{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":31,"total_is_capped":false,"direct_labels_cover":4,"predictions_cover":31,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"a02b92407db7","filters":{"venue":"Data Science Journal"}},"results":[{"id":"W3048186973","doi":"10.5334/dsj-2020-043","title":"The CARE Principles for Indigenous Data Governance","year":2020,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":1206,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Laurentian University","funders":"Rural Development Administration; European Commission","keywords":"Indigenous; Sovereignty; Data governance; Data sharing; Corporate governance; Stewardship (theology); Political science; Public administration; Public relations; Sociology; Law; Business; Medicine; Data quality","authors":[{"name":"Stephanie Russo Carroll","is_ca":false},{"name":"Ibrahim Garba","is_ca":false},{"name":"Óscar Luis Figueroa Rodríguez","is_ca":false},{"name":"Jarita Holbrook","is_ca":false},{"name":"Raymond Lovett","is_ca":false},{"name":"S. A. Materechera","is_ca":false},{"name":"M. A. Parsons","is_ca":false},{"name":"Kay Raseroka","is_ca":false},{"name":"Desi Rodriguez-Lonebear","is_ca":false},{"name":"Robyn Rowe","is_ca":true},{"name":"Rodrigo Sara","is_ca":false},{"name":"Jennifer Walker","is_ca":true},{"name":"Jane Anderson","is_ca":false},{"name":"Māui Hudson","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.3437787918939087,"gpt":0.4254753768765994,"spread":0.08169658498269072,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.05587465,0.0006365927,0.001001057,0.001943982,0.01213862,0.01750332,0.003103263,0.01144437,0.004959771],"category_scores_gemma":[0.08514049,0.0009209531,0.001606744,0.002162196,0.05598981,0.0163346,0.01426311,0.01620205,0.002010472],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.009121288,"about_ca_system_score_gemma":0.03268656,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02035278,"about_ca_topic_score_gemma":0.01580968,"domain_scores_codex":[0.9263034,0.03422971,0.006044667,0.008649219,0.0195547,0.005218301],"domain_scores_gemma":[0.9402927,0.03096393,0.002714178,0.01202404,0.01163825,0.002366907],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000003254591,0.000007987849,0.0001109894,0.00003877791,0.000003869913,0.00003533555,0.004605488,0.00008951381,0.00004789036,0.984363,0.006181141,0.004512767],"study_design_scores_gemma":[0.00001530483,0.00001733183,0.0002965238,0.0003907275,0.00001608252,0.0001789624,0.001939463,0.0004068105,0.0003005191,0.6414441,0.3549539,0.00004029688],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"other","genre_gemma":"methods","genre_scores_codex":[0.0131406,0.004716162,0.3167199,0.2355733,0.003842917,0.00111544,0.0004670209,0.0004498579,0.423975],"genre_scores_gemma":[0.5500107,0.00512363,0.2138239,0.08621753,0.003742975,0.003482447,0.0006747054,0.0006280178,0.1362961],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9968967,"threshold_uncertainty_score":0.2954969,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2036014579","doi":"10.2481/dsj.5.178","title":"Geography Markup Language","year":2006,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Geographic Information Systems Studies","field":"Social Sciences","cited_by":105,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Galdos Systems (Canada)","funders":"","keywords":"Markup language; XML; Geospatial analysis; Geomatics; Computer science; Document type definition; World Wide Web; Interoperability; Information retrieval; Geography; Document Structure Description; Remote sensing","authors":[{"name":"David Burggraf","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02436389931910291,"gpt":0.3337672918235531,"spread":0.3094033925044501,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002352192,0.0008217377,0.0006352651,0.003889889,0.001416384,0.004749239,0.002649446,0.001291073,0.09463055],"category_scores_gemma":[0.01003222,0.0004417314,0.0005514633,0.004963578,0.001081535,0.005352992,0.003693113,0.00178948,0.07719017],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001330654,"about_ca_system_score_gemma":0.002953808,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004846141,"about_ca_topic_score_gemma":0.003361129,"domain_scores_codex":[0.9975782,0.0006349872,0.0004178857,0.000346301,0.0008540907,0.0001685779],"domain_scores_gemma":[0.9955724,0.001201364,0.0003208611,0.001229454,0.001429846,0.000245918],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00009971897,0.00006787583,0.00102125,0.0005879956,0.00002096056,0.0004972304,0.001698612,0.001586931,0.002565013,0.2875538,0.5316368,0.1726639],"study_design_scores_gemma":[0.000005517233,0.000005545333,0.0001609622,0.00006170307,0.000003964557,0.0001808285,0.0001083208,0.0003374364,0.0005495433,0.01223751,0.9863343,0.00001452028],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.005304781,0.002187631,0.2965349,0.008381772,0.002163524,0.001461105,0.08105385,0.03589705,0.5670153],"genre_scores_gemma":[0.06493755,0.005029398,0.3821552,0.006716901,0.0009130063,0.002365183,0.179839,0.01329479,0.3447489],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.09463055,"threshold_uncertainty_score":0.3165707,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2908272686","doi":"10.5334/dsj-2019-001","title":"Understanding Human Mobility Patterns in a Developing Country Using Mobile Phone Data","year":2019,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Human Mobility and Location-Based Analysis","field":"Social Sciences","cited_by":47,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"TRIPS architecture; Mobile phone; Phone; Computer science; Destinations; Centroid; Entropy (arrow of time); Statistics; Geography; Mathematics; Telecommunications","authors":[{"name":"Merkebe Getachew Demissie","is_ca":true},{"name":"Santi Phithakkitnukoon","is_ca":false},{"name":"Lina Kattan","is_ca":true},{"name":"Ali Farhan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.3301334090330965,"gpt":0.4317011090950576,"spread":0.1015677000619611,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003274026,0.0002878335,0.0002570508,0.002065933,0.0002715841,0.0007188908,0.0002436427,0.0002829134,0.0005542535],"category_scores_gemma":[0.002107563,0.0001537341,0.0003089469,0.003501452,0.0002399812,0.0008188784,0.0005507016,0.0003419852,0.0002438891],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004339497,"about_ca_system_score_gemma":0.0003914974,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03618468,"about_ca_topic_score_gemma":0.03896552,"domain_scores_codex":[0.9997957,0.00006920911,0.00002050889,0.00005353286,0.00002997045,0.00003107788],"domain_scores_gemma":[0.9993669,0.0003049817,0.0001286738,0.00007469594,0.00008900053,0.00003576292],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00009762185,0.00009675542,0.9055762,0.0001227811,0.0001424259,0.0006312665,0.0009757191,0.03376721,0.002357573,0.001904626,0.001224224,0.05310364],"study_design_scores_gemma":[0.000009187259,0.0001144301,0.8486183,0.00007422709,0.00008989269,0.0004366706,0.005079732,0.1373098,0.001280597,0.002186795,0.004761817,0.00003849239],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9835955,0.0001772316,0.009527948,0.0001919846,0.000006304042,0.00003581026,0.004796187,0.00004233143,0.001626706],"genre_scores_gemma":[0.9909603,0.0002729563,0.00587584,0.00001981504,0.000005504312,0.00002737359,0.002637381,0.000003974871,0.0001966813],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.03618468,"threshold_uncertainty_score":0.07194811,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2125590838","doi":"10.2481/dsj.4.106","title":"The long-term preservation of accurate and authentic digital data: the INTERPARES project","year":2005,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Digital and Traditional Archives Management","field":"Arts and Humanities","cited_by":46,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Scope (computer science); Computer science; Usability; Transparency (behavior); Implementation; Metadata; Digital preservation; Open data; Data science; Reuse; World Wide Web; Data publishing; Electronic publishing; Publishing; Open science; Data curation; Political science; Software engineering; Engineering; The Internet; Computer security","authors":[{"name":"Luciana Duranti","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1383611772689184,"gpt":0.3153712717204733,"spread":0.1770100944515549,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.06982692,0.0006299266,0.0005880071,0.003152549,0.005809201,0.01492321,0.004568054,0.003499574,0.006327204],"category_scores_gemma":[0.08304152,0.0005338699,0.0007219851,0.005604105,0.01534222,0.02622312,0.03255798,0.004356535,0.002476266],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003160061,"about_ca_system_score_gemma":0.01730598,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004108653,"about_ca_topic_score_gemma":0.002075919,"domain_scores_codex":[0.9617088,0.01709342,0.00181435,0.002683629,0.01510446,0.001595383],"domain_scores_gemma":[0.8547962,0.03842978,0.006405311,0.07489886,0.01899135,0.006478575],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002130494,0.0002272409,0.002834624,0.0004418601,0.00004772075,0.0003069922,0.008089432,0.001639616,0.001338333,0.6049085,0.04427912,0.3356736],"study_design_scores_gemma":[0.0000581473,0.0002860236,0.003387054,0.0008596717,0.00006078693,0.00107708,0.01240981,0.00635315,0.01110808,0.2279108,0.736396,0.00009337153],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.11134,0.01259107,0.4766323,0.07940268,0.002625917,0.001170479,0.002761638,0.006886721,0.3065891],"genre_scores_gemma":[0.4569836,0.009956756,0.4270383,0.004315638,0.001473346,0.0009781152,0.006742221,0.00331329,0.08919878],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9850768,"threshold_uncertainty_score":0.3692845,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2891686592","doi":"10.5334/dsj-2018-020","title":"Using