829,0.000008296055,0.017969534,0.00016069392,0.000038559403,0.000029784864,0.00010446665,0.0007339105,0.00032106997,0.000577434,0.9798473,0.00003409469],"about_ca_topic_score_codex":0.69597816,"about_ca_topic_score_gemma":0.86692166,"teacher_disagreement_score":0.30402184,"about_ca_system_score_codex":0.0067484854,"about_ca_system_score_gemma":0.011284344,"threshold_uncertainty_score":0.61162454},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","score_opus":0.025534151802320264,"score_gpt":0.2710167980926327,"score_spread":0.2454826462903124,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888494408","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003202305,0.000040758703,0.000045700286,0.00004937695,0.00001118864,0.000010606979,0.99864894,0.00008724202,0.0007859417],"genre_scores_gemma":[0.00087174226,0.00004734031,0.00016475415,0.000025406976,0.00000576839,0.000051040304,0.9977077,0.000029818231,0.0010964201],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992607,0.000047515572,0.000061081,0.00019809659,0.0003030067,0.00012951337],"domain_scores_gemma":[0.9979367,0.00012696035,0.00017052902,0.00025614141,0.0012703019,0.00023939389],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005612444,0.0010797308,0.0010971522,0.0025939262,0.0012082505,0.0014157164,0.0019170237,0.0008096224,0.033682376],"category_scores_gemma":[0.0030237255,0.00045140812,0.0007186784,0.007765421,0.00041446072,0.00073586975,0.0014195692,0.0009153399,0.029128652],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000051395775,0.000013022055,0.0032043792,0.00029717875,0.000041336873,0.000021029917,0.000033841712,0.0001725375,0.000121645724,0.00035238574,0.99351245,0.0021787721],"study_design_scores_gemma":[0.00016129951,0.00000938233,0.032784324,0.00016868503,0.000046517474,0.00003289247,0.00013591228,0.00040404254,0.00030931502,0.00059235585,0.96531904,0.000036288497],"about_ca_topic_score_codex":0.6804371,"about_ca_topic_score_gemma":0.85750747,"teacher_disagreement_score":0.3195629,"about_ca_system_score_codex":0.006939236,"about_ca_system_score_gemma":0.010977968,"threshold_uncertainty_score":0.64288974},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","score_opus":0.03618750031423847,"score_gpt":0.285837592773914,"score_spread":0.24965009245967554,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888495679","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00011644135,0.000020319423,0.000018007197,0.00002183976,0.0000071136988,0.000006185725,0.99933344,0.000071826435,0.0004048452],"genre_scores_gemma":[0.00030512403,0.00002353894,0.00012771973,0.000017504737,0.0000020451794,0.000026513868,0.99904984,0.00003122724,0.00041655742],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99906725,0.000058456073,0.00008019786,0.00023565079,0.00037033105,0.0001882497],"domain_scores_gemma":[0.9965996,0.00025823605,0.00017434791,0.0004860184,0.0020870427,0.00039474957],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007887727,0.0012958149,0.0010616083,0.0045568924,0.0017722347,0.0020231523,0.0028362006,0.0010754643,0.040188015],"category_scores_gemma":[0.005177229,0.000587016,0.0008295639,0.0125304265,0.0005114897,0.0008131883,0.0016069226,0.0016074054,0.040105995],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024534018,0.000007759661,0.001340613,0.00017924659,0.000019296147,0.000012642889,0.000022951637,0.00012779512,0.00007174812,0.00026894172,0.99672866,0.0011957925],"study_design_scores_gemma":[0.00014707354,0.0000068555114,0.025861172,0.00026497382,0.00003517375,0.000036342306,0.00020383435,0.00038133914,0.0002663211,0.0006545559,0.9720997,0.00004260524],"about_ca_topic_score_codex":0.8020492,"about_ca_topic_score_gemma":0.9122696,"teacher_disagreement_score":0.19795078,"about_ca_system_score_codex":0.0064497767,"about_ca_system_score_gemma":0.0174873,"threshold_uncertainty_score":0.39823306},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Diatom; Sample (material); Hydrology (agriculture); Paleolimnology; JSON","score_opus":0.03292175507514862,"score_gpt":0.3033355232338611,"score_spread":0.2704137681587125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888513827","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00035829828,0.000039181545,0.000027113047,0.0000333541,0.000010547208,0.000011936642,0.99886084,0.00008864754,0.0005699894],"genre_scores_gemma":[0.0010246127,0.000035759167,0.00017636245,0.000023591207,0.000003032765,0.00007423937,0.9973028,0.00003405626,0.001325647],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990715,0.00007400595,0.00008569895,0.0002800629,0.00029121185,0.00019748135],"domain_scores_gemma":[0.99706334,0.00019780164,0.0001904211,0.00048362603,0.0017404736,0.00032438707],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005856834,0.0013223252,0.0012030568,0.003075114,0.0019340164,0.0018276027,0.0023096302,0.0010949299,0.040474333],"category_scores_gemma":[0.00306328,0.0006376681,0.00081127323,0.0072886324,0.00060174096,0.00089247915,0.0019127222,0.0011911052,0.04272282],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001021591,0.000019797308,0.0053468132,0.0004861075,0.000060046845,0.000030174739,0.00007389044,0.00028262933,0.00028964586,0.00037947734,0.9896825,0.0032468124],"study_design_scores_gemma":[0.00011670103,0.000016225042,0.05516823,0.0003002413,0.00007818031,0.000048417412,0.00037315566,0.00044403077,0.0006917774,0.0005063498,0.9421718,0.00008499141],"about_ca_topic_score_codex":0.73804104,"about_ca_topic_score_gemma":0.9052848,"teacher_disagreement_score":0.26195896,"about_ca_system_score_codex":0.008983649,"about_ca_system_score_gemma":0.010799997,"threshold_uncertainty_score":0.52700335},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal Ontario Museum","funders":"","keywords":"Pollen; Vegetation (pathology); Paleoecology; Taxon","score_opus":0.047350425793273025,"score_gpt":0.3511017711469031,"score_spread":0.3037513453536301,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888548473","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008488255,0.000057899637,0.00008407067,0.00005487497,0.000022630811,0.000012423411,0.9976903,0.00046449254,0.0007644323],"genre_scores_gemma":[0.0006217784,0.000026243562,0.00018648834,0.000017144306,0.000004887861,0.000032969452,0.99845266,0.00005758135,0.00060029194],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99954396,0.00004754604,0.000051381543,0.00012036831,0.00015558438,0.000081095954],"domain_scores_gemma":[0.9988624,0.00022721979,0.00008416036,0.00022858559,0.00046122613,0.00013650302],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005447148,0.0013033089,0.0008823036,0.0024734773,0.0005896279,0.0013173447,0.0018409508,0.0011591751,0.025039347],"category_scores_gemma":[0.0029489247,0.00045606215,0.00070231996,0.0035904997,0.00030114016,0.0008708444,0.0013116172,0.0013346608,0.037093688],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044424207,0.000017849892,0.0015549122,0.0003118153,0.000036754576,0.00003327346,0.000030577037,0.00041474172,0.00016096448,0.00031135528,0.9942608,0.0028225062],"study_design_scores_gemma":[0.0002177856,0.000014763885,0.012230128,0.00014894082,0.000035655095,0.000059788315,0.00012867572,0.0011397537,0.00054437545,0.00072920555,0.98471874,0.00003224046],"about_ca_topic_score_codex":0.040740438,"about_ca_topic_score_gemma":0.11324737,"teacher_disagreement_score":0.040740438,"about_ca_system_score_codex":0.000982379,"about_ca_system_score_gemma":0.0021416375,"threshold_uncertainty_score":0.08376497},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Fauna; Vertebrate; Paleoecology; Invertebrate","score_opus":0.02905637955873201,"score_gpt":0.3173664998443533,"score_spread":0.28831012028562125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888554482","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00033264837,0.000046652425,0.000034722543,0.000020755579,0.000008893214,0.000006167003,0.99892896,0.0001668032,0.00045423672],"genre_scores_gemma":[0.00061355357,0.000026968735,0.00012539778,0.000009826992,0.000002553101,0.0000210666,0.9986951,0.000027822134,0.0004778178],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996001,0.000031219926,0.000032268203,0.00011173208,0.00014354155,0.00008109471],"domain_scores_gemma":[0.9986712,0.00013959748,0.00011709612,0.00021669708,0.0006322012,0.0002232336],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003786906,0.0014361973,0.0009223528,0.0026880559,0.000726759,0.0012408213,0.00179749,0.0010691595,0.036243133],"category_scores_gemma":[0.0024927906,0.00047398603,0.00066464813,0.004839061,0.00035617026,0.00062223675,0.0012231593,0.0010500826,0.034046832],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048797883,0.000010304963,0.00179972,0.00033843095,0.00003638888,0.00002773825,0.000040815034,0.00029898604,0.00014153057,0.0002899019,0.9947918,0.002175598],"study_design_scores_gemma":[0.0001082323,0.0000106118,0.0236631,0.000199325,0.000044058997,0.00006257688,0.00012478174,0.00075343286,0.00037814616,0.00045011484,0.9741769,0.00002870585],"about_ca_topic_score_codex":0.3205775,"about_ca_topic_score_gemma":0.6105521,"teacher_disagreement_score":0.6794225,"about_ca_system_score_codex":0.0029439323,"about_ca_system_score_gemma":0.0037049307,"threshold_uncertainty_score":0.63742304},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); JSON; Hydrology (agriculture); Raw data","score_opus":0.0360434516052837,"score_gpt":0.2991243810822944,"score_spread":0.26308092947701067,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888558042","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00092483935,0.000026084337,0.00006252177,0.00004885313,0.000013491283,0.00001516066,0.9976234,0.000075818556,0.0012098833],"genre_scores_gemma":[0.0019046924,0.000046866127,0.00024877515,0.000043886346,0.0000034700126,0.00008683897,0.99543446,0.000047580605,0.0021834376],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99955696,0.000044549302,0.000049650258,0.0001161876,0.00013043144,0.000102137594],"domain_scores_gemma":[0.9976501,0.00017518018,0.00011087721,0.00035579037,0.0015223585,0.00018568333],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051848905,0.0007482486,0.0006368522,0.0022578056,0.00077108963,0.0012974871,0.0012031868,0.000596704,0.037809305],"category_scores_gemma":[0.0026102455,0.00039140525,0.0005592118,0.006799828,0.00032534555,0.0006206189,0.0013484135,0.00094021345,0.03545912],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007405534,0.000016480824,0.009447698,0.0002448362,0.00007362711,0.000049384253,0.00006401862,0.00041359477,0.00036502726,0.00066957896,0.98366344,0.004918368],"study_design_scores_gemma":[0.00012979681,0.000008223391,0.0740141,0.00034266894,0.00004822552,0.000047300557,0.00059495564,0.0006911766,0.0007170365,0.0012298182,0.9221176,0.0000591416],"about_ca_topic_score_codex":0.5348463,"about_ca_topic_score_gemma":0.8006487,"teacher_disagreement_score":0.4651537,"about_ca_system_score_codex":0.003691174,"about_ca_system_score_gemma":0.010625313,"threshold_uncertainty_score":0.9357861},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Fauna; Vertebrate; Government (linguistics); JSON","score_opus":0.008006067275777419,"score_gpt":0.19679940274108668,"score_spread":0.18879333546530927,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888560324","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00012661579,0.000027942957,0.000029293378,0.000031062966,0.000008171729,0.0000068209542,0.9988005,0.00014072626,0.00082882895],"genre_scores_gemma":[0.00042935193,0.000044833054,0.00015776315,0.000017894996,0.0000026157136,0.00002326556,0.99851805,0.000042784817,0.00076343137],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988697,0.00005176258,0.00006576322,0.00019759405,0.0005561097,0.00025908335],"domain_scores_gemma":[0.99534625,0.00021777354,0.00019632535,0.0004768002,0.0032759875,0.00048683063],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00064720545,0.0015264519,0.0011950465,0.005330352,0.0020709934,0.0022938596,0.0028012134,0.00092843134,0.03549214],"category_scores_gemma":[0.0050315903,0.0006499726,0.00077313144,0.014442686,0.00057459367,0.0010064221,0.0015506387,0.0015610482,0.03805408],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026713158,0.000006505675,0.00085583417,0.00013736436,0.000015250325,0.000011219733,0.000021101741,0.000115124545,0.000049977905,0.00029825175,0.99702543,0.0014371068],"study_design_scores_gemma":[0.000073852796,0.0000050959793,0.01688917,0.0002076219,0.00002764055,0.00003079098,0.00014464062,0.0004440242,0.0003060369,0.0004984513,0.9813363,0.00003633857],"about_ca_topic_score_codex":0.92964077,"about_ca_topic_score_gemma":0.9636479,"teacher_disagreement_score":0.07035923,"about_ca_system_score_codex":0.012437085,"about_ca_system_score_gemma":0.024467148,"threshold_uncertainty_score":0.14154714},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Sample (material); Paleoecology; Hydrology (agriculture); Surface (topology); Raw data","score_opus":0.03754192152436427,"score_gpt":0.3141483203499923,"score_spread":0.27660639882562804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888564111","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00047981073,0.00003688251,0.00003211662,0.00002131878,0.000008934841,0.000008333758,0.99884975,0.00015038415,0.00041234543],"genre_scores_gemma":[0.00091606274,0.000027471793,0.00015124855,0.0000092357905,0.0000021609392,0.000035544326,0.9980556,0.000034768695,0.0007679721],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99940825,0.000047832822,0.000058061592,0.00015870378,0.00018795027,0.00013922076],"domain_scores_gemma":[0.99771094,0.00022520669,0.00015093527,0.00039322226,0.0012653538,0.00025436553],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041195354,0.0014009752,0.0010355953,0.0030742355,0.0009548491,0.0014305797,0.0020132381,0.0010681943,0.02538728],"category_scores_gemma":[0.002486266,0.00054257066,0.0008418061,0.0058549126,0.00052246405,0.00070806465,0.0012417211,0.0011567643,0.029026033],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007702427,0.000021111051,0.0041872477,0.00037912186,0.000049537004,0.000030003508,0.00005386522,0.000528614,0.0002175054,0.00029134154,0.9913476,0.0028169968],"study_design_scores_gemma":[0.00011301163,0.000013041053,0.03945498,0.00022748434,0.000062555635,0.000056993955,0.00029671023,0.0010613297,0.00065857335,0.00042951686,0.95756376,0.00006214766],"about_ca_topic_score_codex":0.5815806,"about_ca_topic_score_gemma":0.8293304,"teacher_disagreement_score":0.41841942,"about_ca_system_score_codex":0.0044227284,"about_ca_system_score_gemma":0.006574404,"threshold_uncertainty_score":0.84176713},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Hydrology (agriculture); JSON; Glacial period","score_opus":0.040793541837072275,"score_gpt":0.3170129130044492,"score_spread":0.27621937116737694,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888578027","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014538155,0.0000768953,0.0001022627,0.00006319195,0.000021554113,0.000029809053,0.99689853,0.0001452221,0.0012086491],"genre_scores_gemma":[0.0016298423,0.000044030407,0.00036423324,0.000026293887,0.000004928353,0.00008630277,0.9965223,0.000048408,0.0012737025],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999637,0.00003047005,0.00003007471,0.00010620164,0.000118503944,0.00007786723],"domain_scores_gemma":[0.9991949,0.000051108003,0.000060637405,0.00010996615,0.00048278173,0.00010066916],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036428188,0.00083712063,0.0007469673,0.0022075074,0.0008552097,0.0010258182,0.001268939,0.00055100356,0.017513955],"category_scores_gemma":[0.0021404966,0.0003705072,0.0005363564,0.0042288336,0.00026538206,0.00051795924,0.0013530414,0.0007013418,0.017960263],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000081305276,0.000016991116,0.0055112788,0.000286616,0.0000488187,0.000054062424,0.000060253617,0.00024755514,0.00040169753,0.0004243099,0.98761916,0.005247894],"study_design_scores_gemma":[0.00010828623,0.000009909605,0.06170475,0.00017510403,0.000035774723,0.00006357756,0.00023100067,0.000574757,0.0004376935,0.0004505077,0.9361764,0.00003223552],"about_ca_topic_score_codex":0.4472226,"about_ca_topic_score_gemma":0.65955013,"teacher_disagreement_score":0.5527774,"about_ca_system_score_codex":0.0023478277,"about_ca_system_score_gemma":0.004060801,"threshold_uncertainty_score":0.88923895},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","score_opus":0.027190837573171856,"score_gpt":0.269282397587167,"score_spread":0.24209156001399512,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6888610478","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003240473,0.00003565872,0.000041888805,0.000046869503,0.000010991291,0.0000105774425,0.99861264,0.00008253436,0.00083486026],"genre_scores_gemma":[0.00085305626,0.000044678563,0.00016249678,0.000024632607,0.000005307945,0.000049920884,0.9976484,0.000030447363,0.0011810611],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99928063,0.000044698256,0.000060274062,0.0001915502,0.00029954108,0.00012321475],"domain_scores_gemma":[0.998073,0.00011395457,0.00015498266,0.00023636292,0.001200714,0.00022087195],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051760837,0.0010574638,0.0010418067,0.0024209726,0.0012011251,0.0013743443,0.0018654873,0.00078162487,0.034098182],"category_scores_gemma":[0.0027471178,0.00045830786,0.0007168794,0.007567824,0.00040571572,0.0007084976,0.0014147055,0.0008997263,0.029234316],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053559037,0.00001371571,0.0035136484,0.00031202377,0.000043187938,0.000021773743,0.00003877901,0.00018903517,0.00014184353,0.00040077962,0.9929283,0.0023432346],"study_design_scores_gemma":[0.00014288818,0.000008150383,0.031708922,0.00015019004,0.000042087122,0.000029462944,0.00013723559,0.00038416428,0.0003140392,0.00059745024,0.966452,0.000033544155],"about_ca_topic_score_codex":0.6894192,"about_ca_topic_score_gemma":0.8638693,"teacher_disagreement_score":0.3105808,"about_ca_system_score_codex":0.006914089,"about_ca_system_score_gemma":0.011825456,"threshold_uncertainty_score":0.6248197},"labels":[],"label_agreement":null},{"id":"W6907175648","doi":"10.21233/n3js7k","title":"Hayes Lake pollen dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal Ontario Museum","funders":"","keywords":"Pollen; Vegetation (pathology); Raw data","score_opus":0.052115665267661114,"score_gpt":0.3579092766264014,"score_spread":0.3057936113587403,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907175648","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013114528,0.0000845031,0.00015488979,0.00006911955,0.000042311614,0.000019629553,0.99676275,0.00077754795,0.0007776351],"genre_scores_gemma":[0.0007798426,0.000024004778,0.000300488,0.000019325038,0.000006852096,0.00004350021,0.9983182,0.000065665576,0.0004421749],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993505,0.00008359525,0.000064679385,0.0001799407,0.00022400453,0.000097332784],"domain_scores_gemma":[0.9991171,0.00014422314,0.00007588322,0.00018992499,0.00035568004,0.00011721412],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007393221,0.0017662571,0.001108633,0.0025920616,0.00070570206,0.0012869647,0.0021579117,0.0014419622,0.023017902],"category_scores_gemma":[0.002920595,0.00051181985,0.0010711618,0.0036487163,0.0003352737,0.0010413835,0.0015132432,0.0016434501,0.036523245],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006353309,0.00003473841,0.0016469497,0.00030673045,0.00005070403,0.0000342046,0.00002817205,0.00056535634,0.00024049527,0.00029567067,0.9931271,0.0036063937],"study_design_scores_gemma":[0.0004943447,0.00003450455,0.015499997,0.00017714874,0.000057292535,0.00010293476,0.00013729342,0.003055016,0.0007287817,0.0013322616,0.9783328,0.000047625137],"about_ca_topic_score_codex":0.020534448,"about_ca_topic_score_gemma":0.056131072,"teacher_disagreement_score":0.023017902,"about_ca_system_score_codex":0.00078709685,"about_ca_system_score_gemma":0.0017099582,"threshold_uncertainty_score":0.077002585},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Cyberinfrastructure; Water chemistry; Raw data; JSON","score_opus":0.041679250930137496,"score_gpt":0.30025391767260456,"score_spread":0.2585746667424671,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907177935","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0004831156,0.000076089455,0.00009650685,0.0001201455,0.000023496743,0.000019540903,0.9979849,0.00019266951,0.0010036078],"genre_scores_gemma":[0.0020134458,0.000067523906,0.0005103407,0.00013932705,0.000015392,0.00010468336,0.9957495,0.00007285029,0.0013270037],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99916756,0.00012163826,0.000067982364,0.00022898099,0.0002600025,0.00015376562],"domain_scores_gemma":[0.9980275,0.00022601626,0.00016957922,0.0003525901,0.0009959408,0.00022838058],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008277123,0.0010198603,0.000897835,0.0024975424,0.00071906106,0.0011653862,0.0024577335,0.0009799149,0.03225411],"category_scores_gemma":[0.00464816,0.00036683163,0.00070993963,0.006024932,0.00034294426,0.0009468321,0.0016435339,0.001109917,0.026109388],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000050018738,0.0000130364015,0.0019981258,0.00024829188,0.000034399018,0.000018310031,0.000017805856,0.0002918053,0.00017302284,0.00044746185,0.9931804,0.0035274047],"study_design_scores_gemma":[0.00015330527,0.000014957475,0.023044517,0.00019230787,0.00004884418,0.000037085763,0.000106847394,0.0010648743,0.00060624187,0.0011633612,0.97350085,0.00006683017],"about_ca_topic_score_codex":0.36221063,"about_ca_topic_score_gemma":0.52985466,"teacher_disagreement_score":0.36221063,"about_ca_system_score_codex":0.0030334985,"about_ca_system_score_gemma":0.0048820567,"threshold_uncertainty_score":0.72020465},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Sample (material); Biogeosciences; Hydrology (agriculture)","score_opus":0.042390730293461035,"score_gpt":0.3058067063964637,"score_spread":0.26341597610300266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907193336","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007276983,0.00003949745,0.000055046155,0.00004269703,0.000010150809,0.000013952892,0.9978834,0.00010044368,0.0011270044],"genre_scores_gemma":[0.0013020321,0.000039851377,0.00018778206,0.000015785045,0.0000040306654,0.000055602584,0.99696285,0.00003974211,0.0013923505],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993693,0.00003179642,0.00004619826,0.0001426515,0.00027396413,0.00013597812],"domain_scores_gemma":[0.99784255,0.00012809812,0.00017559002,0.00026194192,0.001336,0.00025585745],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046351063,0.00090613955,0.0007839914,0.0027256603,0.001357755,0.001352,0.0014598345,0.0006132024,0.031131906],"category_scores_gemma":[0.0027755692,0.00041119175,0.0006201521,0.0068362174,0.0004432677,0.0006115067,0.001469606,0.000712881,0.024757659],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000060627404,0.000013737768,0.0055020503,0.00030734364,0.000038741156,0.000031426218,0.000075165466,0.00018878725,0.00026072923,0.00036547316,0.98998016,0.0031757024],"study_design_scores_gemma":[0.00011186155,0.000009380785,0.058805197,0.00016460706,0.00004014286,0.000035438956,0.00027351212,0.00039879326,0.0003582715,0.00039706955,0.9393718,0.00003394174],"about_ca_topic_score_codex":0.72220016,"about_ca_topic_score_gemma":0.9058696,"teacher_disagreement_score":0.72220016,"about_ca_system_score_codex":0.0063200817,"about_ca_system_score_gemma":0.010292808,"threshold_uncertainty_score":0.55887175},"labels":[],"label_agreement":null},{"id":"W6907228932","doi":"10.21233/ajdn-hf16","title":"Lake Manitoba/Balsam 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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Sample (material); Biogeosciences; Hydrology (agriculture)","score_opus":0.04577166984365422,"score_gpt":0.3126496723621249,"score_spread":0.26687800251847066,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907228932","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00083347154,0.000039523216,0.0000508767,0.00006309201,0.000015318292,0.000015611826,0.99804986,0.00012642532,0.00080584356],"genre_scores_gemma":[0.0011344042,0.000033630364,0.00021648688,0.0000270448,0.0000043100636,0.000084955864,0.9974542,0.000037923695,0.0010069577],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99961203,0.000040427738,0.00003362847,0.00010613024,0.00012390026,0.00008394896],"domain_scores_gemma":[0.99839777,0.00010465592,0.00012772233,0.00020993047,0.0009518798,0.00020808035],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00048525177,0.0010680945,0.0008124297,0.003112028,0.00095042796,0.0011779043,0.0017557582,0.00059277774,0.02889241],"category_scores_gemma":[0.002409163,0.00044237784,0.00058525486,0.007909798,0.00034830728,0.00053465716,0.0014666952,0.0009204149,0.028323501],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007545314,0.000027101729,0.0070576826,0.00027229128,0.000050627335,0.000033170632,0.000059960035,0.00021364093,0.00028693338,0.00034709243,0.988214,0.0033620873],"study_design_scores_gemma":[0.00017474477,0.000013915234,0.0640558,0.0002603545,0.000047366582,0.000049997503,0.00042543985,0.0005761635,0.00051916717,0.0006388235,0.9331995,0.000038725786],"about_ca_topic_score_codex":0.3611439,"about_ca_topic_score_gemma":0.62516487,"teacher_disagreement_score":0.6388561,"about_ca_system_score_codex":0.0030607078,"about_ca_system_score_gemma":0.006248844,"threshold_uncertainty_score":0.7180836},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Vegetation (pathology); Assemblage (archaeology); Table (database)","score_opus":0.053420458326485905,"score_gpt":0.3535709180005279,"score_spread":0.300150459674042,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907243641","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009965557,0.00006769334,0.0001321237,0.00005857024,0.000033541775,0.000017354641,0.997326,0.00065345404,0.00071478],"genre_scores_gemma":[0.0006998956,0.00002109055,0.00026017596,0.000020400063,0.000005940007,0.00004266356,0.9984067,0.00007643172,0.00046672366],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993363,0.00008879756,0.000069921836,0.00018374364,0.0002217585,0.00009950367],"domain_scores_gemma":[0.9990326,0.00016636025,0.000093114024,0.00019840826,0.00036626472,0.00014330499],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00066885754,0.0015602234,0.0009758808,0.0026651553,0.0006727932,0.0013349156,0.0020136598,0.0012115211,0.027128503],"category_scores_gemma":[0.0030111577,0.0005234677,0.00094828644,0.0036920337,0.0003278133,0.0009815533,0.0017826589,0.001429171,0.038721215],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006248659,0.000025472315,0.001740375,0.00032060925,0.00004612911,0.000032082502,0.00002971519,0.0004639464,0.00021440424,0.00026394942,0.993586,0.003214803],"study_design_scores_gemma":[0.00038506676,0.000023971223,0.018344004,0.00016455042,0.000054686265,0.0000907272,0.000119771095,0.0023550559,0.00067939673,0.0009903343,0.9767495,0.00004284154],"about_ca_topic_score_codex":0.02431441,"about_ca_topic_score_gemma":0.06706012,"teacher_disagreement_score":0.027128503,"about_ca_system_score_codex":0.00086823746,"about_ca_system_score_gemma":0.0018827899,"threshold_uncertainty_score":0.09075385},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","score_opus":0.03591422543043152,"score_gpt":0.28562268429791043,"score_spread":0.24970845886747892,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907246276","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00012110209,0.00002060294,0.000018205712,0.000022387472,0.0000074853247,0.0000065174945,0.9993131,0.00006977307,0.0004208137],"genre_scores_gemma":[0.00029668421,0.000021993399,0.000122066114,0.000017392154,0.0000021213984,0.000027179858,0.999054,0.000029495495,0.0004290162],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990295,0.00006184124,0.00008219343,0.00025595084,0.00037999335,0.00019055874],"domain_scores_gemma":[0.9965384,0.0002639366,0.00017873055,0.00050151465,0.0021144701,0.00040293168],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00080493564,0.0013272421,0.0010808726,0.0045793513,0.0017601463,0.0020677496,0.0028416598,0.0011108372,0.04096225],"category_scores_gemma":[0.0052571855,0.00060994376,0.0008186536,0.012363191,0.0004986647,0.0008473219,0.0016140161,0.0016235737,0.041908003],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024884752,0.00000844388,0.0014098315,0.00018700946,0.000019950832,0.000013263567,0.000023124398,0.00012956792,0.000073392875,0.00026843773,0.99664515,0.0011969365],"study_design_scores_gemma":[0.00015310381,0.00000728494,0.025879856,0.00026510365,0.000035650348,0.000037351125,0.00020470409,0.00038226056,0.0002666384,0.000621607,0.97210395,0.0000424424],"about_ca_topic_score_codex":0.7795087,"about_ca_topic_score_gemma":0.90329206,"teacher_disagreement_score":0.22049129,"about_ca_system_score_codex":0.006183922,"about_ca_system_score_gemma":0.016619444,"threshold_uncertainty_score":0.44357955},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Vegetation (pathology); Apoidea; Phenology","score_opus":0.05385351384362163,"score_gpt":0.3582293441080741,"score_spread":0.3043758302644525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907265724","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008952769,0.000057987418,0.00012259252,0.000053706466,0.000024575718,0.0000145334825,0.997629,0.0005449968,0.0006572969],"genre_scores_gemma":[0.00062456314,0.000018508556,0.00023883913,0.000018009026,0.0000052018368,0.000043809392,0.9986481,0.000055263557,0.00034776167],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993414,0.00009986896,0.00007264811,0.00017328253,0.00021461237,0.00009812541],"domain_scores_gemma":[0.99895644,0.00020921585,0.00010106612,0.00021960854,0.00038051966,0.00013308311],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000730722,0.0014696562,0.0010183157,0.0028617042,0.0006987849,0.0013340019,0.0019358525,0.0013749301,0.027357204],"category_scores_gemma":[0.0029693465,0.00049423473,0.00096443505,0.003964009,0.00033755388,0.0009471727,0.0015339953,0.0016277343,0.04017307],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006605426,0.000033712866,0.0019785007,0.00034559195,0.00003934591,0.000030794552,0.000028806735,0.00050741114,0.00021165401,0.0003013278,0.9932929,0.0031638856],"study_design_scores_gemma":[0.00043370438,0.000029120909,0.016653057,0.00021468449,0.000049080907,0.00009669117,0.00012375118,0.0023866873,0.0006919415,0.0011044426,0.97817516,0.00004168578],"about_ca_topic_score_codex":0.016959596,"about_ca_topic_score_gemma":0.048179436,"teacher_disagreement_score":0.027357204,"about_ca_system_score_codex":0.000835589,"about_ca_system_score_gemma":0.0019169624,"threshold_uncertainty_score":0.09151894},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","score_opus":0.03522344810136093,"score_gpt":0.28442288738439814,"score_spread":0.2491994392830372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907269138","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00011526868,0.000020274754,0.000017519702,0.000022389924,0.000007484895,0.0000061981577,0.9993243,0.00006545913,0.0004210764],"genre_scores_gemma":[0.0002975599,0.000024102112,0.00012037662,0.000016854348,0.0000021086134,0.00002595236,0.99904877,0.000028426459,0.00043588795],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99902725,0.00006144247,0.00008329204,0.00024049231,0.0003908422,0.00019675246],"domain_scores_gemma":[0.9964026,0.00026935778,0.00018168634,0.0004967197,0.0022353681,0.00041419524],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008207087,0.0013023199,0.0010607066,0.004633917,0.0017915256,0.00203154,0.0028107143,0.00107934,0.039981123],"category_scores_gemma":[0.005342597,0.0006116711,0.0007643077,0.012987191,0.00050268375,0.00082808384,0.0016345754,0.0015813179,0.041133832],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021713886,0.0000073976694,0.0013245483,0.00016917221,0.000017051587,0.000011993411,0.000021761485,0.00011877431,0.00006189045,0.00025394667,0.99682826,0.0011635217],"study_design_scores_gemma":[0.00013612678,0.0000066643433,0.02703642,0.00026008397,0.00003203421,0.00003359706,0.00020940731,0.00035706308,0.00024646294,0.00059715606,0.97104317,0.000041854386],"about_ca_topic_score_codex":0.8068377,"about_ca_topic_score_gemma":0.91451484,"teacher_disagreement_score":0.19316232,"about_ca_system_score_codex":0.0065732114,"about_ca_system_score_gemma":0.017562158,"threshold_uncertainty_score":0.38859975},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Water chemistry; Hydrology (agriculture); Paleolimnology; Biogeosciences","score_opus":0.025028360615342228,"score_gpt":0.28802691535721436,"score_spread":0.2629985547418721,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907283470","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00081794657,0.00008979758,0.00009403421,0.000080887876,0.000026580461,0.000032451702,0.99746895,0.00014900588,0.0012404326],"genre_scores_gemma":[0.0015852737,0.0000676545,0.000394053,0.00004407462,0.000008071042,0.000090993504,0.9961398,0.000045915836,0.0016241106],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99958736,0.000040371953,0.00003556369,0.0001244805,0.00014302431,0.00006933267],"domain_scores_gemma":[0.9992155,0.00004613691,0.00005723555,0.00009596135,0.00048162727,0.00010355137],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004368767,0.00088996725,0.0008700896,0.0022013497,0.00081026606,0.001058623,0.0014955179,0.0005158379,0.021192605],"category_scores_gemma":[0.0023936022,0.00036567278,0.00051218027,0.0047801333,0.00022328776,0.00062883133,0.0012420813,0.0007606372,0.018323885],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006547986,0.000015793805,0.0031152342,0.0003145522,0.00004828357,0.000036937985,0.00003797143,0.00022718853,0.00020752141,0.0004891949,0.99082434,0.004617667],"study_design_scores_gemma":[0.00011463129,0.000007985801,0.031013027,0.00012935948,0.000032635508,0.000045183224,0.000113537004,0.00050356303,0.00035161758,0.0005415661,0.96711695,0.000029901821],"about_ca_topic_score_codex":0.5226698,"about_ca_topic_score_gemma":0.66558206,"teacher_disagreement_score":0.4773302,"about_ca_system_score_codex":0.0031340537,"about_ca_system_score_gemma":0.005111835,"threshold_uncertainty_score":0.9602826},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Fauna; Vertebrate; Biodiversity; Biogeography","score_opus":0.032909564353330874,"score_gpt":0.28285446870590614,"score_spread":0.24994490435257527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907318632","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00044621387,0.000047796355,0.000030473138,0.000033046374,0.0000108535905,0.000010370545,0.9983321,0.000074642936,0.0010145549],"genre_scores_gemma":[0.0008781684,0.000051389048,0.00014961806,0.000018249317,0.000003172,0.000045193003,0.9972875,0.000025205658,0.0015413944],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994981,0.000026592075,0.000046505043,0.00011729617,0.00020568086,0.00010582405],"domain_scores_gemma":[0.9980101,0.00012497378,0.00012752572,0.00022207713,0.0012363477,0.00027900812],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00052491995,0.00081083615,0.0007755857,0.0035380982,0.0016289427,0.0016705483,0.0014902465,0.00054710475,0.03098401],"category_scores_gemma":[0.003002347,0.00052927964,0.00042546945,0.007826223,0.00031453752,0.0005726377,0.0010647269,0.0012844263,0.021696184],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028312663,0.0000104368855,0.0026508037,0.00016477118,0.000021739728,0.000017965738,0.000027513872,0.00009864633,0.00005668695,0.00021002264,0.99453235,0.0021807745],"study_design_scores_gemma":[0.00008453056,0.000006452237,0.07004451,0.00034642656,0.000044502056,0.00005084665,0.00022004804,0.00044994417,0.0002936486,0.00039529553,0.9280265,0.000037242622],"about_ca_topic_score_codex":0.7915096,"about_ca_topic_score_gemma":0.92566687,"teacher_disagreement_score":0.20849037,"about_ca_system_score_codex":0.005384606,"about_ca_system_score_gemma":0.011474864,"threshold_uncertainty_score":0.4194364},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Surface (topology); Hydrology (agriculture); Raw data","score_opus":0.0464830546314058,"score_gpt":0.3116888561157952,"score_spread":0.26520580148438944,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907320565","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011191267,0.000041492975,0.000061415056,0.00004313805,0.000012882273,0.000012567557,0.99793696,0.00023656161,0.0005358862],"genre_scores_gemma":[0.0010724037,0.000026723737,0.0002102167,0.000016439635,0.000003124284,0.000056387296,0.9978284,0.000036406618,0.0007498982],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996544,0.00003654507,0.000028678181,0.000090769296,0.00011361928,0.00007596292],"domain_scores_gemma":[0.99871016,0.00010094922,0.000093533075,0.00018120641,0.00075333,0.00016067186],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046314497,0.0014645527,0.00088635075,0.002736575,0.0007990082,0.0010201951,0.0017511785,0.00077838445,0.01962398],"category_scores_gemma":[0.0019636818,0.0005030034,0.0006667198,0.0058300416,0.0003599917,0.0005160265,0.0013004136,0.0010321357,0.023690375],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009203247,0.000038265563,0.0068720547,0.00035716867,0.0000618382,0.000042949294,0.00007124079,0.00041674223,0.00047980825,0.0002792618,0.9875236,0.0037650797],"study_design_scores_gemma":[0.00023520492,0.00002146973,0.069840215,0.00024838917,0.0000678072,0.00006702886,0.0003417893,0.0014763903,0.0008263291,0.00060090184,0.92622983,0.00004460216],"about_ca_topic_score_codex":0.29636958,"about_ca_topic_score_gemma":0.56302434,"teacher_disagreement_score":0.70363045,"about_ca_system_score_codex":0.002289529,"about_ca_system_score_gemma":0.0043084384,"threshold_uncertainty_score":0.58928907},"labels":[],"label_agreement":null},{"id":"W6907340645","doi":"10.21233/zbha-zj40","title":"Lake Manitoba water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Water chemistry; Paleolimnology; Hydrology (agriculture); Biogeosciences; Raw data","score_opus":0.028477342781000246,"score_gpt":0.2822142490333903,"score_spread":0.2537369062523901,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907340645","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00093962275,0.00006845067,0.00007280763,0.00012074203,0.000027107886,0.000027767328,0.99723935,0.00025036375,0.0012538481],"genre_scores_gemma":[0.0016289195,0.000053332442,0.00028033621,0.000062396495,0.0000083042805,0.00012383594,0.9960204,0.000045419663,0.0017771666],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995018,0.000062496896,0.000040319403,0.00014458045,0.00016415783,0.000086624226],"domain_scores_gemma":[0.9985171,0.00007797169,0.00010437337,0.00019682404,0.0009162377,0.00018755288],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005814875,0.0012635629,0.00097283314,0.0030248684,0.0011137961,0.0011599167,0.0021644784,0.00066998234,0.034584112],"category_scores_gemma":[0.0025655718,0.0004807294,0.00063622545,0.00831487,0.00033301962,0.0007629236,0.0015941839,0.0011543691,0.032951787],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007817442,0.000033276574,0.005623614,0.00029027901,0.00006353153,0.00003283899,0.000058193546,0.00022459835,0.00028373272,0.00042793807,0.9889613,0.003922375],"study_design_scores_gemma":[0.00018151611,0.000018490908,0.052806642,0.00023221335,0.000054755514,0.000050981813,0.00030163623,0.0008213892,0.00053505576,0.0008482395,0.9441006,0.00004849978],"about_ca_topic_score_codex":0.42385897,"about_ca_topic_score_gemma":0.6549268,"teacher_disagreement_score":0.576141,"about_ca_system_score_codex":0.004127628,"about_ca_system_score_gemma":0.006843693,"threshold_uncertainty_score":0.8427837},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","score_opus":0.046392339334393795,"score_gpt":0.3404525676084768,"score_spread":0.294060228274083,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907444215","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00019927608,0.000013647496,0.00005801101,0.000015789274,0.000007974875,0.0000073377246,0.99904627,0.00010890413,0.0005427649],"genre_scores_gemma":[0.0004998649,0.000013463518,0.00023660928,0.000014057694,0.0000040222612,0.00005002953,0.99841607,0.00007462508,0.0006911661],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992761,0.00008787332,0.00007961293,0.00017383808,0.0002797916,0.000102802434],"domain_scores_gemma":[0.99733704,0.0004357044,0.0003135787,0.00053541525,0.001136223,0.0002418794],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009537139,0.0010152787,0.0009880371,0.0028385185,0.0005467505,0.0017039712,0.0017287849,0.0010257497,0.05314314],"category_scores_gemma":[0.0044772867,0.00054746727,0.0007119132,0.0056331386,0.00028375347,0.00085850456,0.0013660282,0.0013580836,0.07421243],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000050636652,0.000015166806,0.0015937489,0.0002864057,0.000027476768,0.000015977657,0.00002847497,0.00017838657,0.00016800995,0.00037510027,0.9953113,0.0019494552],"study_design_scores_gemma":[0.00022952388,0.00001434943,0.013048762,0.00014726951,0.00002859075,0.000033116605,0.000101534635,0.00037598697,0.0004246573,0.0007001762,0.9848719,0.000024187415],"about_ca_topic_score_codex":0.027816266,"about_ca_topic_score_gemma":0.057917867,"teacher_disagreement_score":0.05314314,"about_ca_system_score_codex":0.0010389363,"about_ca_system_score_gemma":0.0025317818,"threshold_uncertainty_score":0.17778152},"labels":[],"label_agreement":null},{"id":"W6925638841","doi":"10.21233/dm5z-m225","title":"Lake Ontario/Wellers Bay water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"COVID-19 Impact on Reproduction","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","score_opus":0.03327612165398636,"score_gpt":0.3163501767722987,"score_spread":0.28307405511831235,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6925638841","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0002523321,0.000038705602,0.000040037936,0.000051298015,0.00001059461,0.000010901042,0.99881893,0.00007001746,0.00070727064],"genre_scores_gemma":[0.0008035027,0.000049037193,0.00015230478,0.00002766182,0.0000061815113,0.000055839522,0.9978054,0.000026209049,0.0010738139],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99929655,0.00005015174,0.00006504685,0.00018866679,0.00027577882,0.00012379256],"domain_scores_gemma":[0.9978393,0.00015795765,0.00019198116,0.00027321724,0.001268927,0.00026860763],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005979084,0.0010593241,0.0010432021,0.0024331594,0.0010827122,0.0013812502,0.001934101,0.00087825913,0.034877084],"category_scores_gemma":[0.0035210636,0.00044377038,0.0007376559,0.006828821,0.00042031653,0.0006497647,0.0013442982,0.00094109453,0.028893963],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000054082844,0.000012294721,0.0032109623,0.00029983983,0.0000406074,0.000022232503,0.000026836831,0.00018010243,0.000085778,0.000351727,0.9938134,0.0019021257],"study_design_scores_gemma":[0.00019951582,0.000009938681,0.030586302,0.0001963962,0.000049993443,0.00004040727,0.00012008426,0.00036538663,0.00025730013,0.0005996659,0.9675415,0.000033571276],"about_ca_topic_score_codex":0.625245,"about_ca_topic_score_gemma":0.8234915,"teacher_disagreement_score":0.37475502,"about_ca_system_score_codex":0.0061781895,"about_ca_system_score_gemma":0.010948847,"threshold_uncertainty_score":0.7539241},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Sample (material); Paleoecology; Hydrology (agriculture); Surface (topology); Raw data","score_opus":0.1328001093702441,"score_gpt":0.3985061333075233,"score_spread":0.2657060239372792,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6925831778","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00058569206,0.000023602492,0.000028889046,0.000022713883,0.00000836065,0.0000128610955,0.9988023,0.00010626395,0.00040933117],"genre_scores_gemma":[0.0013050743,0.000024912315,0.00017576705,0.000011523529,0.00000223514,0.00006436743,0.99751306,0.00003098514,0.0008721538],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99937624,0.000046603924,0.00007189289,0.00015033175,0.00020592858,0.00014900831],"domain_scores_gemma":[0.99705255,0.00027292295,0.0002000498,0.0004634339,0.001704557,0.0003064622],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045027115,0.0011432666,0.00088605937,0.0029187948,0.00095544517,0.0013073444,0.0019344507,0.0008930179,0.023594031],"category_scores_gemma":[0.0029016566,0.00050462206,0.000763512,0.0057088328,0.0005245221,0.00063233025,0.0011653507,0.0010869561,0.02026335],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011411963,0.000026381624,0.005292101,0.00036698254,0.000052518488,0.000036415702,0.00006153457,0.0006543806,0.0002699727,0.00038230303,0.9895079,0.0032354053],"study_design_scores_gemma":[0.00015220806,0.000015893494,0.05618062,0.00024073386,0.00007202125,0.000057652444,0.00035343884,0.0012226116,0.00089929,0.00048125957,0.94025034,0.000073967785],"about_ca_topic_score_codex":0.64957076,"about_ca_topic_score_gemma":0.8617076,"teacher_disagreement_score":0.35042924,"about_ca_system_score_codex":0.0057906965,"about_ca_system_score_gemma":0.009368999,"threshold_uncertainty_score":0.704986},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Diatom; Sample (material); Hydrology (agriculture); JSON; Paleolimnology","score_opus":0.06992012292253903,"score_gpt":0.3300284588049191,"score_spread":0.2601083358823801,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6925886978","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00046916498,0.000037105598,0.00006655678,0.000047670117,0.000023081328,0.000014383746,0.9985215,0.00017578425,0.0006447184],"genre_scores_gemma":[0.00080282555,0.000025529465,0.0002492349,0.000034252924,0.00000780621,0.0000629466,0.9982122,0.00003891717,0.00056640344],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99931824,0.00008273457,0.00007084685,0.0001890172,0.00022181669,0.00011733256],"domain_scores_gemma":[0.99852717,0.00019332912,0.0001388776,0.00030268,0.00063854276,0.0001994219],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007775535,0.0009625105,0.00081014604,0.0022031844,0.00071755366,0.0012971122,0.0017987535,0.0009196019,0.030059146],"category_scores_gemma":[0.0037122087,0.00038644506,0.00061170774,0.004231099,0.00029859232,0.00071566063,0.001727138,0.0011472655,0.04099048],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055436656,0.000018638395,0.0024695315,0.00023482976,0.000024636798,0.000022400656,0.000021481508,0.0001653755,0.00021960695,0.00028567621,0.9936807,0.002801613],"study_design_scores_gemma":[0.00017818957,0.000021692513,0.028548468,0.0002071171,0.000041027943,0.000061610626,0.0001524498,0.0007955405,0.0007069637,0.0008585257,0.96838254,0.000045851062],"about_ca_topic_score_codex":0.05645567,"about_ca_topic_score_gemma":0.120255,"teacher_disagreement_score":0.05645567,"about_ca_system_score_codex":0.0010361184,"about_ca_system_score_gemma":0.0022063544,"threshold_uncertainty_score":0.11225414},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Peninsula; Pollen; Shandong peninsula; Raw data","score_opus":0.23291275259776664,"score_gpt":0.48405788104162856,"score_spread":0.25114512844386194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926007326","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015978536,0.000083596795,0.000062630046,0.000069470734,0.000022557117,0.000017488666,0.99733245,0.00027900698,0.00053498015],"genre_scores_gemma":[0.001438302,0.000037599526,0.00023733127,0.000022058854,0.0000053332105,0.000060852453,0.9975878,0.000041868454,0.0005688673],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99968374,0.000042259162,0.000034225057,0.000094847645,0.00008921403,0.000055734927],"domain_scores_gemma":[0.9992932,0.00014002318,0.000080228245,0.00012841936,0.0002644334,0.00009374964],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006107716,0.0011864862,0.00088186533,0.002096586,0.00075458037,0.0011998147,0.0015084129,0.0010820796,0.01999011],"category_scores_gemma":[0.0020519267,0.0004610324,0.00059472333,0.0031243213,0.00027013355,0.00064448116,0.0010627867,0.0010768899,0.021830456],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000100507976,0.00003947587,0.003575816,0.00039634405,0.00005012057,0.00005995639,0.000045441087,0.00048216668,0.00025633493,0.00028746342,0.99083763,0.0038688078],"study_design_scores_gemma":[0.00058551534,0.0000321695,0.04671363,0.0003245869,0.00006508335,0.00012917697,0.00021906389,0.0022564696,0.0007055423,0.00084771885,0.94808245,0.000038632017],"about_ca_topic_score_codex":0.035551477,"about_ca_topic_score_gemma":0.08099873,"teacher_disagreement_score":0.035551477,"about_ca_system_score_codex":0.00093556894,"about_ca_system_score_gemma":0.0017668769,"threshold_uncertainty_score":0.07068908},"labels":[],"label_agreement":null},{"id":"W6926021723","doi":"10.21233/hf2z-ye48","title":"Lake Ontario/Salmon Point water chemistry dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrology (agriculture); Water chemistry; Biogeosciences; Paleolimnology","score_opus":0.044636070416648,"score_gpt":0.34629003133739694,"score_spread":0.3016539609207489,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926021723","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0004404674,0.000038813043,0.00006551748,0.000059608734,0.000013218326,0.000016859354,0.99840623,0.000086678854,0.00087264297],"genre_scores_gemma":[0.0010624509,0.00004342679,0.0002272829,0.000029451538,0.0000063523325,0.00008091882,0.99708,0.000029962042,0.001440103],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99927396,0.000057878224,0.0000626036,0.00018177634,0.00030199005,0.0001217623],"domain_scores_gemma":[0.9977608,0.00013044065,0.00016622295,0.0002492921,0.0014468576,0.00024633185],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006535517,0.00089107075,0.000905277,0.002058344,0.0011308921,0.0012102978,0.0018116324,0.0008074647,0.027967645],"category_scores_gemma":[0.003336639,0.00041098314,0.0007005806,0.0058009783,0.00041997308,0.00061289215,0.0013544126,0.00090753974,0.023998713],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005066168,0.000015714955,0.0041191704,0.00026049456,0.000036915306,0.000021730402,0.000036968893,0.00020654683,0.000107286076,0.00038003546,0.9924367,0.0023278329],"study_design_scores_gemma":[0.00017199347,0.000010713965,0.037636187,0.00017280714,0.000043002383,0.00003326112,0.00015943593,0.00049225334,0.0002913759,0.0006391673,0.96031183,0.00003805617],"about_ca_topic_score_codex":0.6941217,"about_ca_topic_score_gemma":0.87230426,"teacher_disagreement_score":0.6941217,"about_ca_system_score_codex":0.0064128903,"about_ca_system_score_gemma":0.0116709275,"threshold_uncertainty_score":0.6153593},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Assemblage (archaeology); Table (database); Phenology","score_opus":0.02573800130060465,"score_gpt":0.3020086824278549,"score_spread":0.27627068112725023,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926064083","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007377319,0.000071107526,0.00014094834,0.000043343924,0.00001866967,0.0000103658695,0.99744534,0.0007019951,0.00083047256],"genre_scores_gemma":[0.00066530844,0.000028499018,0.00032343625,0.00002602039,0.000004134363,0.000040871768,0.99835736,0.00009024206,0.00046404504],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995074,0.00007030964,0.000048120524,0.00014996012,0.00015844943,0.00006573934],"domain_scores_gemma":[0.9988801,0.00030554447,0.00010334459,0.00024404051,0.00035013343,0.00011690056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006608306,0.0011879117,0.000702653,0.002711823,0.00060491316,0.0015692485,0.0016599108,0.0012554342,0.0363758],"category_scores_gemma":[0.0034439724,0.00033310155,0.0008266234,0.0037610158,0.00028876774,0.0007846727,0.0014027148,0.0013451473,0.048129406],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000093549585,0.000032427695,0.0019672874,0.0004882305,0.000055223474,0.000051167943,0.000034172688,0.00059348665,0.0005537322,0.00078534806,0.9899558,0.0053895484],"study_design_scores_gemma":[0.00021297023,0.000015520807,0.00634412,0.0001474358,0.00003279707,0.00006448854,0.00006455924,0.00096617703,0.0006291767,0.0009964994,0.99050397,0.00002216952],"about_ca_topic_score_codex":0.013379832,"about_ca_topic_score_gemma":0.034712896,"teacher_disagreement_score":0.0363758,"about_ca_system_score_codex":0.0008920625,"about_ca_system_score_gemma":0.0018023838,"threshold_uncertainty_score":0.12168914},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","score_opus":0.045172043693490246,"score_gpt":0.335582708193329,"score_spread":0.29041066449983877,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926068258","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00019947281,0.000012459812,0.00005947754,0.000015049683,0.000007779608,0.0000075559756,0.99906117,0.000103146485,0.0005340303],"genre_scores_gemma":[0.0005280513,0.000013026328,0.00024558682,0.000013940428,0.000003886227,0.00005371053,0.9983069,0.00007730142,0.0007576154],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99928397,0.0000817402,0.00008029976,0.00017381924,0.00027891915,0.00010138747],"domain_scores_gemma":[0.9972953,0.00041453505,0.00032245592,0.00052886875,0.001187009,0.00025193274],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00097354886,0.0009905584,0.0009876041,0.0027610417,0.00055744103,0.0016760008,0.0017009362,0.0009568768,0.05306874],"category_scores_gemma":[0.0044698524,0.0005562898,0.0006743624,0.0058692913,0.0002851719,0.0008329723,0.0013680147,0.0013259035,0.072197415],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052812447,0.000015276082,0.0017277285,0.00027757406,0.000026736927,0.000015987058,0.000028595114,0.00017332859,0.00016304763,0.00036222234,0.9951556,0.002001143],"study_design_scores_gemma":[0.0002361244,0.000014531303,0.014505046,0.00014746835,0.000029216519,0.000032841122,0.000105368366,0.00037469363,0.00044342244,0.000689081,0.9833973,0.000024897605],"about_ca_topic_score_codex":0.032949638,"about_ca_topic_score_gemma":0.06514713,"teacher_disagreement_score":0.05306874,"about_ca_system_score_codex":0.0011229118,"about_ca_system_score_gemma":0.0027367065,"threshold_uncertainty_score":0.17753261},"labels":[],"label_agreement":null},{"id":"W6926104656","doi":"10.21233/nwxd-e552","title":"Lake Winnipeg water chemistry 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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Water chemistry; Hydrology (agriculture); Water quality; Water pollution; Raw data","score_opus":0.019963110606361016,"score_gpt":0.2736524536360762,"score_spread":0.2536893430297152,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926104656","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00066165207,0.00007995987,0.00007156676,0.000057143545,0.000022918895,0.000020395493,0.9982364,0.00021584326,0.0006341213],"genre_scores_gemma":[0.0009051487,0.000049770024,0.00024553985,0.000027195107,0.000004292399,0.000045621437,0.9979638,0.000032282056,0.00072634907],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99965584,0.000031765096,0.000031068463,0.00010403679,0.000112946276,0.00006427695],"domain_scores_gemma":[0.9993507,0.00005678401,0.000045584613,0.00008433291,0.00036893354,0.00009362816],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003785417,0.0013518668,0.0010442382,0.0020121157,0.00082550803,0.0010860949,0.0018240162,0.0008535897,0.016899664],"category_scores_gemma":[0.0020834913,0.00043512403,0.0006679775,0.0040745665,0.00029371175,0.00063077523,0.0012135404,0.0010644185,0.017232228],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008499028,0.000023894661,0.0021481554,0.00038533076,0.000054561537,0.00004411151,0.000027884851,0.00043607643,0.00027326075,0.00037774126,0.99222916,0.0039148508],"study_design_scores_gemma":[0.00021541861,0.00001119695,0.017473169,0.00013006198,0.00004513817,0.000047507117,0.000084971165,0.0010411718,0.0005425142,0.0006418607,0.97973746,0.000029457753],"about_ca_topic_score_codex":0.40434235,"about_ca_topic_score_gemma":0.5817844,"teacher_disagreement_score":0.59565765,"about_ca_system_score_codex":0.0026872964,"about_ca_system_score_gemma":0.0047079744,"threshold_uncertainty_score":0.8039776},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Water chemistry; Paleolimnology; Hydrology (agriculture); Biogeosciences; Raw data","score_opus":0.017203400787349974,"score_gpt":0.26888004372333696,"score_spread":0.25167664293598696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926108429","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005425776,0.00005646363,0.00006399235,0.00007327404,0.000014037308,0.000020396094,0.99798405,0.00010976545,0.0011354042],"genre_scores_gemma":[0.0011916559,0.0000585598,0.00021675859,0.000031448042,0.0000069946977,0.000091308706,0.9967565,0.000030026311,0.0016166744],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992207,0.000056273468,0.00006269131,0.00019161912,0.00034235168,0.00012639296],"domain_scores_gemma":[0.99806577,0.00011291566,0.00014384458,0.00022048397,0.0012530964,0.00020390344],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000594735,0.0009848614,0.00095238496,0.0023228982,0.0012273578,0.0012898367,0.0016742382,0.000792402,0.028011285],"category_scores_gemma":[0.003332207,0.00042032773,0.00065437757,0.006265415,0.00037071548,0.0007536691,0.00134597,0.00094684516,0.023139816],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055798726,0.000016098546,0.0037808365,0.00030105448,0.000039757404,0.00002570755,0.00004117924,0.00020724288,0.00013278832,0.00040419016,0.99189234,0.0031030304],"study_design_scores_gemma":[0.00013541187,0.000010130971,0.0336026,0.00015081916,0.000042339907,0.000037654336,0.00012839993,0.0005103036,0.00029770943,0.00054779445,0.9645038,0.000032960863],"about_ca_topic_score_codex":0.68136656,"about_ca_topic_score_gemma":0.8662782,"teacher_disagreement_score":0.31863344,"about_ca_system_score_codex":0.006942444,"about_ca_system_score_gemma":0.010703953,"threshold_uncertainty_score":0.6410198},"labels":[],"label_agreement":null},{"id":"W6926111188","doi":"10.21233/rsq3-3r91","title":"Lake