{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":24,"total_is_capped":false,"direct_labels_cover":7,"predictions_cover":24,"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":"09653c74ed9a","filters":{"venue":"Journal of Informetrics"}},"results":[{"id":"W2191720934","doi":"10.1016/j.joi.2015.09.003","title":"Are top-cited papers more interdisciplinary?","year":2015,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Interdisciplinary Research and Collaboration","field":"Decision Sciences","cited_by":144,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"Humanities and Social Sciences Youth Foundation, Ministry of Education of the People's Republic of China; Fonds de Recherche du Québec-Société et Culture; National Office for Philosophy and Social Sciences; Ministry of Education of the People's Republic of China","keywords":"Citation; Bibliometrics; Citation impact; Rank (graph theory); Computer science; Focus (optics); Citation analysis; Percentile; Data science; Percentile rank; Library science; Information retrieval; Statistics; Mathematics","authors":[{"name":"Shiji Chen","is_ca":false},{"name":"Clément Arsenault","is_ca":true},{"name":"Vincent Larivière","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1625741757646505,"gpt":0.461230812010258,"spread":0.2986566362456075,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.04412194,0.001095891,0.003798116,0.0340602,0.004615027,0.03859451,0.003084013,0.006140408,0.03510064],"category_scores_gemma":[0.2865244,0.001039573,0.002279851,0.06463293,0.003376351,0.01876061,0.006949789,0.00323831,0.007043443],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002698384,"about_ca_system_score_gemma":0.003670086,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009325774,"about_ca_topic_score_gemma":0.001827303,"domain_scores_codex":[0.9340636,0.01706407,0.01237098,0.00937392,0.02274743,0.004380147],"domain_scores_gemma":[0.579525,0.2003029,0.09303771,0.02722623,0.06718995,0.0327182],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.002799093,0.0009395782,0.4881817,0.01148927,0.009849382,0.00155426,0.007998778,0.0008383203,0.004929084,0.02611699,0.05663872,0.3886648],"study_design_scores_gemma":[0.0009175303,0.0007570809,0.660415,0.005953767,0.005940383,0.003731455,0.02395787,0.001167495,0.003967683,0.09708612,0.1957124,0.0003931454],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4748386,0.2225298,0.01604164,0.1257103,0.02609414,0.0004510092,0.009209718,0.0007991634,0.1243258],"genre_scores_gemma":[0.9096039,0.03750307,0.006877793,0.009971501,0.01941473,0.0002970185,0.004139919,0.0004272281,0.01176488],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9659398,"threshold_uncertainty_score":0.2333419,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2129840456","doi":"10.1016/j.joi.2009.03.010","title":"Modeling a century of citation distributions","year":2009,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Complex Systems and Time Series Analysis","field":"Economics, Econometrics and Finance","cited_by":143,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Citation; Computer science; Function (biology); Simple (philosophy); Set (abstract data type); Mathematical economics; Econometrics; Mathematics; Epistemology; Library science; Philosophy","authors":[{"name":"Matthew L. Wallace","is_ca":true},{"name":"Vincent Larivière","is_ca":true},{"name":"Yves Gingras","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03318613078372981,"gpt":0.2246175135931572,"spread":0.1914313828094274,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.004713895,0.0004891062,0.0009685613,0.001953257,0.0008408341,0.004183922,0.002227148,0.003335601,0.003860443],"category_scores_gemma":[0.039685,0.0009284079,0.0008134763,0.002110213,0.002979392,0.009792574,0.002137594,0.003598968,0.0004682223],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002237394,"about_ca_system_score_gemma":0.001209915,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007700848,"about_ca_topic_score_gemma":0.006858651,"domain_scores_codex":[0.99847,0.0009229802,0.0000715208,0.0002448423,0.0001902113,0.0001004515],"domain_scores_gemma":[0.9814813,0.01429737,0.0009247342,0.001765968,0.000932797,0.0005977563],"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":[0.00006113195,0.00009979922,0.005341992,0.00007418973,0.0001254599,0.0001224334,0.0005884637,0.2866986,0.0001942732,0.659683,0.006349077,0.0406616],"