Classified and Unclassified Land Cover Data to Estimate the Footprint of Human Settlement","year":2018,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Impact of Light on Environment and Health","field":"Environmental Science","cited_by":45,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Esri (Canada)","funders":"","keywords":"Geography; Population; Footprint; Land cover; Cartography; Settlement (finance); Orthophoto; Ancillary data; Georeference; Census; Remote sensing; Land use; Computer science; Physical geography; Archaeology; Ecology; World Wide Web; Demography","authors":[{"name":"Charlie Frye","is_ca":true},{"name":"E. A. Nordstrand","is_ca":true},{"name":"Dawn J. Wright","is_ca":true},{"name":"Carmelle Terborgh","is_ca":true},{"name":"Jeanne Foust","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2387426298426108,"gpt":0.4369638167012191,"spread":0.1982211868586084,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000993353,0.0007241688,0.0004582712,0.006529471,0.0004195073,0.001714612,0.0006832057,0.0004578173,0.004073131],"category_scores_gemma":[0.004426804,0.0002496977,0.00079023,0.007787972,0.0003290937,0.001279093,0.0009626477,0.0003663275,0.002594963],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001236083,"about_ca_system_score_gemma":0.0008966416,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04552886,"about_ca_topic_score_gemma":0.07940274,"domain_scores_codex":[0.9990262,0.0001852077,0.00006891646,0.0001683911,0.0004689227,0.00008238621],"domain_scores_gemma":[0.9982635,0.0003836732,0.0003373714,0.0003356274,0.0005959752,0.00008388689],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0001527748,0.0003890194,0.4531622,0.0004246672,0.0005081232,0.0003506819,0.001265865,0.08538615,0.00343159,0.007926283,0.02330695,0.4236957],"study_design_scores_gemma":[0.00005523742,0.0001492098,0.6148163,0.0004267483,0.0001412699,0.0005330163,0.00382152,0.2914033,0.00522926,0.01632239,0.06693761,0.000164134],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6065686,0.001112739,0.2105041,0.0005544308,0.0002506842,0.0009081913,0.0931197,0.00279209,0.08418944],"genre_scores_gemma":[0.7846709,0.0006527941,0.153462,0.0001393501,0.00006264909,0.0005759603,0.05471718,0.0002939409,0.005425246],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04552886,"threshold_uncertainty_score":0.09052771,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1978333903","doi":"10.2481/dsj.1.45","title":"A Mexican case study on a centralised database from world natural history museums","year":2002,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":33,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Distribution (mathematics); Geography; Ornithology; Natural history; Species richness; Endemism; Database; Collections management; Library science; Ethnology; Ecology; Archaeology; History; Computer science; Biology; Southern Hemisphere","authors":[{"name":"Adolfo G. Navarro‐Sigüenza","is_ca":false},{"name":"A. Townsend Peterson","is_ca":false},{"name":"Alejandro Gordillo‐Martínez","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1465141139966224,"gpt":0.3052122749726442,"spread":0.1586981609760218,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006922219,0.0002503731,0.0003630652,0.003965226,0.00221088,0.002472747,0.001376063,0.00101117,0.00415249],"category_scores_gemma":[0.02030538,0.0003505138,0.0003472352,0.01107062,0.0008242017,0.002496745,0.001895661,0.000695776,0.0004643288],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003297231,"about_ca_system_score_gemma":0.002407501,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.08720831,"about_ca_topic_score_gemma":0.1779962,"domain_scores_codex":[0.9947995,0.002302106,0.0005350628,0.0006262479,0.001326,0.0004111337],"domain_scores_gemma":[0.9828722,0.007948549,0.002478328,0.002761707,0.0033792,0.000560018],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0005921125,0.0008143357,0.8394696,0.0009254605,0.0001426627,0.009956315,0.02962001,0.001449011,0.001893776,0.005357296,0.01422022,0.09555928],"study_design_scores_gemma":[0.0001311671,0.0003690942,0.8521852,0.0006360211,0.0001881241,0.006864709,0.0547823,0.006406948,0.001857611,0.0007663254,0.07572561,0.00008691803],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.977797,0.0007697146,0.001744424,0.001153172,0.00001818289,0.0004822142,0.008052336,0.00006978461,0.009913264],"genre_scores_gemma":[0.9782681,0.0007534347,0.008134049,0.0001767724,0.00002658258,0.0004304789,0.009985401,0.00003977478,0.002185347],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.08720831,"threshold_uncertainty_score":0.1734014,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2133007173","doi":"10.2481/dsj.4.127","title":"Multi-sensor data fusion for land vehicle attitude estimation using a fuzzy expert system","year":2005,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Inertial Sensor and Navigation","field":"Engineering","cited_by":31,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Gyroscope; Accelerometer; Kalman filter; Sensor fusion; Compass; Computer science; Inertial measurement unit; Fuzzy logic; Dead reckoning; Vibrating structure gyroscope; Artificial intelligence; Position (finance); Field (mathematics); Inertial navigation system; Global Positioning System; Computer vision; Control theory (sociology); Real-time computing; Inertial frame of reference; Engineering; Mathematics; Geography; Telecommunications; Aerospace engineering","authors":[{"name":"Jau-Hsiung Wang","is_ca":true},{"name":"Yang Gao","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1043321370756325,"gpt":0.3515910503201103,"spread":0.2472589132444778,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005313078,0.0002696709,0.0004091786,0.0002936205,0.00031921,0.0003993479,0.0003760149,0.0005474477,0.0008697403],"category_scores_gemma":[0.001118199,0.0001459989,0.000328966,0.0001831469,0.0001714196,0.0004749899,0.0002704673,0.0003580634,0.0002171111],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003297077,"about_ca_system_score_gemma":0.0004038693,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003534306,"about_ca_topic_score_gemma":0.005396499,"domain_scores_codex":[0.9998112,0.00003863182,0.00001952416,0.00004033343,0.00007368004,0.00001659357],"domain_scores_gemma":[0.9997095,0.00009940357,0.00002458483,0.00002405034,0.0001303598,0.0000119906],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004730909,0.0002822831,0.00256619,0.0002050954,0.0001371056,0.0003131059,0.0003039647,0.2607127,0.09140826,0.003778806,0.001896807,0.6379225],"study_design_scores_gemma":[0.00002448701,0.0001097805,0.00098484,0.00000850573,0.00003066053,0.00004863798,0.00002323383,0.9870422,0.009745273,0.001083529,0.0008868171,0.00001211144],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.06330845,0.0002012013,0.9343819,0.0001113967,0.00003492998,0.00006028853,0.00003364361,0.000451655,0.001416604],"genre_scores_gemma":[0.7812287,0.0001225167,0.2171819,0.00008032825,0.0000255424,0.00008053421,0.00005369536,0.0000102696,0.001216508],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003534306,"threshold_uncertainty_score":0.007027507,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2585472482","doi":"10.5334/dsj-2017-003","title":"Legal and Ethical Issues around Incorporating Traditional Knowledge in Polar Data Infrastructures","year":2017,"lang":"en","type":"article","venue":"Data Science Journal","topic":"International Maritime Law Issues","field":"Environmental Science","cited_by":21,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University; University of Ottawa","funders":"","keywords":"Traditional knowledge; Acknowledgement; Knowledge management; Interoperability; Context (archaeology); Sociology of scientific knowledge; Knowledge sharing; Inclusion (mineral); Body of knowledge; Computer science; Engineering ethics; Indigenous; Sociology; Social science; World Wide Web; Engineering; Computer security; Geography","authors":[{"name":"Teresa Scassa","is_ca":true},{"name":"Fraser Taylor","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07870823258626229,"gpt":0.3662956398552014,"spread":0.2875874072689391,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.1351551,0.0004725307,0.001019191,0.002625891,0.01637431,0.02798202,0.005717576,0.01750294,0.002687],"category_scores_gemma":[0.1623022,0.001131367,0.001228649,0.003921771,0.09489617,0.03418604,0.02087659,0.01973644,0.0009816034],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01159756,"about_ca_system_score_gemma":0.02612342,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01165454,"about_ca_topic_score_gemma":0.01093738,"domain_scores_codex":[0.818396,0.1171107,0.01205617,0.009376273,0.0359281,0.007132874],"domain_scores_gemma":[0.7388613,0.1862122,0.012583,0.03739481,0.02056195,0.004386812],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001258481,0.00002123228,0.0006847049,0.00007019992,0.000007974475,0.0003445234,0.03241635,0.0002781151,0.0002502765,0.9556655,0.002490772,0.007757797],"study_design_scores_gemma":[0.00001945485,0.00003205749,0.0009921301,0.00137085,0.00002302431,0.0006907563,0.03506592,0.0009419386,0.001023364,0.8084858,0.1512728,0.00008178957],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"commentary","genre_gemma":"other","genre_scores_codex":[0.07668497,0.005392122,0.1896438,0.3902367,0.001778976,0.0007778162,0.0002245063,0.0002035613,0.3350575],"genre_scores_gemma":[0.8532903,0.002919268,0.07999864,0.04237929,0.0010462,0.001117711,0.0001655571,0.0001882916,0.01889469],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.9942824,"threshold_uncertainty_score":0.7147769,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2141093030","doi":"10.2481/dsj.3.202","title":"Appraising