Superior/Thunder Bay ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Steroid Chemistry and Biochemistry","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Sample (material); Biogeosciences; Hydrology (agriculture)","score_opus":0.019174622075233443,"score_gpt":0.28073276709881034,"score_spread":0.2615581450235769,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926111188","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009043909,0.000033286007,0.000061572544,0.000036719233,0.000009079318,0.000009302537,0.9982096,0.00016109538,0.0005751016],"genre_scores_gemma":[0.00069713366,0.000019304794,0.00016579873,0.000014468995,0.00000295396,0.00003412632,0.9986123,0.000026560427,0.0004274285],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999595,0.000036098154,0.000041851592,0.00012034806,0.00014035322,0.0000663965],"domain_scores_gemma":[0.99924624,0.00009294671,0.00008562473,0.00014235672,0.00030457924,0.00012819614],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051496003,0.0008704205,0.00066954986,0.0017800024,0.00060219516,0.0009980344,0.0011556899,0.00066877465,0.027397461],"category_scores_gemma":[0.0018832164,0.00029838402,0.0005939088,0.0032759178,0.0002679111,0.0005669143,0.0015232597,0.00091219775,0.029037498],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000102026905,0.000030980977,0.0048148683,0.00034221564,0.00005239368,0.000038881797,0.00003700304,0.00027779714,0.00059171976,0.00036585407,0.98919195,0.0041543213],"study_design_scores_gemma":[0.00028279726,0.000025585849,0.04805854,0.0001579104,0.000057270874,0.00006533875,0.00014315212,0.0007259501,0.00088584813,0.0008453447,0.9487225,0.000029727496],"about_ca_topic_score_codex":0.031527586,"about_ca_topic_score_gemma":0.07929364,"teacher_disagreement_score":0.9684724,"about_ca_system_score_codex":0.0009604709,"about_ca_system_score_gemma":0.0021171565,"threshold_uncertainty_score":0.091653585},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","score_opus":0.02816435944357716,"score_gpt":0.26952324384184284,"score_spread":0.2413588843982657,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926139745","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00018896854,0.000014656287,0.000056363177,0.0000151853155,0.000007904476,0.0000064100423,0.9990676,0.00014097428,0.00050202216],"genre_scores_gemma":[0.0004261902,0.000012822411,0.00020961638,0.000013091981,0.0000029537903,0.000036625774,0.9986816,0.000071607414,0.0005454761],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992988,0.000085780266,0.000077972465,0.00018061667,0.00026501907,0.00009171496],"domain_scores_gemma":[0.997738,0.00043573242,0.00024255586,0.00051290565,0.00086016324,0.0002106976],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009832124,0.0010598839,0.0009448036,0.0027161813,0.00049915427,0.001603551,0.001708278,0.0010841453,0.0477365],"category_scores_gemma":[0.0044368706,0.0005176022,0.00075878535,0.00538867,0.00030523018,0.0008042989,0.0012372902,0.0012692956,0.07127136],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006169203,0.000018148843,0.0014789541,0.000336877,0.000031346968,0.000020360041,0.000025193089,0.00028645483,0.00020982602,0.00038343554,0.99497235,0.0021754315],"study_design_scores_gemma":[0.00030379696,0.000016157868,0.011123994,0.00014081597,0.000029813371,0.000037027272,0.000081706756,0.0005044305,0.0005362956,0.0007648891,0.9864369,0.000024079978],"about_ca_topic_score_codex":0.023098877,"about_ca_topic_score_gemma":0.04871408,"teacher_disagreement_score":0.0477365,"about_ca_system_score_codex":0.00097166124,"about_ca_system_score_gemma":0.0023648343,"threshold_uncertainty_score":0.1596945},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Hydrography; Pollen; Georgian; Sample (material); Service (business)","score_opus":0.03314748601294594,"score_gpt":0.2896120739771524,"score_spread":0.25646458796420646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926272728","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00014741026,0.000020823547,0.000019123181,0.000022169743,0.0000054937263,0.000005762463,0.99932075,0.00007091848,0.00038742562],"genre_scores_gemma":[0.00046215212,0.00002677545,0.0001350703,0.00001929919,0.0000022062482,0.00002108887,0.998752,0.000024556946,0.0005569182],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99920136,0.000046557037,0.00006725474,0.00017809986,0.00035038198,0.00015622005],"domain_scores_gemma":[0.99651355,0.00027131787,0.00019351493,0.0003458765,0.002335083,0.00034062413],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00070872315,0.0013490437,0.0008884689,0.0040406673,0.001341563,0.0016832832,0.0024480356,0.0010702687,0.027021548],"category_scores_gemma":[0.0051508704,0.0005861793,0.0007629065,0.01105325,0.00044216157,0.0005710552,0.0010891597,0.0014007611,0.025935689],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033452845,0.00001061286,0.0021082284,0.00016244924,0.000021991867,0.000017597,0.000019589505,0.00029446493,0.00006270985,0.0002754102,0.99567825,0.0013152825],"study_design_scores_gemma":[0.0002401507,0.000009892607,0.048436914,0.00027644966,0.000046455807,0.00004676055,0.00018889664,0.0011492468,0.00040836164,0.00072415144,0.9484201,0.000052557552],"about_ca_topic_score_codex":0.8846955,"about_ca_topic_score_gemma":0.94553983,"teacher_disagreement_score":0.11530453,"about_ca_system_score_codex":0.007427975,"about_ca_system_score_gemma":0.01851958,"threshold_uncertainty_score":0.23196709},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","score_opus":0.02623565170144206,"score_gpt":0.2573196254817599,"score_spread":0.23108397378031784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926310424","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00019485204,0.000028573066,0.000025703046,0.000029920882,0.0000099492645,0.000009194758,0.9989555,0.00010159472,0.00064463547],"genre_scores_gemma":[0.00040782764,0.000029679119,0.00014593202,0.000019622314,0.000002482212,0.000027747581,0.9988412,0.000031495398,0.00049391994],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990237,0.00004970248,0.00007947852,0.00023040945,0.00042377005,0.00019305703],"domain_scores_gemma":[0.99636275,0.000211039,0.0001667413,0.00044798083,0.0023980425,0.00041344424],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006882141,0.0012775912,0.000964855,0.004800023,0.001881412,0.0019072009,0.00259356,0.0009835402,0.028143961],"category_scores_gemma":[0.005095852,0.00052566425,0.0007047951,0.013251785,0.00057936093,0.0008743515,0.0015419562,0.0015188552,0.034868862],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026322099,0.000009327221,0.0015029309,0.00016010269,0.000016090866,0.000015599939,0.00002449244,0.00013758268,0.00008681105,0.00028044844,0.9962586,0.0014816712],"study_design_scores_gemma":[0.00011993472,0.000007246624,0.030492283,0.00022758567,0.000029824287,0.00003976312,0.00024435364,0.00044133363,0.0003638061,0.0006077129,0.96738225,0.00004398904],"about_ca_topic_score_codex":0.83234257,"about_ca_topic_score_gemma":0.93010306,"teacher_disagreement_score":0.16765743,"about_ca_system_score_codex":0.0075700013,"about_ca_system_score_gemma":0.01928319,"threshold_uncertainty_score":0.33728957},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Hydrology (agriculture); JSON; Glacial period","score_opus":0.03069011364637703,"score_gpt":0.2983547078859254,"score_spread":0.26766459423954836,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6926447289","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012631819,0.000101530444,0.000083458144,0.000064781976,0.0000201002,0.000025560206,0.997349,0.00013201703,0.00096039864],"genre_scores_gemma":[0.0013098646,0.000054988595,0.00028758595,0.000028125898,0.000004630908,0.0000652504,0.99725455,0.000028219318,0.0009668081],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9996606,0.00003115543,0.000029238468,0.00010259152,0.000103391314,0.00007301448],"domain_scores_gemma":[0.9993556,0.00004782629,0.000052412714,0.00009148736,0.00035600932,0.000096661526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00035775133,0.00094596774,0.0007539292,0.0019182512,0.00081353594,0.0010019271,0.0013149449,0.0006562465,0.014293142],"category_scores_gemma":[0.0018276706,0.00032276066,0.0005851014,0.00355578,0.00029099194,0.00045790413,0.0012363348,0.00074063265,0.016125312],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013228947,0.000024166671,0.0057388046,0.00039658367,0.0000614203,0.000068095476,0.00004727274,0.00030554735,0.0005280085,0.00040305415,0.98666114,0.0056336294],"study_design_scores_gemma":[0.00014169562,0.000013611506,0.049013685,0.00017553697,0.00004495759,0.00007803853,0.00017536215,0.0005678267,0.0005246895,0.00043106408,0.9488044,0.00002922616],"about_ca_topic_score_codex":0.35606673,"about_ca_topic_score_gemma":0.575888,"teacher_disagreement_score":0.6439333,"about_ca_system_score_codex":0.002261878,"about_ca_system_score_gemma":0.0039031159,"threshold_uncertainty_score":0.7079884},"labels":[],"label_agreement":null},{"id":"W6944575782","doi":"10.21233/6fc9-k625","title":"Nunavut Mammoths vertebrate fauna dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Fauna; Vertebrate; Taphonomy; Biodiversity","score_opus":0.030818700680713695,"score_gpt":0.3072406250401667,"score_spread":0.276421924359453,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944575782","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00045400887,0.000064893655,0.00004269925,0.000036181573,0.000024160116,0.000010698629,0.9983388,0.00013113393,0.0008973851],"genre_scores_gemma":[0.00073637266,0.00003703325,0.0002134159,0.000023477149,0.0000070048286,0.000053826923,0.99805355,0.0000386786,0.0008366207],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99944407,0.000055036213,0.000053947733,0.00018792028,0.00017246541,0.00008645754],"domain_scores_gemma":[0.9985468,0.00017837944,0.00012955259,0.00025190346,0.00069560466,0.00019771901],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005340788,0.0010778861,0.0008571914,0.0030282352,0.0010720156,0.0012178208,0.0018206666,0.00064512616,0.03616932],"category_scores_gemma":[0.0034071126,0.00043043765,0.0005404872,0.004694485,0.00029891075,0.0007247415,0.001808252,0.00085083966,0.030722015],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000047681093,0.000015510963,0.0043744845,0.00029986963,0.000026348125,0.000023337003,0.000062177074,0.00015505338,0.00014397592,0.0004601894,0.9907957,0.0035956586],"study_design_scores_gemma":[0.00007637272,0.000010478486,0.033513837,0.00035743648,0.000029190967,0.00006812394,0.00021277538,0.00044679773,0.0003384118,0.0006932855,0.9642218,0.000031542408],"about_ca_topic_score_codex":0.21368034,"about_ca_topic_score_gemma":0.41106212,"teacher_disagreement_score":0.7863197,"about_ca_system_score_codex":0.0019938347,"about_ca_system_score_gemma":0.0030490074,"threshold_uncertainty_score":0.4248731},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Diatom; Sample (material); Paleoecology; Hydrology (agriculture); Paleolimnology","score_opus":0.030926823066879085,"score_gpt":0.2992857042299124,"score_spread":0.2683588811630333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944602275","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00028919644,0.000026998921,0.000026092071,0.000021202795,0.000008565576,0.000007335992,0.99918956,0.00012491191,0.00030619488],"genre_scores_gemma":[0.0005925131,0.000022093518,0.00015332723,0.000012105279,0.0000018661533,0.000038789127,0.9983468,0.000031991465,0.0008005192],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993037,0.000056459838,0.00006154182,0.00020380075,0.00021379492,0.00016061211],"domain_scores_gemma":[0.9974809,0.00020606894,0.00015648323,0.0003961494,0.0014644444,0.0002958881],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005445176,0.0018932988,0.0012604509,0.003238545,0.0016324361,0.001662674,0.0023901877,0.0013555256,0.035639476],"category_scores_gemma":[0.0024505951,0.0007763318,0.00096129114,0.006989971,0.00058094185,0.00076016947,0.0015255002,0.0013481439,0.038905796],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009280447,0.00002276183,0.0034706208,0.0004769456,0.00006410135,0.000027685972,0.00005109733,0.00049878046,0.00025218766,0.00026035603,0.9920689,0.0027138093],"study_design_scores_gemma":[0.00019457865,0.000018965087,0.043116078,0.00027888882,0.000099199155,0.00004745459,0.00026485062,0.00097504625,0.000912744,0.0005213257,0.9534852,0.00008569576],"about_ca_topic_score_codex":0.7173388,"about_ca_topic_score_gemma":0.8962933,"teacher_disagreement_score":0.2826612,"about_ca_system_score_codex":0.0069771204,"about_ca_system_score_gemma":0.009479304,"threshold_uncertainty_score":0.5686517},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Surface (topology); Hydrology (agriculture); Raw data","score_opus":0.0464830546314058,"score_gpt":0.3116888561157952,"score_spread":0.26520580148438944,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944604238","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011191267,0.000041492975,0.000061415056,0.00004313805,0.000012882273,0.000012567557,0.99793696,0.00023656161,0.0005358862],"genre_scores_gemma":[0.0010724037,0.000026723737,0.0002102167,0.000016439635,0.000003124284,0.000056387296,0.9978284,0.000036406618,0.0007498982],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9996544,0.00003654507,0.000028678181,0.000090769296,0.00011361928,0.00007596292],"domain_scores_gemma":[0.99871016,0.00010094922,0.000093533075,0.00018120641,0.00075333,0.00016067186],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046314497,0.0014645527,0.00088635075,0.002736575,0.0007990082,0.0010201951,0.0017511785,0.00077838445,0.01962398],"category_scores_gemma":[0.0019636818,0.0005030034,0.0006667198,0.0058300416,0.0003599917,0.0005160265,0.0013004136,0.0010321357,0.023690375],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009203247,0.000038265563,0.0068720547,0.00035716867,0.0000618382,0.000042949294,0.00007124079,0.00041674223,0.00047980825,0.0002792618,0.9875236,0.0037650797],"study_design_scores_gemma":[0.00023520492,0.00002146973,0.069840215,0.00024838917,0.0000678072,0.00006702886,0.0003417893,0.0014763903,0.0008263291,0.00060090184,0.92622983,0.00004460216],"about_ca_topic_score_codex":0.29636958,"about_ca_topic_score_gemma":0.56302434,"teacher_disagreement_score":0.70363045,"about_ca_system_score_codex":0.002289529,"about_ca_system_score_gemma":0.0043084384,"threshold_uncertainty_score":0.58928907},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrology (agriculture); Water chemistry; JSON; Raw data; Biogeosciences","score_opus":0.04957984495907053,"score_gpt":0.30799175714580934,"score_spread":0.2584119121867388,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944749696","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00043957317,0.000032194886,0.0000835212,0.000063733416,0.000020826643,0.000012522548,0.99846905,0.00018918756,0.0006892938],"genre_scores_gemma":[0.0011245753,0.00002873202,0.00030397042,0.000047913352,0.000009204049,0.00005231938,0.9976446,0.00004609145,0.0007425872],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993474,0.00009630463,0.00006766217,0.00017665426,0.00020881624,0.000103171114],"domain_scores_gemma":[0.99855417,0.00018949115,0.00013954651,0.0002718854,0.0006605375,0.00018443237],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007325934,0.0008299569,0.0007695613,0.0018381246,0.00054985745,0.0010662646,0.0019205513,0.00079577236,0.029190019],"category_scores_gemma":[0.0036268132,0.00034804456,0.00057635986,0.004156682,0.00026239915,0.0008241177,0.0013495616,0.001106038,0.032312084],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044076987,0.000017787359,0.0022528688,0.00017341289,0.00002451509,0.000016898908,0.00001526398,0.00025326872,0.00011857103,0.00038851646,0.9940182,0.0026766649],"study_design_scores_gemma":[0.00017031652,0.000017708468,0.024563044,0.000174509,0.000037915906,0.000055454482,0.00011915901,0.0015369979,0.0006007985,0.0012850404,0.9713836,0.0000554309],"about_ca_topic_score_codex":0.092361815,"about_ca_topic_score_gemma":0.17717987,"teacher_disagreement_score":0.092361815,"about_ca_system_score_codex":0.0011758975,"about_ca_system_score_gemma":0.0026296861,"threshold_uncertainty_score":0.18364841},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Water chemistry; Paleoecology; Hydrology (agriculture); Water quality; Paleolimnology; Limnology","score_opus":0.022111377524144386,"score_gpt":0.2359748199573171,"score_spread":0.2138634424331727,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944769255","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006284385,0.000043508124,0.000059413225,0.00007748576,0.000017779217,0.000015304602,0.9982027,0.00025399166,0.00070140115],"genre_scores_gemma":[0.0014394805,0.000044003402,0.0002452434,0.000044992987,0.00000539205,0.000077377736,0.99682343,0.00004772572,0.0012723273],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99962854,0.000043938475,0.000031951196,0.000106924104,0.000117626834,0.00007096767],"domain_scores_gemma":[0.9986896,0.00008974744,0.00009209909,0.00016398939,0.0008067377,0.00015782712],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00050814054,0.0013828811,0.00088984054,0.0025710405,0.0009005204,0.0011829253,0.002145347,0.0008183971,0.030818665],"category_scores_gemma":[0.002286542,0.00051862444,0.0006555201,0.006840904,0.00034606518,0.0007003923,0.0013157887,0.0012118841,0.028616488],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000092924005,0.000032041113,0.0043221097,0.00032228895,0.000065085376,0.000036624355,0.000042669995,0.00044718068,0.00024813405,0.00039267834,0.9907968,0.0032015215],"study_design_scores_gemma":[0.00027596692,0.000015614323,0.031101344,0.00021910536,0.000057755624,0.0000421358,0.00019881548,0.0013780927,0.000629314,0.00085666316,0.9651811,0.00004402297],"about_ca_topic_score_codex":0.42978293,"about_ca_topic_score_gemma":0.65331316,"teacher_disagreement_score":0.5702171,"about_ca_system_score_codex":0.0037118066,"about_ca_system_score_gemma":0.0067823306,"threshold_uncertainty_score":0.85456264},"labels":[],"label_agreement":null},{"id":"W6944776784","doi":"10.21233/wvem-0g61","title":"Site 9, 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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","score_opus":0.03612806788557004,"score_gpt":0.2857749148154152,"score_spread":0.24964684692984518,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944776784","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00011468327,0.00001965676,0.000016804484,0.00002136046,0.0000074818195,0.000006007421,0.99933964,0.00006385728,0.00041044835],"genre_scores_gemma":[0.00029795204,0.000023755883,0.00011984501,0.0000162751,0.000002044703,0.000024994179,0.9990716,0.000027454062,0.00041618233],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99904484,0.000059672355,0.00008349321,0.00023889504,0.0003810872,0.00019194126],"domain_scores_gemma":[0.9965714,0.00025403698,0.00017699695,0.00048108705,0.0021187288,0.00039771278],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007887227,0.0012902312,0.0010431261,0.0045955624,0.001756851,0.0020115832,0.0027933351,0.0010618839,0.03929234],"category_scores_gemma":[0.0052400795,0.00059772714,0.0007618222,0.012715061,0.0004938332,0.00081725273,0.0015978995,0.0015644633,0.04078828],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022909442,0.0000075322,0.0013487863,0.00017097524,0.00001769803,0.00001260616,0.000021611451,0.00012163139,0.000067902845,0.000259794,0.9967623,0.0011862705],"study_design_scores_gemma":[0.00013683717,0.000006698739,0.026739327,0.0002591701,0.000031595515,0.000034176686,0.00020708212,0.00035462648,0.00025477682,0.00060401135,0.9713302,0.000041542182],"about_ca_topic_score_codex":0.80452085,"about_ca_topic_score_gemma":0.9125826,"teacher_disagreement_score":0.19547915,"about_ca_system_score_codex":0.0064560496,"about_ca_system_score_gemma":0.017053844,"threshold_uncertainty_score":0.39326072},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Pollen; Apoidea; Vegetation (pathology); Aerial survey","score_opus":0.02883348087179219,"score_gpt":0.3106010849059356,"score_spread":0.28176760403414336,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944791880","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008413327,0.00003267979,0.000069565715,0.000028694958,0.000015564263,0.00001220733,0.997922,0.00034977522,0.00072815066],"genre_scores_gemma":[0.0007729254,0.000014684004,0.0002479474,0.000012122191,0.000003987529,0.000042368345,0.9983039,0.000053243773,0.00054881524],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995433,0.000046032088,0.000054583663,0.000121230296,0.00017764926,0.00005726017],"domain_scores_gemma":[0.99883586,0.00017129743,0.000121446414,0.00024051986,0.0004906357,0.00014016988],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005980508,0.0012804942,0.0009834995,0.0028069506,0.00050975743,0.0011097504,0.0016921608,0.0010072864,0.027150944],"category_scores_gemma":[0.0024591275,0.0004868423,0.0006713351,0.0047719963,0.00019945395,0.0005959319,0.00077679654,0.0012472526,0.032605994],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052633335,0.000025653493,0.0020033298,0.00022660266,0.000035246987,0.000024317025,0.000021216847,0.0005382826,0.0001944864,0.00021947044,0.9935347,0.0031240657],"study_design_scores_gemma":[0.0003237919,0.000024723999,0.0336136,0.00015035809,0.000045206994,0.00007269927,0.0000932237,0.0024749385,0.000747361,0.0007205467,0.9616957,0.000037834194],"about_ca_topic_score_codex":0.048271257,"about_ca_topic_score_gemma":0.07841603,"teacher_disagreement_score":0.95172876,"about_ca_system_score_codex":0.0009187651,"about_ca_system_score_gemma":0.0019304171,"threshold_uncertainty_score":0.095980585},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","score_opus":0.04014400315304083,"score_gpt":0.28392293276239633,"score_spread":0.2437789296093555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944816815","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000115461015,0.000021788783,0.000016863662,0.000022283375,0.0000075142766,0.0000060505204,0.9993216,0.000066683264,0.0004216746],"genre_scores_gemma":[0.00030111044,0.00002584393,0.000119453034,0.000016669332,0.0000022075178,0.000026461696,0.99903905,0.000029743798,0.00043942285],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99903154,0.000061547944,0.00008220394,0.00023983762,0.00039148773,0.00019344964],"domain_scores_gemma":[0.9963819,0.00026603186,0.00018481522,0.0004962346,0.0022509154,0.00042007965],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00081508205,0.0013321945,0.0010808237,0.004752367,0.0017536151,0.0020249565,0.0028708747,0.0010868893,0.039301235],"category_scores_gemma":[0.0054545463,0.00061198504,0.0007696603,0.013325856,0.00050051115,0.0008274213,0.0016066992,0.0015609041,0.042251147],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021453598,0.0000070591923,0.0012571603,0.00016449964,0.00001706357,0.0000115513285,0.000020868696,0.000114978975,0.000061210994,0.00023913474,0.99694484,0.0011401203],"study_design_scores_gemma":[0.00013846938,0.0000066114944,0.026818013,0.00026032008,0.00003240723,0.000032224598,0.00020279513,0.00034928255,0.00024640243,0.00058333494,0.97128814,0.000041965523],"about_ca_topic_score_codex":0.8104847,"about_ca_topic_score_gemma":0.9148798,"teacher_disagreement_score":0.1895153,"about_ca_system_score_codex":0.006641009,"about_ca_system_score_gemma":0.0178088,"threshold_uncertainty_score":0.38126272},"labels":[],"label_agreement":null},{"id":"W6944873594","doi":"10.21233/gbbv-se79","title":"Riviere Gatineau 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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); JSON; Hydrology (agriculture); Biogeosciences","score_opus":0.04142568637026676,"score_gpt":0.325911727338225,"score_spread":0.2844860409679582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944873594","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015547022,0.0001742098,0.00020822365,0.00003986381,0.000028848057,0.000019557936,0.9958444,0.00036488948,0.0017653066],"genre_scores_gemma":[0.0018013832,0.00005931093,0.00043967552,0.000019496636,0.000009961336,0.000066114524,0.9966018,0.000106469495,0.00089577376],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99926835,0.00010174931,0.000058950805,0.00022480146,0.0002254393,0.000120634104],"domain_scores_gemma":[0.99899,0.00014054037,0.00010767426,0.00024470757,0.00042717016,0.000089937945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006906583,0.0009618469,0.0008879585,0.0028338875,0.0005088906,0.0013222548,0.0012190907,0.0006129304,0.025765987],"category_scores_gemma":[0.0027017198,0.00031200543,0.0008621998,0.003913918,0.00026958864,0.00054253056,0.0013398603,0.0008942846,0.046317484],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011411684,0.000032354972,0.0068848706,0.00045808745,0.0000669331,0.000052576084,0.000059974354,0.00042935522,0.0007035482,0.0006415866,0.9817348,0.0088217445],"study_design_scores_gemma":[0.00009739635,0.00001607537,0.036242247,0.00019734472,0.000029443549,0.00009218012,0.00013638561,0.00047708736,0.00036758117,0.000555021,0.96176505,0.000024235684],"about_ca_topic_score_codex":0.029367488,"about_ca_topic_score_gemma":0.05885086,"teacher_disagreement_score":0.9706325,"about_ca_system_score_codex":0.000783804,"about_ca_system_score_gemma":0.001084324,"threshold_uncertainty_score":0.08619583},"labels":[],"label_agreement":null},{"id":"W6944888775","doi":"10.21233/n3km0h","title":"Lac Martyne pollen dataset","year":2017,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal Ontario Museum","funders":"","keywords":"Pollen; Vegetation (pathology); Bee pollen; Taxonomy (biology)","score_opus":0.05279827150362752,"score_gpt":0.35765451184628755,"score_spread":0.30485624034266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944888775","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011860977,0.00013324716,0.00014088463,0.000073702795,0.000035836423,0.000016043876,0.9964993,0.00095045136,0.00096440513],"genre_scores_gemma":[0.00084316835,0.000034899793,0.00029221937,0.000027549124,0.000008041012,0.00003772542,0.99823016,0.00008135591,0.00044488453],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.999185,0.00009958562,0.00006877298,0.00025482563,0.0002728923,0.000118940276],"domain_scores_gemma":[0.99879503,0.00025150465,0.000110934394,0.0003001832,0.00040159238,0.00014072577],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000732153,0.0018618017,0.0011059637,0.0030494058,0.00076849025,0.0018504692,0.0022997952,0.0017108577,0.023254972],"category_scores_gemma":[0.0034523124,0.00040774795,0.0012285291,0.0041106204,0.0003602495,0.0009883511,0.0015655712,0.0017345218,0.038610913],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008453291,0.000037803726,0.0014946389,0.0003583613,0.0000474397,0.00004885964,0.000029608656,0.00059061794,0.00037760363,0.00037469904,0.9921193,0.0044364305],"study_design_scores_gemma":[0.00027643552,0.000026420523,0.010815047,0.0001824585,0.00004160577,0.000121761106,0.000097951495,0.0020318255,0.00073716143,0.0007931874,0.98484117,0.000035025012],"about_ca_topic_score_codex":0.026386164,"about_ca_topic_score_gemma":0.056534566,"teacher_disagreement_score":0.026386164,"about_ca_system_score_codex":0.0011783779,"about_ca_system_score_gemma":0.0019517674,"threshold_uncertainty_score":0.077795565},"labels":[],"label_agreement":null},{"id":"W6944915319","doi":"10.21233/h7jh-sp94","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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Sample (material); Biogeosciences; Hydrology (agriculture)","score_opus":0.042390730293461035,"score_gpt":0.3058067063964637,"score_spread":0.26341597610300266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944915319","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007276983,0.00003949745,0.000055046155,0.00004269703,0.000010150809,0.000013952892,0.9978834,0.00010044368,0.0011270044],"genre_scores_gemma":[0.0013020321,0.000039851377,0.00018778206,0.000015785045,0.0000040306654,0.000055602584,0.99696285,0.00003974211,0.0013923505],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9993693,0.00003179642,0.00004619826,0.0001426515,0.00027396413,0.00013597812],"domain_scores_gemma":[0.99784255,0.00012809812,0.00017559002,0.00026194192,0.001336,0.00025585745],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046351063,0.00090613955,0.0007839914,0.0027256603,0.001357755,0.001352,0.0014598345,0.0006132024,0.031131906],"category_scores_gemma":[0.0027755692,0.00041119175,0.0006201521,0.0068362174,0.0004432677,0.0006115067,0.001469606,0.000712881,0.024757659],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000060627404,0.000013737768,0.0055020503,0.00030734364,0.000038741156,0.000031426218,0.000075165466,0.00018878725,0.00026072923,0.00036547316,0.98998016,0.0031757024],"study_design_scores_gemma":[0.00011186155,0.000009380785,0.058805197,0.00016460706,0.00004014286,0.000035438956,0.00027351212,0.00039879326,0.0003582715,0.00039706955,0.9393718,0.00003394174],"about_ca_topic_score_codex":0.72220016,"about_ca_topic_score_gemma":0.9058696,"teacher_disagreement_score":0.27779984,"about_ca_system_score_codex":0.0063200817,"about_ca_system_score_gemma":0.010292808,"threshold_uncertainty_score":0.55887175},"labels":[],"label_agreement":null},{"id":"W6945034636","doi":"10.21233/tr20-jj88","title":"Lake Superior/Thunder 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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","score_opus":0.027780072201174087,"score_gpt":0.2879465791948306,"score_spread":0.2601665069936565,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6945034636","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00045047535,0.000024775305,0.000047957423,0.000055847664,0.00001548187,0.000009175925,0.99872595,0.00016949506,0.00050083996],"genre_scores_gemma":[0.0005106682,0.000019045114,0.00014180857,0.000024404459,0.00000578896,0.00003761196,0.99881446,0.000025159146,0.00042107952],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994715,0.00005318347,0.00006121963,0.0001603154,0.00017955464,0.000074240896],"domain_scores_gemma":[0.9990042,0.00010493922,0.0001023401,0.00018680377,0.0004524434,0.000149342],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006298813,0.0011370898,0.00087981456,0.0019546899,0.0006613523,0.0011717767,0.0017055658,0.00089208,0.035873786],"category_scores_gemma":[0.0024320567,0.00034553953,0.0007034604,0.004560785,0.00025539208,0.0008776815,0.0017411627,0.0011766797,0.04203775],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000486157,0.000022386277,0.0023074506,0.00020205144,0.000035007324,0.000019789064,0.00002002598,0.00018795297,0.00015376306,0.00026938174,0.9943023,0.002431293],"study_design_scores_gemma":[0.00037766204,0.000025855765,0.028652279,0.00017362888,0.000052327156,0.000053888813,0.0001414621,0.0008747381,0.000582764,0.0013710655,0.967653,0.000041342584],"about_ca_topic_score_codex":0.045309175,"about_ca_topic_score_gemma":0.084613614,"teacher_disagreement_score":0.9546908,"about_ca_system_score_codex":0.0012029307,"about_ca_system_score_gemma":0.0024711136,"threshold_uncertainty_score":0.12000984},"labels":[],"label_agreement":null},{"id":"W6945038375","doi":"10.21233/f9n9-3a63","title":"Lake Ontario/Jordan Harbour ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Harbour; Biogeosciences; Hydrology (agriculture)","score_opus":0.03898332847075395,"score_gpt":0.27839532939285544,"score_spread":0.2394120009221015,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6945038375","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00070635346,0.000027982298,0.000044370398,0.000035097462,0.000008063937,0.000012020847,0.99808,0.000091600305,0.0009945215],"genre_scores_gemma":[0.0011183335,0.000030982243,0.0001859782,0.000016790798,0.0000035497658,0.00005675402,0.99728286,0.00003309776,0.0012715984],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995808,0.000021576809,0.00003245101,0.00009669345,0.00018132979,0.00008713502],"domain_scores_gemma":[0.9984484,0.00009329211,0.00012719457,0.00018292468,0.00095071236,0.00019743857],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003780808,0.0007500438,0.000640012,0.002364389,0.0010561012,0.0011273012,0.0013260968,0.000532666,0.031722404],"category_scores_gemma":[0.0021715094,0.00035676637,0.0004990358,0.005291371,0.00033163378,0.0005091992,0.00129383,0.00065368536,0.02400433],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005405223,0.00001391629,0.005603497,0.0002969504,0.00003266054,0.000033652883,0.00006401982,0.00016870387,0.00023111152,0.00029139494,0.98997504,0.0032349837],"study_design_scores_gemma":[0.000113869435,0.000009128927,0.06001169,0.00016351693,0.000032625292,0.000033499717,0.0002302535,0.00039580368,0.00036791872,0.00035843134,0.9382569,0.000026341215],"about_ca_topic_score_codex":0.5645206,"about_ca_topic_score_gemma":0.82638675,"teacher_disagreement_score":0.4354794,"about_ca_system_score_codex":0.0043319464,"about_ca_system_score_gemma":0.0076209907,"threshold_uncertainty_score":0.876088},"labels":[],"label_agreement":null},{"id":"W6945109322","doi":"10.21233/zxf1-1023","title":"Ottawa River ostracode surface sample dataset","year":2018,"lang":"en","type":"dataset","venue":"Neotoma Paleoecological Database","topic":"Historical Economic and Legal Thought","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Hydrology (agriculture); JSON; Raw data","score_opus":0.039561971866568436,"score_gpt":0.32570263367156627,"score_spread":0.2861406618049978,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6945109322","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006296596,0.000039934537,0.000033286527,0.000040740822,0.000012742913,0.0000150694905,0.99835557,0.00009523239,0.0007776338],"genre_scores_gemma":[0.0013162456,0.000033613993,0.00017290225,0.00001784754,0.000003769893,0.00007016114,0.99692106,0.00003364391,0.0014307672],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99926966,0.000068231,0.00007316775,0.00018769818,0.00023126583,0.0001700382],"domain_scores_gemma":[0.997191,0.00022895297,0.00019411834,0.0005049608,0.0015768976,0.00030414067],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046695597,0.000992772,0.00084849866,0.002877728,0.0011446849,0.0015379266,0.0019428171,0.000926211,0.032453705],"category_scores_gemma":[0.0032779037,0.0004308133,0.00072242075,0.005728766,0.0005528674,0.00073682587,0.001452454,0.0010989928,0.03370791],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006127001,0.000017926583,0.004592064,0.00022191553,0.000034253393,0.000027947848,0.000050413706,0.0002921947,0.00012794159,0.00038284573,0.9916334,0.002557935],"study_design_scores_gemma":[0.00008657286,0.000012169468,0.047305368,0.00023802124,0.000047696765,0.00005684153,0.00041999784,0.0006670612,0.00042220097,0.0005032974,0.9501789,0.00006188936],"about_ca_topic_score_codex":0.60546565,"about_ca_topic_score_gemma":0.84867024,"teacher_disagreement_score":0.39453435,"about_ca_system_score_codex":0.005125095,"about_ca_system_score_gemma":0.0077335294,"threshold_uncertainty_score":0.79371566},"labels":[],"label_agreement":null},{"id":"W6963065975","doi":"10.21233/e049-qy37","title":"Lake Ontario/Salmon Point 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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Point (geometry); JSON; Hydrology (agriculture); Raw data","score_opus":0.03982793848208121,"score_gpt":0.3010011508704095,"score_spread":0.2611732123883283,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963065975","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0004931464,0.00003277044,0.000052294,0.00003430116,0.000009786445,0.000012184655,0.99845374,0.00010930367,0.0008025738],"genre_scores_gemma":[0.000840751,0.000030692216,0.00016032385,0.000015580365,0.000003608962,0.00004883841,0.99772614,0.000033249933,0.0011407941],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993363,0.000038856317,0.000048923244,0.00016625377,0.00027563807,0.00013399096],"domain_scores_gemma":[0.9978809,0.00012643298,0.00015129882,0.00025513436,0.0013474756,0.00023880773],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004968991,0.0010664063,0.0009415671,0.0024507502,0.0013289872,0.0012967316,0.0016986738,0.0007563655,0.03385219],"category_scores_gemma":[0.0026853445,0.00049839355,0.0007582212,0.0060638757,0.0004436143,0.0006271181,0.0014958412,0.0008354572,0.032720473],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005160655,0.000014068126,0.004024218,0.0002816049,0.000036105255,0.000021335525,0.0000482573,0.00016743132,0.0001888159,0.00025370065,0.9924333,0.0024795902],"study_design_scores_gemma":[0.00014999733,0.000012908763,0.050747983,0.0001917533,0.000048468537,0.000037110727,0.00021903696,0.0005105554,0.00038849766,0.00046641578,0.94718784,0.00003945823],"about_ca_topic_score_codex":0.7266147,"about_ca_topic_score_gemma":0.8943057,"teacher_disagreement_score":0.7266147,"about_ca_system_score_codex":0.006048775,"about_ca_system_score_gemma":0.010791482,"threshold_uncertainty_score":0.5499906},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Harbour; Biogeosciences; Hydrology (agriculture)","score_opus":0.04191524403828403,"score_gpt":0.30219736330528507,"score_spread":0.26028211926700107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963197343","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007333828,0.000037360915,0.00005727453,0.000037422076,0.000009529128,0.00001248829,0.998063,0.00009220753,0.0009573803],"genre_scores_gemma":[0.0014702345,0.00003941142,0.00020174481,0.000015251045,0.000003862126,0.00005530122,0.9968501,0.00003608382,0.0013279472],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994518,0.00003007086,0.000042177617,0.00012469363,0.00023182078,0.000119350116],"domain_scores_gemma":[0.99814737,0.00011930841,0.00015271863,0.00023305848,0.0011304633,0.00021707555],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00044716068,0.0008680765,0.0007603626,0.0027186417,0.0012251678,0.0012004628,0.0013790808,0.0006004886,0.02767165],"category_scores_gemma":[0.0026370636,0.00040308788,0.00064655417,0.006118007,0.00041189126,0.00052966096,0.0014048852,0.0006882367,0.02045933],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007227782,0.00001464437,0.007494111,0.00039244504,0.000052114898,0.00003852733,0.00008191438,0.00023720715,0.0002977086,0.00038346596,0.9873284,0.0036071783],"study_design_scores_gemma":[0.00011943206,0.000010714517,0.07037837,0.00017805849,0.000046998208,0.00003802896,0.00026312377,0.00043520302,0.00039778164,0.0004270786,0.92766815,0.00003715142],"about_ca_topic_score_codex":0.7163743,"about_ca_topic_score_gemma":0.88951373,"teacher_disagreement_score":0.28362572,"about_ca_system_score_codex":0.0054130475,"about_ca_system_score_gemma":0.008908925,"threshold_uncertainty_score":0.57059205},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Hydrology (agriculture); Raw data; JSON","score_opus":0.03457881301172576,"score_gpt":0.30721720054941315,"score_spread":0.2726383875376874,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963240565","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00048552736,0.000052614556,0.000087273576,0.00004452888,0.000016563516,0.000018053943,0.9979997,0.00022945835,0.0010661863],"genre_scores_gemma":[0.00091240124,0.000029804,0.00026482943,0.000020228339,0.000005836355,0.00006193899,0.9976615,0.00007324609,0.0009702522],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991984,0.000072289266,0.00004753595,0.00022339512,0.00029667557,0.00016166514],"domain_scores_gemma":[0.9976375,0.00020729782,0.00014786593,0.0004184485,0.0013293714,0.00025957415],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006433032,0.0012278968,0.00092548627,0.0034610773,0.0010863749,0.0015629327,0.0023919358,0.00080983265,0.04811331],"category_scores_gemma":[0.0044951094,0.00045675086,0.00077742373,0.006909853,0.00041730958,0.00079247414,0.0015123845,0.0013000625,0.04357418],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002781444,0.0000115477515,0.0023745166,0.00011315239,0.00002704363,0.000015814838,0.000021856968,0.00019808114,0.00012873208,0.00030937963,0.9939594,0.0028127525],"study_design_scores_gemma":[0.00008978566,0.000009389559,0.035270687,0.00015767613,0.000030860818,0.00003661707,0.00012852851,0.00078974693,0.00033523503,0.00060686236,0.9624984,0.0000461987],"about_ca_topic_score_codex":0.50788414,"about_ca_topic_score_gemma":0.74464506,"teacher_disagreement_score":0.49211586,"about_ca_system_score_codex":0.0035278879,"about_ca_system_score_gemma":0.0061198743,"threshold_uncertainty_score":0.990028},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Fauna; Vertebrate; Taxon; Ectotherm; Biodiversity","score_opus":0.028532410114349944,"score_gpt":0.30354710061711926,"score_spread":0.27501469050276933,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963481671","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00009331623,0.000012541075,0.000022261975,0.000017052544,0.0000072327134,0.0000041774674,0.99942964,0.000082073966,0.00033173486],"genre_scores_gemma":[0.00034366446,0.00002256894,0.00013265983,0.000022902082,0.0000018368387,0.000036185163,0.9985807,0.00003659725,0.00082294905],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994449,0.00005147282,0.000067094545,0.00013413772,0.00015872384,0.00014375927],"domain_scores_gemma":[0.9969807,0.0002860493,0.00015131524,0.00043445642,0.0018574421,0.00028989904],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00065033627,0.0016652095,0.0010876438,0.003761438,0.00075443013,0.0017357293,0.0024497793,0.0013921964,0.06277223],"category_scores_gemma":[0.004020967,0.00066137593,0.0008318072,0.010377226,0.00038165212,0.0009523288,0.0014318104,0.0016925283,0.06915227],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000041741332,0.000009194152,0.0014294075,0.00022834103,0.000033873155,0.000017829125,0.000020969414,0.0002375671,0.0000920911,0.0002993519,0.9960007,0.0015888653],"study_design_scores_gemma":[0.00018171803,0.0000064677156,0.016028168,0.00030383066,0.00003724974,0.000029523231,0.00017323259,0.00053298316,0.00043554077,0.0010991478,0.98112696,0.00004519435],"about_ca_topic_score_codex":0.47366697,"about_ca_topic_score_gemma":0.7028732,"teacher_disagreement_score":0.52633303,"about_ca_system_score_codex":0.0049407296,"about_ca_system_score_gemma":0.010833323,"threshold_uncertainty_score":0.94181985},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Pollen; Point (geometry); Raw data; Vegetation (pathology)","score_opus":0.046059271998525025,"score_gpt":0.35426566239592994,"score_spread":0.30820639039740494,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963483783","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0010923382,0.00011657553,0.00014103529,0.000041431886,0.000034094854,0.000016942151,0.9969325,0.00076958904,0.0008553889],"genre_scores_gemma":[0.00069786294,0.000028228107,0.00028081707,0.000014094073,0.000005306064,0.000031527117,0.9985455,0.00006170891,0.0003349028],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99935395,0.00007963615,0.000050700215,0.0001934558,0.00023631466,0.00008588479],"domain_scores_gemma":[0.9990792,0.00014890378,0.00007589857,0.00023584449,0.00034368853,0.00011647091],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006396591,0.002080734,0.0012636526,0.003116645,0.00071105687,0.001408238,0.002429202,0.001613489,0.022567065],"category_scores_gemma":[0.002401812,0.00044851264,0.0011290712,0.0045502847,0.0003293514,0.0008132709,0.0013542888,0.0016414743,0.037101854],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008393165,0.000038269198,0.001631191,0.00041733866,0.000054184762,0.00004847355,0.000029855115,0.0007495486,0.00035534543,0.00036017073,0.9918933,0.0043385616],"study_design_scores_gemma":[0.00036078406,0.000027942968,0.017966341,0.00018946912,0.000053159194,0.00016476944,0.00010542046,0.0022560535,0.00085890543,0.0010354884,0.9769403,0.00004148607],"about_ca_topic_score_codex":0.022004459,"about_ca_topic_score_gemma":0.046004876,"teacher_disagreement_score":0.9779955,"about_ca_system_score_codex":0.00083222415,"about_ca_system_score_gemma":0.0014920407,"threshold_uncertainty_score":0.07549435},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Hydrology (agriculture); JSON; Raw data","score_opus":0.030799825111113493,"score_gpt":0.27511161074332063,"score_spread":0.24431178563220715,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963501954","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007683532,0.000021205253,0.00005610177,0.00004033459,0.0000106598945,0.00001254704,0.99783784,0.0000684181,0.00118447],"genre_scores_gemma":[0.0017536634,0.000042202108,0.00022791192,0.00003801025,0.0000028943862,0.00008171446,0.9955521,0.00004241478,0.0022590929],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9995988,0.000036465368,0.000044140477,0.000108335626,0.00012124038,0.000091119924],"domain_scores_gemma":[0.9979067,0.00016411858,0.000099131415,0.0003103447,0.0013478266,0.00017195116],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00044569292,0.0006847121,0.00056698755,0.002030335,0.00066451635,0.0011789724,0.001134846,0.0005378674,0.03952868],"category_scores_gemma":[0.0025508325,0.00036812155,0.00044395868,0.006565142,0.00028023048,0.00065351155,0.0012231283,0.00084185257,0.0348895],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059014437,0.0000142038,0.0073256376,0.00020452605,0.000052767984,0.000041273866,0.000051741143,0.00036876884,0.0002664667,0.000623062,0.9868838,0.0041087167],"study_design_scores_gemma":[0.00011748118,0.000006620462,0.06569979,0.0003259957,0.00004004937,0.000041705443,0.00052826863,0.0007583773,0.0007268923,0.0012098824,0.9304942,0.000050857878],"about_ca_topic_score_codex":0.53263015,"about_ca_topic_score_gemma":0.8038205,"teacher_disagreement_score":0.46736985,"about_ca_system_score_codex":0.0036621075,"about_ca_system_score_gemma":0.010213667,"threshold_uncertainty_score":0.94024456},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Raw data; Hydrology (agriculture); Biogeosciences","score_opus":0.03471035648026747,"score_gpt":0.30895113188960266,"score_spread":0.27424077540933517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963527136","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005062549,0.000046785346,0.00007247486,0.000050851457,0.000014700787,0.0000182904,0.9981647,0.00018907741,0.0009369545],"genre_scores_gemma":[0.000926324,0.000025419151,0.00022810593,0.000021721386,0.000005758971,0.00006812957,0.9977575,0.000057966463,0.00090908323],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992893,0.00006318048,0.00004327617,0.00019246707,0.00025850514,0.00015330229],"domain_scores_gemma":[0.99787915,0.00017014053,0.00013586188,0.00032193132,0.001250294,0.00024255778],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00061365316,0.0011905999,0.00088976754,0.0033908356,0.0010863639,0.0014531694,0.0023415913,0.00080457376,0.046898276],"category_scores_gemma":[0.004214411,0.00044391042,0.00073803955,0.0068423585,0.00041018345,0.00078254304,0.0015716741,0.0011954221,0.038346652],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003135326,0.000010673637,0.002288653,0.000118919765,0.00002616175,0.000015170761,0.00002228263,0.0001581944,0.00011834989,0.0002712686,0.9944283,0.0025106792],"study_design_scores_gemma":[0.00013120707,0.000010630554,0.043321416,0.00017270331,0.00003613611,0.000035046305,0.00015154008,0.00079563766,0.00036184685,0.0006265159,0.95430887,0.000048354443],"about_ca_topic_score_codex":0.5348114,"about_ca_topic_score_gemma":0.7618529,"teacher_disagreement_score":0.46518862,"about_ca_system_score_codex":0.0039464845,"about_ca_system_score_gemma":0.0064628986,"threshold_uncertainty_score":0.9358564},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","score_opus":0.026029152017918732,"score_gpt":0.2734806788983457,"score_spread":0.247451526880427,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963627274","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0002803246,0.00003896283,0.000045537527,0.00004886046,0.000011245978,0.0000108171735,0.9986889,0.00008061699,0.00079482544],"genre_scores_gemma":[0.000848032,0.00005027323,0.00017611375,0.000027288674,0.000005971287,0.000055853576,0.9976417,0.00003147705,0.0011634155],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992336,0.000049292063,0.0000659388,0.00020548176,0.0003152113,0.00013047893],"domain_scores_gemma":[0.9976936,0.00013743427,0.00018717066,0.00027666835,0.001455923,0.00024919165],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005845805,0.0010677563,0.0011237791,0.002646074,0.0011989265,0.0014621476,0.0019260993,0.00080813206,0.03433149],"category_scores_gemma":[0.0031573393,0.00046071436,0.0007083811,0.008449857,0.00041061302,0.0007444736,0.0014093629,0.00092145463,0.029997647],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004705511,0.000011981517,0.0029964338,0.0002901721,0.000038599723,0.00001969211,0.00003202072,0.00016100175,0.000108424225,0.0003568031,0.9938036,0.0021341832],"study_design_scores_gemma":[0.0001510318,0.0000080771715,0.029622756,0.00016975311,0.00004261917,0.000029355488,0.0001309031,0.00036921512,0.000284229,0.0005746254,0.96858305,0.000034285364],"about_ca_topic_score_codex":0.70040846,"about_ca_topic_score_gemma":0.8667265,"teacher_disagreement_score":0.29959154,"about_ca_system_score_codex":0.007197079,"about_ca_system_score_gemma":0.011733199,"threshold_uncertainty_score":0.6027118},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Hydrography; Pollen; Sample (material); Georgian; Service (business)","score_opus":0.03618938238550142,"score_gpt":0.28586680674619386,"score_spread":0.24967742436069243,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963651761","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0001200726,0.000021469847,0.000018100127,0.00002145781,0.000007344277,0.0000063042785,0.99931204,0.000068336005,0.00042489468],"genre_scores_gemma":[0.00031277156,0.000023551362,0.00012552098,0.000016832699,0.0000020155087,0.000024962672,0.9990472,0.00002786748,0.00041915174],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990601,0.000060971444,0.000081344544,0.00023499985,0.0003718743,0.00019081646],"domain_scores_gemma":[0.9966377,0.00025614558,0.00017312662,0.00047087815,0.002072101,0.00038997913],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007889836,0.001269766,0.0010354858,0.0045775175,0.0017251907,0.001977263,0.002767913,0.0010627513,0.040213976],"category_scores_gemma":[0.0052008573,0.0005841124,0.0007771149,0.012323226,0.0004887296,0.0008110084,0.001550376,0.001529393,0.039423756],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000023285951,0.000007835416,0.0013599017,0.00018296648,0.00001880915,0.000012314972,0.000022857805,0.0001274844,0.00006794793,0.00027372674,0.9966738,0.0012291089],"study_design_scores_gemma":[0.00013600147,0.0000068574413,0.026327718,0.00026458804,0.00003334783,0.000033891563,0.00020389157,0.00036451934,0.00025311054,0.00060438074,0.971731,0.00004062968],"about_ca_topic_score_codex":0.7970244,"about_ca_topic_score_gemma":0.9102489,"teacher_disagreement_score":0.20297557,"about_ca_system_score_codex":0.0063628564,"about_ca_system_score_gemma":0.016711134,"threshold_uncertainty_score":0.40834183},"labels":[],"label_agreement":null},{"id":"W6963685958","doi":"10.21233/vk9c-qp66","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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Sample (material); Hydrology (agriculture); JSON; Raw data","score_opus":0.03754192152436427,"score_gpt":0.3141483203499923,"score_spread":0.27660639882562804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963685958","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005655217,0.000040749514,0.00003661985,0.000031932632,0.000011510963,0.000012643479,0.9983734,0.00012769521,0.00079984],"genre_scores_gemma":[0.0011447048,0.000032717726,0.00016896943,0.000014954358,0.0000029599312,0.000054403292,0.9972594,0.000043881264,0.0012779838],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992293,0.00006462653,0.00007538607,0.00021265657,0.00024751143,0.00017052011],"domain_scores_gemma":[0.99712425,0.00022806205,0.00019097871,0.0005431529,0.0016082503,0.00030522706],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00045941758,0.0010454167,0.0009037047,0.0030137815,0.0011807813,0.0015620928,0.0019134863,0.0008421864,0.036104698],"category_scores_gemma":[0.0031696216,0.00046246185,0.0007342072,0.0062596546,0.00052360346,0.00082192494,0.0015042673,0.0009988672,0.03847881],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006814971,0.000015052708,0.004550625,0.00026455955,0.000042041032,0.000026408097,0.00005551719,0.00029481176,0.00018889738,0.00036159644,0.9911869,0.002945383],"study_design_scores_gemma":[0.000072156676,0.000011116174,0.04417942,0.00022527394,0.000049457667,0.00005298908,0.00035900634,0.00055122055,0.0004924117,0.0004451135,0.95349807,0.00006380685],"about_ca_topic_score_codex":0.61048114,"about_ca_topic_score_gemma":0.85035205,"teacher_disagreement_score":0.38951886,"about_ca_system_score_codex":0.0049952907,"about_ca_system_score_gemma":0.0076946896,"threshold_uncertainty_score":0.7836256},"labels":[],"label_agreement":null},{"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Paleoecology; Bay; Biogeosciences; Hydrology (agriculture); Paleolimnology; Water chemistry","score_opus":0.027190837573171856,"score_gpt":0.269282397587167,"score_spread":0.24209156001399512,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963807274","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003240473,0.00003565872,0.000041888805,0.000046869503,0.000010991291,0.0000105774425,0.99861264,0.00008253436,0.00083486026],"genre_scores_gemma":[0.00085305626,0.000044678563,0.00016249678,0.000024632607,0.000005307945,0.000049920884,0.9976484,0.000030447363,0.0011810611],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99928063,0.000044698256,0.000060274062,0.0001915502,0.00029954108,0.00012321475],"domain_scores_gemma":[0.998073,0.00011395457,0.00015498266,0.00023636292,0.001200714,0.00022087195],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00051760837,0.0010574638,0.0010418067,0.0024209726,0.0012011251,0.0013743443,0.0018654873,0.00078162487,0.034098182],"category_scores_gemma":[0.0027471178,0.00045830786,0.0007168794,0.007567824,0.00040571572,0.0007084976,0.0014147055,0.0008997263,0.029234316],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053559037,0.00001371571,0.0035136484,0.00031202377,0.000043187938,0.000021773743,0.00003877901,0.00018903517,0.00014184353,0.00040077962,0.9929283,0.0023432346],"study_design_scores_gemma":[0.00014288818,0.000008150383,0.031708922,0.00015019004,0.000042087122,0.000029462944,0.00013723559,0.00038416428,0.0003140392,0.00059745024,0.966452,0.000033544155],"about_ca_topic_score_codex":0.6894192,"about_ca_topic_score_gemma":0.8638693,"teacher_disagreement_score":0.3105808,"about_ca_system_score_codex":0.006914089,"about_ca_system_score_gemma":0.011825456,"threshold_uncertainty_score":0.6248197},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Natural Resources Canada","funders":"","keywords":"Pollen; Sample (material); Surface (topology); Raw data","score_opus":0.03245494995724853,"score_gpt":0.3113474159590036,"score_spread":0.2788924660017551,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963860407","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00026299126,0.000023357166,0.00006625862,0.000018654326,0.000010356905,0.000005969147,0.9989386,0.00019299275,0.00048083396],"genre_scores_gemma":[0.000514121,0.00001947781,0.0002187337,0.000012996295,0.0000031996044,0.000030648946,0.9987255,0.000057674257,0.00041765312],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99942315,0.00007328703,0.00005965289,0.00014876453,0.0002189842,0.00007617179],"domain_scores_gemma":[0.9985045,0.00027438754,0.00015617948,0.00035485026,0.0005852562,0.00012478294],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007437203,0.0013228869,0.00096895674,0.0025663967,0.0005197932,0.0015174369,0.0016361239,0.0013117109,0.03164386],"category_scores_gemma":[0.0033398375,0.0005085866,0.0009434578,0.0049867607,0.00029729304,0.0007909664,0.0010631962,0.0012670972,0.05638451],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000645411,0.000026563557,0.0017462415,0.00041797257,0.000046574452,0.000027899176,0.000023847551,0.0006300793,0.00027058957,0.00036271356,0.99356294,0.0028200722],"study_design_scores_gemma":[0.00034814334,0.000022542255,0.013272013,0.00016214227,0.000042619366,0.000050054074,0.00009069416,0.0011210534,0.00067868864,0.0009830924,0.98319614,0.00003264513],"about_ca_topic_score_codex":0.023387354,"about_ca_topic_score_gemma":0.045888454,"teacher_disagreement_score":0.03164386,"about_ca_system_score_codex":0.00087209744,"about_ca_system_score_gemma":0.0017468334,"threshold_uncertainty_score":0.10585928},"labels":[],"label_agreement":null}]}