digital records for long-term preservation","year":2004,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Digital and Traditional Archives Management","field":"Arts and Humanities","cited_by":18,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Scope (computer science); Computer science; Digital preservation; Usability; Transparency (behavior); Metadata; Implementation; Open data; World Wide Web; Data science; Reuse; Data curation; Data publishing; Electronic publishing; Publishing; Political science; Software engineering; The Internet; Computer security; Engineering","authors":[{"name":"Terry Eastwood","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1352581856858978,"gpt":0.3013021351824907,"spread":0.1660439494965929,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.1137528,0.0004744061,0.000787747,0.01847992,0.005636349,0.0369664,0.003496414,0.003820792,0.01700765],"category_scores_gemma":[0.4708805,0.0004558018,0.0006725704,0.0290136,0.009744972,0.03905672,0.0108176,0.003350127,0.003787803],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.008637528,"about_ca_system_score_gemma":0.0163177,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006021678,"about_ca_topic_score_gemma":0.007641024,"domain_scores_codex":[0.8960173,0.04637995,0.01275654,0.002073284,0.04062507,0.002147928],"domain_scores_gemma":[0.4002926,0.2848288,0.03540327,0.07445246,0.1969945,0.008028423],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0006033457,0.0001274144,0.02484963,0.004037304,0.0001565838,0.0005220695,0.03175721,0.0009380378,0.0006311468,0.3657289,0.09898904,0.4716594],"study_design_scores_gemma":[0.00008067887,0.000386479,0.02100213,0.01330335,0.0002918669,0.0006851436,0.07437645,0.002041463,0.002422626,0.1024148,0.7828389,0.0001560247],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1248956,0.06472272,0.06938496,0.2443825,0.01378694,0.002310936,0.004905284,0.00121832,0.4743927],"genre_scores_gemma":[0.8398021,0.02943483,0.06699903,0.009179274,0.005463096,0.0008016879,0.003849968,0.0008363868,0.04363348],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1137528,"threshold_uncertainty_score":0.6015896,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2123356220","doi":"10.2481/dsj.5.174","title":"The Crystallographic Information File (CIF)","year":2006,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University; Brockhouse Institute for Materials Research","funders":"","keywords":"Computer science; Scope (computer science); Transparency (behavior); Usability; Open data; Metadata; World Wide Web; Data science; Reuse; Data publishing; Data curation; Implementation; Software; Publishing; Political science; Software engineering; Computer security","authors":[{"name":"I. D. Brown","is_ca":true},{"name":"Brian T. McMahon","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01438017795474202,"gpt":0.2244732268565835,"spread":0.2100930489018415,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.008213235,0.007152257,0.02777502,0.01032586,0.009478618,0.01014454,0.02089088,0.009696944,0.884328],"category_scores_gemma":[0.02125653,0.005102168,0.00449828,0.03730476,0.003007604,0.01231079,0.006646013,0.01231604,0.5554547],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.009308064,"about_ca_system_score_gemma":0.0193391,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02201495,"about_ca_topic_score_gemma":0.02517699,"domain_scores_codex":[0.993912,0.0009705888,0.0005158784,0.0010342,0.002386665,0.001180777],"domain_scores_gemma":[0.9896823,0.003693382,0.001103751,0.001388112,0.003093311,0.00103927],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001487891,0.0000786793,0.0001723295,0.002712935,0.00005517149,0.00007440735,0.00008504299,0.000515536,0.000609506,0.006282578,0.9798123,0.009452823],"study_design_scores_gemma":[0.002951345,0.0002267114,0.001571309,0.001207082,0.000214522,0.000434373,0.0004254331,0.002800321,0.00136217,0.01487084,0.9735723,0.0003636031],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"methods","genre_scores_codex":[0.0009184635,0.003206887,0.006812198,0.003420752,0.007034515,0.001479539,0.8959472,0.006926744,0.07425359],"genre_scores_gemma":[0.007614367,0.00526221,0.01638372,0.001566553,0.001716612,0.004091658,0.9338439,0.003106746,0.02641427],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.884328,"threshold_uncertainty_score":0.164992,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3127642832","doi":"10.5334/dsj-2021-007","title":"Stewardship Maturity Assessment Tools for Modernization of Climate Data Management","year":2021,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Climate variability and models","field":"Environmental Science","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Environment and Climate Change Canada","funders":"Environment and Climate Change Canada; Centrum fÖr Personcentrerad Vård; National Centers for Environmental Information; National Oceanic and Atmospheric Administration; Grains Research and Development Corporation; National Aeronautics and Space Administration","keywords":"Stewardship (theology); Data management; Maturity (psychological); Scope (computer science); Data quality; Computer science; Process (computing); Quality (philosophy); Usability; Process management; Environmental resource management; Business; Data science; Database; Environmental science; Political science","authors":[{"name":"Robert Dunn","is_ca":false},{"name":"Christina Lief","is_ca":false},{"name":"Ge Peng","is_ca":false},{"name":"William E. Wright","is_ca":false},{"name":"Omar Baddour","is_ca":false},{"name":"Markus G. Donat","is_ca":false},{"name":"Brigitte Dubuisson","is_ca":false},{"name":"Jean‐François Legeais","is_ca":false},{"name":"P. Siegmund","is_ca":false},{"name":"Reinaldo Silveira","is_ca":false},{"name":"Xiaolan L. Wang","is_ca":true},{"name":"Markus Ziese","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.132977886381932,"gpt":0.3695392086253232,"spread":0.2365613222433913,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.06702811,0.001818128,0.0008809016,0.01949688,0.00256715,0.01178093,0.002727371,0.001504677,0.008000716],"category_scores_gemma":[0.1933127,0.00108657,0.00231012,0.01139637,0.001570618,0.01648218,0.009145082,0.004929029,0.002882995],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006069065,"about_ca_system_score_gemma":0.0122671,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006120121,"about_ca_topic_score_gemma":0.008141815,"domain_scores_codex":[0.9556321,0.01534082,0.009628084,0.002915379,0.01508084,0.001402759],"domain_scores_gemma":[0.7991746,0.0875411,0.02238633,0.02474849,0.06061643,0.005533036],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001805907,0.0008248533,0.03693908,0.002072768,0.0002312234,0.0002507309,0.01360639,0.01057737,0.004264905,0.1437232,0.0655937,0.7217353],"study_design_scores_gemma":[0.0001655622,0.0006451487,0.04811697,0.005358081,0.0002559234,0.0008074783,0.01770723,0.119172,0.01590688,0.2615226,0.5296992,0.0006429623],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03421797,0.0009922758,0.8816442,0.00739603,0.0003181824,0.004818649,0.005332114,0.01605628,0.04922415],"genre_scores_gemma":[0.07501145,0.0004777446,0.9125686,0.0003141311,0.00005650412,0.00229364,0.005752932,0.0007836084,0.002741354],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.06702811,"threshold_uncertainty_score":0.3544827,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3111376437","doi":"10.5334/dsj-2020-047","title":"39 Hints to Facilitate the Use of Semantics for Data on Agriculture and Nutrition","year":2020,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Agriculture and Agri-Food Canada; Rural Development Administration; Agence Nationale de la Recherche; Ministry of Agriculture of the People's Republic of China; Department for International Development; Bill and Melinda Gates Foundation","keywords":"Computer science; Semantic interoperability; Interoperability; Conceptualization; Linked data; Knowledge management; Data science; Semantics (computer science); World Wide Web; Data sharing; Standardization; Semantic Web","authors":[{"name":"Caterina Caracciolo","is_ca":false},{"name":"Sophie Aubin","is_ca":false},{"name":"Clément Jonquet","is_ca":false},{"name":"Emna Amdouni","is_ca":false},{"name":"Leyla Jael Castro","is_ca":false},{"name":"Brandon Whitehead","is_ca":false},{"name":"Catherine Roussey","is_ca":false},{"name":"Armando Stellato","is_ca":false},{"name":"Ferdinando Villa","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2965619348551438,"gpt":0.3499252635567615,"spread":0.0533633287016177,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.08452573,0.001227408,0.00116062,0.008987781,0.002543629,0.01083357,0.003214331,0.004687839,0.004940648],"category_scores_gemma":[0.1299845,0.001790141,0.00271647,0.00771523,0.005826201,0.02421181,0.01462578,0.005302946,0.002974814],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004124928,"about_ca_system_score_gemma":0.009599124,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008073497,"about_ca_topic_score_gemma":0.01184388,"domain_scores_codex":[0.9369188,0.03146348,0.01499577,0.003260403,0.01197868,0.001382834],"domain_scores_gemma":[0.8023643,0.1217816,0.007157207,0.04042386,0.02545759,0.002815383],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0002755122,0.0001721869,0.008395811,0.002604551,0.0001054278,0.001401869,0.02379814,0.00456765,0.01600407,0.7778568,0.02572139,0.1390966],"study_design_scores_gemma":[0.00003892743,0.00009079972,0.002409453,0.002167059,0.000075187,0.0009278917,0.005150912,0.007037683,0.0091742,0.207338,0.7654444,0.0001454324],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02093625,0.001281852,0.9029949,0.03007964,0.0003474382,0.001706153,0.005110662,0.005539419,0.03200371],"genre_scores_gemma":[0.05460829,0.000724679,0.9319012,0.00246995,0.00007450722,0.0006977443,0.006009987,0.0009165869,0.002597124],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.08452573,"threshold_uncertainty_score":0.4470202,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4301396488","doi":"10.5334/dsj-2022-017","title":"A Survey on Publicly Available Open Datasets Derived From Electronic Health Records (EHRs) of Patients with Neuroblastoma","year":2022,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Neuroblastoma Research and Treatments","field":"Medicine","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Neuroblastoma; Computer science; Health records; Open science; License; The Internet; Electronic health record; Open data; World Wide Web; Internet privacy; Medicine; Information retrieval; Data mining; Political science; Mathematics; Statistics; Health care","authors":[{"name":"Davide Chicco","is_ca":true},{"name":"Gabriel Cerono","is_ca":false},{"name":"Davide Cangelosi","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07305559528067233,"gpt":0.3504970068041415,"spread":0.2774414115234691,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.01021409,0.0006292441,0.0009444067,0.008501466,0.000736537,0.002241986,0.001486131,0.001364635,0.003123239],"category_scores_gemma":[0.06579199,0.0003146983,0.001036238,0.01566376,0.000608556,0.003039402,0.003758419,0.0009651149,0.001658581],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00127497,"about_ca_system_score_gemma":0.003022275,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006003798,"about_ca_topic_score_gemma":0.008588018,"domain_scores_codex":[0.9803949,0.005603066,0.004827285,0.002252272,0.006014243,0.0009082492],"domain_scores_gemma":[0.9168954,0.04896139,0.01248089,0.006739492,0.01279014,0.002132711],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"observational","study_design_scores_codex":[0.0009726149,0.0002032,0.3698104,0.0102693,0.0007011017,0.0004307862,0.001907235,0.001310113,0.001968987,0.006593823,0.4383466,0.1674857],"study_design_scores_gemma":[0.00008956951,0.000116908,0.4225538,0.004722445,0.0003544003,0.001405545,0.003630559,0.001526093,0.00203295,0.003519495,0.5598889,0.0001593129],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.102447,0.01776408,0.005774887,0.01040243,0.0005151989,0.0003713639,0.8513907,0.0009353296,0.01039908],"genre_scores_gemma":[0.08552543,0.008559856,0.007349293,0.003592142,0.0002725078,0.0008958807,0.8925357,0.0002217718,0.001047476],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9985139,"threshold_uncertainty_score":0.0540179,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2154198926","doi":"10.2481/dsj.14-047","title":"Data-PE: A Framework for Evaluating Data Publication Policies at Scholarly Journals","year":2015,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Scope (computer science); Usability; Transparency (behavior); Metadata; Open data; Data science; Data publishing; Implementation; Reuse; Publishing; World Wide Web; Political science; Software engineering; Engineering","authors":[{"name":"Nathan Moles","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.7646680752783171,"gpt":0.5850843913237225,"spread":0.1795836839545947,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3122008,0.002756205,0.004003277,0.0652151,0.007161181,0.03232554,0.006991662,0.005364609,0.01639744],"category_scores_gemma":[0.5277312,0.002130366,0.004241704,0.05426259,0.007134519,0.04492564,0.01646706,0.005208471,0.00659668],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01066171,"about_ca_system_score_gemma":0.03589238,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01615812,"about_ca_topic_score_gemma":0.01154823,"domain_scores_codex":[0.7016701,0.162797,0.0487055,0.01601574,0.0633764,0.007435235],"domain_scores_gemma":[0.3911301,0.3929101,0.04601748,0.06088814,0.09610736,0.01294675],"domain_codex":null,"domain_gemma":"reporting","domain_candidate":"reporting","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.001231938,0.0006810657,0.05936128,0.003358299,0.0008921472,0.0002898036,0.00653403,0.0105298,0.001209098,0.5849521,0.05929523,0.2716652],"study_design_scores_gemma":[0.0005211585,0.0008523887,0.02368814,0.004059455,0.0006170687,0.0005680083,0.01148707,0.08725765,0.007459357,0.5048236,0.3579787,0.0006873733],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02591836,0.001571449,0.8178897,0.006866645,0.0008274465,0.0109307,0.03187674,0.01211628,0.09200258],"genre_scores_gemma":[0.1148661,0.0006672394,0.850264,0.0008029277,0.0003272104,0.008598509,0.01833673,0.001202478,0.004934924],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9930083,"threshold_uncertainty_score":0.8481789,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2032869846","doi":"10.2481/dsj.6.od26","title":"Canadian National Consultation on Access to Scientific Research Data","year":2007,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":true},"ca_institutions":"National Research Council Canada; Université de Montréal","funders":"National Research Council Canada; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Strong","keywords":"Scope (computer science); Computer science; Transparency (behavior); Usability; Metadata; Data science; Open data; Reuse; Implementation; World Wide Web; Engineering; Computer security; Software engineering","authors":[{"name":"Michel Sabourin","is_ca":true},{"name":"Bernard Dumouchel","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.6443846780180462,"gpt":0.5804227013872811,"spread":0.06396197663076508,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.1102502,0.001654889,0.003440385,0.01338107,0.01695284,0.01550049,0.0121676,0.02343648,0.04619005],"category_scores_gemma":[0.1963779,0.002313375,0.004270209,0.02635349,0.006532154,0.006143025,0.01236174,0.01841101,0.009258871],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.1641705,"about_ca_system_score_gemma":0.6882334,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.9719106,"about_ca_topic_score_gemma":0.9758616,"domain_scores_codex":[0.8626081,0.01592002,0.01251241,0.006105145,0.08654446,0.01630987],"domain_scores_gemma":[0.6435397,0.08876326,0.004989986,0.01669901,0.2063033,0.03970471],"domain_codex":null,"domain_gemma":"reproducibility","domain_candidate":"reproducibility","domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"qualitative","study_design_scores_codex":[0.00003205033,0.000008492242,0.0002555374,0.0003519864,0.00001740085,0.00007688293,0.0003907171,0.00005952139,0.00007239095,0.01277825,0.9792861,0.006670626],"study_design_scores_gemma":[0.00002711701,0.000004768072,0.002195414,0.0006693461,0.00002687572,0.00003997199,0.0003371645,0.00006756962,0.00006691783,0.001522089,0.9949964,0.00004633212],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.001028106,0.02302877,0.0018116,0.7502089,0.01955619,0.001037114,0.037136,0.0007994415,0.1653939],"genre_scores_gemma":[0.0193063,0.03059763,0.02172632,0.5477103,0.003638262,0.002780676,0.03007909,0.001028865,0.3431326],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9878324,"threshold_uncertainty_score":0.9694437,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1985018337","doi":"10.2481/dsj.4.21","title":"Rescuing and recovering lost or endangered data","year":2005,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Digital and Cyber Forensics","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dominion Astrophysical Observatory; Herzberg Institute of Astrophysics","funders":"","keywords":"Scope (computer science); Computer science; Transparency (behavior); Open data; Metadata; Usability; Reuse; Data science; Implementation; World Wide Web; Data curation; Open science; Engineering; Software engineering; Computer security","authors":[{"name":"Elizabeth Griffin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1006452456693089,"gpt":0.3114684138730922,"spread":0.2108231682037833,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01566988,0.0009272688,0.001143813,0.006391687,0.005175385,0.007431432,0.005667239,0.005158108,0.01068028],"category_scores_gemma":[0.1121215,0.0007494161,0.001316854,0.004248842,0.004186744,0.01086217,0.0124214,0.00515339,0.008839803],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001246685,"about_ca_system_score_gemma":0.008742338,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002447749,"about_ca_topic_score_gemma":0.002358032,"domain_scores_codex":[0.9834173,0.004969651,0.001467271,0.001599985,0.007325082,0.001220738],"domain_scores_gemma":[0.8813122,0.02402538,0.00888492,0.06468382,0.01784136,0.003252357],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0006518177,0.0003422494,0.01636652,0.001126403,0.000239412,0.003603081,0.00691415,0.005678807,0.009325089,0.1370278,0.1307702,0.6879544],"study_design_scores_gemma":[0.00009675656,0.0004760229,0.008501749,0.002903951,0.0003471412,0.008993546,0.009623928,0.03841799,0.0438513,0.2491796,0.637296,0.0003119826],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1187809,0.007384607,0.7044547,0.05369686,0.006823187,0.001348799,0.003190325,0.009874385,0.09444623],"genre_scores_gemma":[0.5578356,0.004805648,0.327034,0.00865272,0.001634204,0.0005491059,0.005086046,0.002606618,0.09179595],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01566988,"threshold_uncertainty_score":0.08287126,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4386830764","doi":"10.5334/dsj-2023-035","title":"Umbrella Data Management Plans to Integrate FAIR Data: Lessons From the ISIDORe and BY-COVID Consortia for Pandemic Preparedness","year":2023,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal; Institute for Research in Immunology and Cancer","funders":"Biotechnology and Biological Sciences Research Council","keywords":"Preparedness; Data management; Computer science; Multidisciplinary approach; Plan (archaeology); Download; Process (computing); Data management plan; Data science; Process management; Knowledge management; Engineering management; World Wide Web; Business; Engineering; Political science; Geography; Database","authors":[{"name":"Romain David","is_ca":false},{"name":"Audrey S. Richard","is_ca":false},{"name":"Claire Connellan","is_ca":false},{"name":"Katharina B. Lauer","is_ca":false},{"name":"Maria Luisa Chiusano","is_ca":false},{"name":"Carole Goble","is_ca":false},{"name":"M Houde","is_ca":true},{"name":"Isabel Kemmer","is_ca":false},{"name":"Antje Keppler","is_ca":false},{"name":"Philippe Lieutaud","is_ca":false},{"name":"Christian Ohmann","is_ca":false},{"name":"Maria Panagiotopoulou","is_ca":false},{"name":"Sara Raza Khan","is_ca":false},{"name":"Arina Rybina","is_ca":false},{"name":"Stian Soiland‐Reyes","is_ca":false},{"name":"Charlotte Wit","is_ca":false},{"name":"Rudolf Wittner","is_ca":false},{"name":"Rafael Andrade Buono","is_ca":false},{"name":"Sarah Marsh","is_ca":false},{"name":"Pauline Audergon","is_ca":false},{"name":"Dylan Bonfils","is_ca":false},{"name":"J.M. Carazo","is_ca":false},{"name":"Rémi N. Charrel","is_ca":false},{"name":"Frederik Coppens","is_ca":false},{"name":"Wolfgang Fecke","is_ca":false},{"name":"Claudia Filippone","is_ca":false},{"name":"Eva G. Álvarez","is_ca":false},{"name":"Sheraz Gul","is_ca":false},{"name":"Henning Hermjakob","is_ca":false},{"name":"Katja Herzog","is_ca":false},{"name":"Petr Holub","is_ca":false},{"name":"Łukasz Kozera","is_ca":false},{"name":"Allyson Lister","is_ca":false},{"name":"José López-Coronado","is_ca":false},{"name":"Bénédicte Madon","is_ca":false},{"name":"Kurt Majcen","is_ca":false},{"name":"William Martin","is_ca":false},{"name":"Wolfgang G. Müller","is_ca":false},{"name":"Elli Papadopoulou","is_ca":false},{"name":"Christine Prat","is_ca":false},{"name":"Paolo Romano","is_ca":false},{"name":"Susanna‐Assunta Sansone","is_ca":false},{"name":"Gary Saunders","is_ca":false},{"name":"Niklas Blomberg","is_ca":false},{"name":"Jonathan J. Ewbank","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.4303456883339171,"gpt":0.4883463625022655,"spread":0.05800067416834848,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.4855037,0.001642959,0.001808199,0.007506551,0.01512245,0.06058188,0.01693589,0.01564576,0.007075835],"category_scores_gemma":[0.3552156,0.002618849,0.004347359,0.01253561,0.03126967,0.06660323,0.03338017,0.0260896,0.003295065],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.02448364,"about_ca_system_score_gemma":0.1132921,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04089735,"about_ca_topic_score_gemma":0.02034111,"domain_scores_codex":[0.7652922,0.1247961,0.0264506,0.0149073,0.05857282,0.009981086],"domain_scores_gemma":[0.504484,0.1814317,0.01248042,0.1934862,0.08014648,0.02797123],"domain_codex":null,"domain_gemma":"reproducibility","domain_candidate":"reproducibility","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.00009111515,0.0002239305,0.002087382,0.0003856092,0.00006440939,0.0002819916,0.009332332,0.00297586,0.00071278,0.8765352,0.05242186,0.05488738],"study_design_scores_gemma":[0.0001042035,0.0001451129,0.001606031,0.001644755,0.0000431691,0.0003428887,0.00609586,0.00561215,0.001552364,0.3865601,0.5960684,0.0002248344],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0135935,0.004080885,0.5516711,0.3606712,0.00390926,0.002617746,0.001750743,0.003238442,0.05846701],"genre_scores_gemma":[0.05936563,0.001994594,0.8888759,0.02285866,0.001050652,0.002333171,0.004062475,0.001911734,0.01754719],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9830641,"threshold_uncertainty_score":0.6344656,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4402883940","doi":"10.5334/dsj-2024-046","title":"Knowledge Infrastructures Are Growing Up: The Case for Institutional (Data) Repositories 10 Years After the Holdren Memo","year":2024,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"University at Buffalo; University of Minnesota","keywords":"Computer science; Business; World Wide Web; Data science","authors":[{"name":"Mikala Narlock","is_ca":false},{"name":"Scout Calvert","is_ca":false},{"name":"Shawna Taylor","is_ca":false},{"name":"Rachel Priesman Marquez","is_ca":false},{"name":"Aundria Parkman","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1327752469239644,"gpt":0.407377605795062,"spread":0.2746023588710976,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.02938407,0.0003224639,0.000439766,0.00400451,0.0203,0.03812424,0.003721845,0.0102471,0.009593083],"category_scores_gemma":[0.05701653,0.0009032578,0.001020668,0.007293689,0.01967755,0.06938031,0.02044056,0.00915975,0.00137369],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01847537,"about_ca_system_score_gemma":0.02890712,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.06135979,"about_ca_topic_score_gemma":0.07739386,"domain_scores_codex":[0.9807259,0.008452728,0.0008858382,0.001991594,0.004401703,0.003542164],"domain_scores_gemma":[0.9321728,0.03406134,0.006428536,0.01044628,0.009386616,0.007504489],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001120831,0.0001006842,0.01692772,0.0001815749,0.00003415188,0.003193193,0.1057165,0.0005043814,0.0003972565,0.7415113,0.05372396,0.07759727],"study_design_scores_gemma":[0.00002874402,0.00006115268,0.008210943,0.0009725318,0.00003806185,0.001416986,0.1243814,0.001062231,0.0007994581,0.07780516,0.7851158,0.0001075577],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.2566548,0.008104803,0.009502914,0.5450062,0.001123191,0.000126778,0.0004955369,0.0004261533,0.1785595],"genre_scores_gemma":[0.9341164,0.005005063,0.009057254,0.01909101,0.0006283389,0.0001388862,0.0003512106,0.0004152845,0.03119655],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9618757,"threshold_uncertainty_score":0.1553997,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2785284360","doi":"10.5334/dsj-2018-001","title":"Science Metadata Management, Interoperability and Data Citations of the National Institute of Polar Research, Japan","year":2018,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Metadata; Data management plan; Interoperability; Directory; Stewardship (theology); Data center; Meta Data Services; Open science; Data management; Data mapping; Library science; Data as a service; Research center; World Wide Web; Metadata management; Computer science; Metadata repository; Database; Political science; Business; Service (business)","authors":[{"name":"Masaki Kanao","is_ca":false},{"name":"Masaki Okada","is_ca":false},{"name":"Julie Friddell","is_ca":false},{"name":"Akira Kadokura","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.478091744716791,"gpt":0.5185561253619065,"spread":0.04046438064511554,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.01740929,0.000469509,0.0007308021,0.0133419,0.004175257,0.009539172,0.001777875,0.0009464631,0.01149589],"category_scores_gemma":[0.03695893,0.000575735,0.0005729672,0.02582136,0.00178171,0.01116898,0.007209508,0.001330968,0.006401403],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005739104,"about_ca_system_score_gemma":0.02472704,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05982715,"about_ca_topic_score_gemma":0.0397801,"domain_scores_codex":[0.9886233,0.002570234,0.002641572,0.001473409,0.004186978,0.0005045775],"domain_scores_gemma":[0.9654808,0.003151226,0.00300763,0.009505793,0.01705086,0.001803694],"domain_codex":null,"domain_gemma":"reproducibility","domain_candidate":"reproducibility","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001394804,0.0001151488,0.02691664,0.0007879495,0.00009147153,0.0002142654,0.004551994,0.0009481848,0.0029122,0.1657662,0.2037809,0.5937756],"study_design_scores_gemma":[0.00002529622,0.00002881407,0.01621783,0.0003704272,0.000101244,0.0001328338,0.002134665,0.002678706,0.003340143,0.0313839,0.9435221,0.00006412128],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"methods","genre_scores_codex":[0.06762857,0.02603973,0.2507673,0.04778459,0.006219449,0.002427851,0.04939566,0.01041179,0.539325],"genre_scores_gemma":[0.3232479,0.02212172,0.3646546,0.002880717,0.00215165,0.002834029,0.1087932,0.00320697,0.1701092],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9904608,"threshold_uncertainty_score":0.1189578,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2059472642","doi":"10.2481/dsj.5.143","title":"A hashing technique using separate binary tree","year":2006,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science; Scope (computer science); Usability; Data science; Metadata; Transparency (behavior); Data publishing; Implementation; World Wide Web; Publishing; Software engineering; Computer security; Political science","authors":[{"name":"Mehedi Masud","is_ca":true},{"name":"Gopal Das","is_ca":false},{"name":"Anisur Rahman","is_ca":false},{"name":"Arunashis Ghose","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05802956779394889,"gpt":0.3299944953690478,"spread":0.2719649275750989,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009575899,0.0006172356,0.001193136,0.002315005,0.001241034,0.001396867,0.001834697,0.001262791,0.01270484],"category_scores_gemma":[0.003846593,0.00050746,0.0008589944,0.003310687,0.0008217053,0.004128381,0.003008406,0.0011441,0.008978047],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004346931,"about_ca_system_score_gemma":0.0009808681,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001017894,"about_ca_topic_score_gemma":0.001035648,"domain_scores_codex":[0.9984341,0.0001894846,0.0001510663,0.000327001,0.000734364,0.0001639809],"domain_scores_gemma":[0.9978063,0.0004368008,0.0001385729,0.0009493836,0.0005630222,0.0001060742],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0007375075,0.0001870171,0.001255398,0.0004272781,0.00008238549,0.0002190411,0.0003589464,0.008193961,0.05123429,0.0279694,0.01521738,0.8941174],"study_design_scores_gemma":[0.0004442316,0.002538812,0.005532608,0.0003447039,0.0003886396,0.007561904,0.0009428406,0.5053044,0.2267939,0.07807256,0.1716384,0.0004370761],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02263773,0.001843338,0.964668,0.0002318936,0.0008023017,0.0002385136,0.000349516,0.002592002,0.00663672],"genre_scores_gemma":[0.231056,0.001209539,0.7406788,0.0003651432,0.0003923819,0.0002867487,0.001811099,0.0004771073,0.02372326],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01270484,"threshold_uncertainty_score":0.04250193,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2021867077","doi":"10.2481/dsj.8.27","title":"CISTI'S Activities in Support of Scientific Data Management in Canada 2008-2010","year":2009,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"National Research Council Canada","funders":"","keywords":"Scope (computer science); Computer science; Transparency (behavior); Usability; Metadata; Implementation; Open data; Reuse; Data management; Data science; World Wide Web; Publishing; Data publishing; Political science; Database; Engineering; Software engineering; Computer security","authors":[{"name":"Mary Zborowski","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1662144302179339,"gpt":0.3827343700927204,"spread":0.2165199398747866,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.01750647,0.0009591047,0.0008994233,0.006863173,0.00767073,0.007392597,0.003481832,0.001331164,0.020765],"category_scores_gemma":[0.03915985,0.0006778871,0.0005786565,0.01813403,0.00291422,0.002846597,0.004792187,0.00247288,0.007944568],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.08695627,"about_ca_system_score_gemma":0.3716304,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.9673039,"about_ca_topic_score_gemma":0.9491015,"domain_scores_codex":[0.9847771,0.001257191,0.0006747678,0.001162505,0.01040085,0.001727667],"domain_scores_gemma":[0.9127129,0.004026918,0.002059551,0.002768007,0.0653463,0.01308642],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0004998741,0.0001616578,0.0162106,0.0007570591,0.00006699267,0.0006775447,0.004984661,0.001668803,0.005040487,0.04230788,0.6491129,0.2785115],"study_design_scores_gemma":[0.00003116539,0.00002763379,0.007612954,0.0001323817,0.00001180854,0.00009294243,0.001320962,0.001882201,0.00194149,0.0009758349,0.9859416,0.00002899367],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.0636505,0.02408808,0.05265517,0.2205657,0.01304146,0.004648612,0.06577227,0.01430489,0.5412734],"genre_scores_gemma":[0.2155928,0.02437572,0.1311391,0.01057671,0.00263935,0.001460121,0.06226817,0.002869708,0.5490783],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9965182,"threshold_uncertainty_score":0.6309148,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2031989156","doi":"10.2481/dsj.2.25","title":"Data integration and knowledge discovery in biomedical databases. Reliable information from unreliable sources","year":2003,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Artificial Intelligence in Healthcare","field":"Health Professions","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Scope (computer science); Computer science; Data science; Usability; Metadata; Transparency (behavior); Open data; Reuse; Implementation; World Wide Web; Data discovery; Data publishing; Publishing; Software engineering; Political science; Engineering","authors":[{"name":"Arnold Mitnitski","is_ca":true},{"name":"Alexander Mogilner","is_ca":true},{"name":"Chris MacKnight","is_ca":true},{"name":"Kenneth Rockwood","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2754938637313639,"gpt":0.5016500590355036,"spread":0.2261561953041397,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04343435,0.0009461007,0.003337821,0.0158275,0.002686394,0.01741557,0.003775336,0.003931697,0.004110878],"category_scores_gemma":[0.1114113,0.001948764,0.002696335,0.0268997,0.006749413,0.02651634,0.01253912,0.003672997,0.003175957],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002636592,"about_ca_system_score_gemma":0.006708412,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003334626,"about_ca_topic_score_gemma":0.002785309,"domain_scores_codex":[0.9604355,0.01842936,0.004587048,0.003094553,0.01279717,0.0006563522],"domain_scores_gemma":[0.9305956,0.0471747,0.004474113,0.01159568,0.00530367,0.0008560392],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002291806,0.0001108598,0.004448676,0.004272627,0.0005929797,0.0009035059,0.003104127,0.008177101,0.001313908,0.5193773,0.03352737,0.4239423],"study_design_scores_gemma":[0.00004389499,0.00004620248,0.001313227,0.001656992,0.0002440066,0.0008978816,0.0009914953,0.0338826,0.002417628,0.8319884,0.1264325,0.00008509575],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.006767988,0.0765029,0.8550107,0.0347089,0.001304398,0.000595469,0.002768955,0.001863523,0.02047709],"genre_scores_gemma":[0.1494963,0.04474839,0.7807166,0.005439686,0.002483658,0.0008797764,0.007352399,0.0005720817,0.008311111],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.04343435,"threshold_uncertainty_score":0.2297056,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2955083888","doi":"10.5334/dsj-2019-027","title":"Developing a Model Guidelines Addressing Legal Impediments to Open Access to Publicly Funded Research Data in Malaysia","year":2019,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Data access; Computer science; Business; Knowledge management; Database","authors":[{"name":"Haswira Nor Mohamad Hashim","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.8726138537959646,"gpt":0.6560074045530535,"spread":0.2166064492429111,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.04547009,0.0007873643,0.0008200341,0.006250395,0.003073116,0.01254213,0.003387281,0.004101526,0.003662149],"category_scores_gemma":[0.08172981,0.001025306,0.001518752,0.004417846,0.003486194,0.01091832,0.005953216,0.004273558,0.001023464],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01423107,"about_ca_system_score_gemma":0.0602402,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03721255,"about_ca_topic_score_gemma":0.04956111,"domain_scores_codex":[0.9555452,0.0197691,0.006082439,0.002594065,0.0132849,0.002724301],"domain_scores_gemma":[0.9080643,0.04199493,0.01068667,0.005824239,0.03066777,0.002762173],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00008924114,0.0004767223,0.01549487,0.001191053,0.00006202026,0.0009010655,0.007178247,0.03676515,0.001179866,0.8386798,0.01784686,0.08013511],"study_design_scores_gemma":[0.0002494274,0.0005073796,0.01327477,0.005646003,0.0002317269,0.0009705171,0.02014584,0.1190502,0.006108728,0.5163679,0.3170168,0.0004307348],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1222643,0.002442684,0.5560185,0.07100996,0.0005218898,0.009929707,0.00358895,0.001174725,0.2330493],"genre_scores_gemma":[0.3761988,0.002455103,0.5954316,0.002745833,0.00007340434,0.006099239,0.002088667,0.0001236352,0.01478368],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9966127,"threshold_uncertainty_score":0.2404717,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2765696161","doi":"10.5334/dsj-2017-048","title":"The Northern Voice: Listening to Indigenous and Northern Perspectives on Management of Data in Canada","year":2017,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Indigenous Studies and Ecology","field":"Health Professions","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"University of Waterloo","funders":"","keywords":"Indigenous; Interoperability; Active listening; Class (philosophy); The Internet; Geography; World Wide Web; Computer science; Sociology; Ecology","authors":[{"name":"Dana Church","is_ca":true},{"name":"Julie Friddell","is_ca":true},{"name":"E. LeDrew","is_ca":true},{"name":"Gabrielle Alix","is_ca":true},{"name":"Garret Reid","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08398072553422269,"gpt":0.4148932033926937,"spread":0.330912477858471,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0136653,0.0008001674,0.0009757855,0.002320538,0.08171867,0.0233549,0.004374409,0.005318858,0.009804279],"category_scores_gemma":[0.0173352,0.0007760563,0.0007659493,0.005580664,0.02700537,0.007110958,0.01320817,0.00972347,0.001025702],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.1709126,"about_ca_system_score_gemma":0.2860342,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.993673,"about_ca_topic_score_gemma":0.9961344,"domain_scores_codex":[0.9809094,0.004634552,0.0004663675,0.001430739,0.005188025,0.007370921],"domain_scores_gemma":[0.9748281,0.007399664,0.000694151,0.0005072539,0.00660372,0.009967269],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.00005172187,0.00002335709,0.002997389,0.0001555383,0.00001872068,0.001176006,0.9043437,0.0001166491,0.0005041998,0.009531697,0.06583686,0.01524421],"study_design_scores_gemma":[0.000003503342,0.000006817526,0.00222258,0.000163167,0.000009175235,0.0001047277,0.8431591,0.00005820452,0.000084214,0.0007794496,0.1533634,0.00004566278],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.2716922,0.01943492,0.002666008,0.4480141,0.005915409,0.0002383232,0.001086983,0.0002723291,0.2506798],"genre_scores_gemma":[0.8486395,0.01054519,0.002180251,0.06001551,0.0005109308,0.00014025,0.0003502119,0.0003730834,0.07724506],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9956256,"threshold_uncertainty_score":0.9616238,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2083696640","doi":"10.2481/dsj.009-026","title":"Data Management Activities of Canada's National Science Library - 2010 Update and Prospective","year":2010,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"National Research Council Canada","funders":"","keywords":"Mandate; Political science; Transformational leadership; Parliament; Research council; Library science; Engineering management; Public relations; Management; Business; Engineering; Computer science; Government (linguistics)","authors":[{"name":"Mary Zborowski","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0732256112792575,"gpt":0.3498536390072697,"spread":0.2766280277280122,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.01975334,0.001109215,0.001195874,0.02162627,0.01205673,0.01761134,0.004884149,0.002048965,0.01806989],"category_scores_gemma":[0.03430227,0.001296034,0.0008349642,0.04639519,0.00282337,0.004980625,0.005064605,0.002582053,0.01045914],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.1585398,"about_ca_system_score_gemma":0.5169505,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.9901803,"about_ca_topic_score_gemma":0.9899283,"domain_scores_codex":[0.970715,0.001489619,0.002000112,0.001092601,0.02204766,0.002655011],"domain_scores_gemma":[0.861553,0.004218617,0.003356393,0.005933209,0.1112161,0.01372268],"domain_codex":null,"domain_gemma":"reproducibility","domain_candidate":"reproducibility","domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"observational","study_design_scores_codex":[0.00007149452,0.0001044044,0.009634683,0.0006449599,0.00002669151,0.00009591444,0.00162689,0.0002940461,0.0007151058,0.007921405,0.8346992,0.1441653],"study_design_scores_gemma":[0.00000499206,0.000006716583,0.008887796,0.0001615708,0.0000104424,0.00003522733,0.0008795902,0.0001470176,0.0005701429,0.0001514331,0.9891198,0.000025292],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.02035623,0.0406983,0.006484218,0.1267812,0.004640496,0.001927356,0.2723682,0.00754426,0.5191998],"genre_scores_gemma":[0.08134204,0.06074526,0.04540927,0.01250259,0.0009132732,0.000840023,0.3120472,0.002296504,0.4839039],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9823887,"threshold_uncertainty_score":0.9759746,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2610869427","doi":"10.5334/dsj-2017-024","title":"All or Nothing: The False Promise of Anonymity","year":2017,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Juvenile Diabetes Research Foundation","funders":"","keywords":"Anonymity; Data sharing; Computer science; Identification (biology); Data anonymization; Process (computing); Internet privacy; State (computer science); Computer security; Information privacy; Medicine","authors":[{"name":"Neil Walker","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.8656067754571136,"gpt":0.6914985035282208,"spread":0.1741082719288928,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.1079848,0.0006178404,0.001700838,0.001241437,0.01143477,0.02642872,0.004096669,0.02369952,0.00478692],"category_scores_gemma":[0.2052194,0.0008880303,0.001570673,0.001560637,0.09596907,0.03743938,0.01474111,0.036858,0.002712166],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006218098,"about_ca_system_score_gemma":0.01762907,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002301231,"about_ca_topic_score_gemma":0.001469483,"domain_scores_codex":[0.7982061,0.1396655,0.006279104,0.01061763,0.03867518,0.006556436],"domain_scores_gemma":[0.7424797,0.2008637,0.008596747,0.03116518,0.01138735,0.005507454],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00002931514,0.000008918996,0.0001569098,0.0000650198,0.00001377329,0.00009749985,0.003593053,0.0000990805,0.00005544682,0.9409947,0.04601479,0.008871526],"study_design_scores_gemma":[0.00003166134,0.00003467807,0.0001664594,0.0004189717,0.00001573698,0.0003052216,0.00178058,0.0002816788,0.0002244455,0.7969933,0.1996952,0.00005206818],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.003797431,0.004345627,0.01996389,0.8937511,0.006947522,0.00006069186,0.0001341755,0.00008827296,0.07091116],"genre_scores_gemma":[0.3405105,0.006759926,0.01539014,0.5916201,0.01840108,0.0006559378,0.0001760992,0.0002759733,0.02621025],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1079848,"threshold_uncertainty_score":0.5710847,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2132929844","doi":"10.2481/dsj.008-011","title":"Structured Query Translation in Peer to Peer Database Sharing Systems","year":2009,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Advanced Database Systems and Queries","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science; Scope (computer science); Usability; Transparency (behavior); Data sharing; World Wide Web; Metadata; Implementation; Open data; Reuse; Data science; Database; Software engineering","authors":[{"name":"Mehedi Masud","is_ca":true},{"name":"Fazel Keshtkar","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07014835883326626,"gpt":0.3407390784558296,"spread":0.2705907196225634,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03210694,0.0009328728,0.002548773,0.002439138,0.004113552,0.008625732,0.00564177,0.004660389,0.009147139],"category_scores_gemma":[0.076874,0.00157015,0.001596522,0.005797531,0.005284288,0.0148388,0.01336157,0.002277393,0.0055611],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002043445,"about_ca_system_score_gemma":0.005086763,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004350523,"about_ca_topic_score_gemma":0.002507634,"domain_scores_codex":[0.9578718,0.02095555,0.00442705,0.00302987,0.01172461,0.001991146],"domain_scores_gemma":[0.924545,0.04162369,0.002044756,0.02111477,0.009082134,0.001589549],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.002275603,0.0009550753,0.004058785,0.001700533,0.0003815055,0.002492342,0.01088378,0.1057265,0.01249821,0.4743846,0.04516045,0.3394825],"study_design_scores_gemma":[0.0005292504,0.0003986256,0.000526113,0.00018855,0.0001011794,0.0006262182,0.002478104,0.3987891,0.01194997,0.5389469,0.04527861,0.0001874587],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04084652,0.001281501,0.9320475,0.002557968,0.0004146836,0.001867778,0.0008395397,0.007727283,0.0124172],"genre_scores_gemma":[0.4809901,0.00107666,0.4948366,0.001194443,0.0004633707,0.001608273,0.003073304,0.001505622,0.0152516],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.03210694,"threshold_uncertainty_score":0.1697998,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4382141301","doi":"10.5334/dsj-2023-018","title":"Polar Data Forum IV – An Ocean of Opportunities","year":2023,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University; Inuvialuit Regional Corporation; Nunavut Research Institute; University of Calgary","funders":"LifeWatch – Niclas Öberg Foundation; Belgian Federal Science Policy Office","keywords":"Interoperability; Polar; Data management; Computer science; Library science; World Wide Web; Data science; Database","authors":[{"name":"Annemie R. Janssen","is_ca":false},{"name":"P. Bricher","is_ca":false},{"name":"Karen Payne","is_ca":false},{"name":"Renuka Badhe","is_ca":false},{"name":"Nicole Biebow","is_ca":false},{"name":"T. De Bruin","is_ca":false},{"name":"P. Elshout","is_ca":false},{"name":"A. G. Gaylord","is_ca":false},{"name":"Øystein Godøy","is_ca":false},{"name":"Patrick Gorringe","is_ca":false},{"name":"Damien Guihen","is_ca":false},{"name":"Johnathan Kool","is_ca":false},{"name":"Jan René Larsen","is_ca":false},{"name":"Joseph E. Nolan","is_ca":false},{"name":"Antonio Novellino","is_ca":false},{"name":"William Manley","is_ca":false},{"name":"Andreas Marouchos","is_ca":false},{"name":"Molly McCammon","is_ca":false},{"name":"M. S. Murray","is_ca":true},{"name":"Jenn Parrott","is_ca":true},{"name":"Jay Pearlman","is_ca":false},{"name":"Helen J. Peat","is_ca":false},{"name":"Peter Pulsifer","is_ca":true},{"name":"L.E.A. Symons","is_ca":false},{"name":"Marten Tacoma","is_ca":false},{"name":"Soulaine Theocharides","is_ca":false},{"name":"Stein Tronstad","is_ca":false},{"name":"Chantelle Verhey","is_ca":false},{"name":"Anton Van de Putte","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.5412294111931125,"gpt":0.4607770822650198,"spread":0.08045232892809268,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.09554195,0.0007804437,0.0008488545,0.004028848,0.01426747,0.03940298,0.002640809,0.01217998,0.03440339],"category_scores_gemma":[0.08819683,0.000963646,0.001118251,0.005423149,0.01048931,0.03404235,0.03671108,0.01603298,0.008753772],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007266753,"about_ca_system_score_gemma":0.02040277,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004687643,"about_ca_topic_score_gemma":0.004349437,"domain_scores_codex":[0.9544919,0.01973094,0.00141908,0.002927679,0.01449021,0.006940245],"domain_scores_gemma":[0.8848644,0.05135402,0.004603917,0.01239591,0.01243135,0.03435029],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001916108,0.0001245194,0.001768789,0.0004742582,0.00001537604,0.0006828904,0.0118623,0.0003232724,0.0009414383,0.4296494,0.4583956,0.09557054],"study_design_scores_gemma":[0.000007266848,0.0000147992,0.0002667602,0.0002125048,0.000001019146,0.00006493997,0.002530167,0.00003941829,0.0001067909,0.01369783,0.9830402,0.00001841135],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"other","genre_scores_codex":[0.01414086,0.02105004,0.02324745,0.6972238,0.05626727,0.0004488891,0.00182506,0.0008919664,0.1849047],"genre_scores_gemma":[0.2634888,0.028927,0.0447496,0.2331233,0.04112488,0.00228003,0.007875144,0.004471741,0.3739595],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.09554195,"threshold_uncertainty_score":0.5052801,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4401463404","doi":"10.5334/dsj-2024-042","title":"Decentralised Semantics: A Semantic Engine User Perspective","year":2024,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"Canada First Research Excellence Fund","keywords":"Computer science; Perspective (graphical); Semantics (computer science); Information retrieval; Semantic computing; World Wide Web; Semantic Web; Programming language; Artificial intelligence","authors":[{"name":"Carly Huitema","is_ca":false},{"name":"P. F. Knowles","is_ca":false},{"name":"Philippe Page","is_ca":false},{"name":"A. Michelle Edwards","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04095387098060591,"gpt":0.3302790571581251,"spread":0.2893251861775192,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03422497,0.000986331,0.001826379,0.0029559,0.00229131,0.01966344,0.004902651,0.005069066,0.007557066],"category_scores_gemma":[0.02826203,0.001602139,0.002687666,0.003190132,0.01211492,0.04015477,0.01183433,0.007336187,0.003367061],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004605157,"about_ca_system_score_gemma":0.005365734,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005795412,"about_ca_topic_score_gemma":0.004261301,"domain_scores_codex":[0.9846135,0.006330896,0.001765486,0.001914406,0.00430879,0.001066829],"domain_scores_gemma":[0.9778541,0.008614488,0.0006126818,0.00856657,0.003326021,0.001026155],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.00009627085,0.00004728219,0.0006321993,0.0001451063,0.00003563663,0.0002544081,0.005409517,0.002820316,0.001558357,0.9679331,0.00400163,0.01706604],"study_design_scores_gemma":[0.00005234917,0.00006624134,0.0002714605,0.0002676413,0.00006668923,0.0005744269,0.002421208,0.03283108,0.004960718,0.7103665,0.2480454,0.0000763663],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01350947,0.0006711522,0.9216243,0.007822931,0.0002513232,0.0001970295,0.0004220857,0.004768962,0.05073274],"genre_scores_gemma":[0.3833212,0.00146414,0.5828202,0.002890595,0.0004346492,0.0004136057,0.001368348,0.004661364,0.02262577],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.03422497,"threshold_uncertainty_score":0.1810011,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4380538161","doi":"10.5334/dsj-2023-015","title":"Legal Regulation of State Electronic Services: Relevant Issues and Ways of Improvement","year":2023,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Corruption and Economic Development","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Government (linguistics); Business; Politics; Public relations; Public sector; Quality (philosophy); State (computer science); Order (exchange); Service (business); Public administration; The Republic; Political science; Marketing; Computer science; Law; Finance","authors":[{"name":"Akzhan G. Duisenkul","is_ca":false},{"name":"Dzhamilya Ospanova","is_ca":false},{"name":"Gaziz D. Taigamitov","is_ca":false},{"name":"Saule M. Madykhan","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05167142070269107,"gpt":0.3291735194967818,"spread":0.2775020987940907,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.09442486,0.00130052,0.002017959,0.007858914,0.01039072,0.04180332,0.007742224,0.02223169,0.01053142],"category_scores_gemma":[0.1326965,0.001011536,0.00224581,0.0101629,0.04900141,0.05913773,0.009824604,0.02130171,0.001832673],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.02482846,"about_ca_system_score_gemma":0.08330019,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03403898,"about_ca_topic_score_gemma":0.02183337,"domain_scores_codex":[0.926603,0.04503669,0.004664769,0.004023843,0.01278834,0.006883353],"domain_scores_gemma":[0.8186083,0.1035303,0.01000006,0.01757874,0.04269275,0.007589684],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00003141161,0.0002515388,0.001985773,0.0008546576,0.00003591344,0.0002011609,0.006993507,0.00102915,0.0001763831,0.9056286,0.02390859,0.05890341],"study_design_scores_gemma":[0.00007211637,0.0001800798,0.004126939,0.007288152,0.00008435627,0.0003737759,0.04525118,0.002971815,0.000732157,0.5639488,0.3747337,0.0002368991],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.006539661,0.0455323,0.03184442,0.8510121,0.002708938,0.000466212,0.0001516782,0.0003594011,0.06138536],"genre_scores_gemma":[0.6041532,0.08333214,0.1349632,0.1408801,0.006248418,0.002051797,0.0004983486,0.0003582532,0.02751454],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.09442486,"threshold_uncertainty_score":0.4993724,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1446508819","doi":"","title":"The Canadian Enhanced Polar Outflow Probe (e-POP) Mission: Current Status and Planned Observations and Data Distribution:Current Status and Planned Observations and Data Distribution","year":2009,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Methane Hydrates and Related Phenomena","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Current (fluid); Outflow; Distribution (mathematics); Polar; Environmental science; Meteorology; Geography; Physics; Engineering; Electrical engineering; Mathematics","authors":[{"name":"A. W. Yau","is_ca":false},{"name":"H. G. James","is_ca":false},{"name":"P. A. Bernhardt","is_ca":false},{"name":"L. L. Cogger","is_ca":false},{"name":"Greg Enno","is_ca":false},{"name":"Hirofumi HAYAKAWA","is_ca":false},{"name":"E. P. King","is_ca":false},{"name":"D. J. Knudsen","is_ca":false},{"name":"Richard B. Langley","is_ca":false},{"name":"R. Rankin","is_ca":false},{"name":"R. H. Hum","is_ca":false},{"name":"D. D. Wallis","is_ca":false},{"name":"A. White","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08234495579074003,"gpt":0.3189444292911663,"spread":0.2365994735004263,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002782102,0.001170043,0.0007633232,0.003881898,0.00293969,0.001490959,0.002879068,0.0009693034,0.01013964],"category_scores_gemma":[0.002060664,0.0004597663,0.0004015029,0.006455013,0.000958647,0.001425783,0.001438057,0.001018684,0.00362211],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.009020347,"about_ca_system_score_gemma":0.03310571,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.8599607,"about_ca_topic_score_gemma":0.8964268,"domain_scores_codex":[0.9984816,0.00005700749,0.00003276398,0.0001388313,0.0009075619,0.0003823207],"domain_scores_gemma":[0.9946137,0.0002225131,0.0003066396,0.0003659985,0.003533788,0.000957387],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.004553875,0.0007478161,0.1317016,0.001720218,0.0003618937,0.0003029008,0.001021744,0.006820066,0.06847171,0.003511057,0.4077387,0.3730484],"study_design_scores_gemma":[0.0008067451,0.0004009506,0.3315531,0.0003321204,0.0002623868,0.0001238658,0.0007535378,0.0100061,0.0323543,0.001317532,0.6218203,0.0002690422],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"methods","genre_scores_codex":[0.180125,0.003922798,0.01984725,0.004012659,0.0005601712,0.003287516,0.6300001,0.0103694,0.1478751],"genre_scores_gemma":[0.3463162,0.004666182,0.07436365,0.001679577,0.0003128536,0.001979341,0.5274842,0.001577403,0.04162059],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8599607,"threshold_uncertainty_score":0.2817279,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}