study_design_scores_gemma":[0.000009976581,0.000008127442,0.0004875993,0.00003096166,0.00001535155,0.0000589192,0.00008510101,0.5196727,0.00007048108,0.474066,0.005477374,0.00001749996],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2071691,0.0141591,0.7197979,0.03519363,0.001162564,0.00004531632,0.001094123,0.0006501411,0.02072816],"genre_scores_gemma":[0.9208571,0.007428183,0.05809332,0.001089277,0.001207658,0.00009871828,0.0003620575,0.0002686452,0.01059507],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9980468,"threshold_uncertainty_score":0.02492976,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2908106500","doi":"10.1016/j.joi.2018.12.005","title":"Comparing journal and paper level classifications of science","year":2019,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":95,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal; McGill University","funders":"","keywords":"Computer science; Bibliometrics; Information retrieval; Robustness (evolution); Discipline; Data science; Data mining; Social science; Sociology","authors":[{"name":"Fei Shu","is_ca":false},{"name":"Charles‐Antoine Julien","is_ca":true},{"name":"Lin Zhang","is_ca":false},{"name":"Junping Qiu","is_ca":false},{"name":"Jing Zhang","is_ca":false},{"name":"Vincent Larivière","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.7615079244057587,"gpt":0.5814149810599263,"spread":0.1800929433458324,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.008995476,0.0003486038,0.0007097425,0.02938005,0.000947903,0.007389792,0.000780716,0.001199644,0.009577801],"category_scores_gemma":[0.08237549,0.0001839045,0.0008798636,0.02283659,0.001141633,0.003876116,0.002630001,0.0009221354,0.00187462],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001471844,"about_ca_system_score_gemma":0.001447757,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002813439,"about_ca_topic_score_gemma":0.003477289,"domain_scores_codex":[0.9910662,0.003542936,0.001180268,0.0007073335,0.002645825,0.000857416],"domain_scores_gemma":[0.8528819,0.09209782,0.02181768,0.007111813,0.02076471,0.005326133],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.001328902,0.0001775618,0.9411668,0.0003746686,0.000521725,0.00006331469,0.001737058,0.0005574893,0.002812274,0.008738606,0.003124365,0.03939745],"study_design_scores_gemma":[0.0001058152,0.0003353108,0.968101,0.0002241162,0.0004357786,0.00009912118,0.006453929,0.002772652,0.001488146,0.01383153,0.006106219,0.0000464763],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9689168,0.001378244,0.002834061,0.0006015692,0.0001705621,0.00007199525,0.004497651,0.00008423939,0.02144496],"genre_scores_gemma":[0.9907596,0.0003866412,0.002801701,0.0001011388,0.0001220191,0.00006881448,0.004482437,0.00004107529,0.001236584],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.97062,"threshold_uncertainty_score":0.04757315,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1221429143","doi":"10.1016/j.joi.2015.08.002","title":"How to become an important player in scientific collaboration networks?","year":2015,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":88,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Publication; Computer science; Productivity; Quality (philosophy); Position (finance); Work (physics); Data science; Knowledge management; Business; Engineering","authors":[{"name":"Ashkan Ebadi","is_ca":true},{"name":"Andrea Schiffauerova","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02861058399649574,"gpt":0.2999498346330988,"spread":0.271339250636603,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.01452403,0.0007058191,0.0007795714,0.002633189,0.007956987,0.01840775,0.002585858,0.008296563,0.02791222],"category_scores_gemma":[0.05559898,0.0006646373,0.0005913301,0.002243471,0.009345828,0.04867399,0.01303246,0.007153327,0.01039347],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002661714,"about_ca_system_score_gemma":0.007135647,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001622235,"about_ca_topic_score_gemma":0.002650933,"domain_scores_codex":[0.9909757,0.004483064,0.0002425073,0.001262519,0.001509433,0.001526834],"domain_scores_gemma":[0.9296727,0.01588762,0.004303812,0.003267101,0.00834432,0.0385245],"domain_codex":null,"domain_gemma":"incentives","domain_candidate":"incentives","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003438628,0.0004843895,0.01493069,0.0005824085,0.0001445918,0.0009465191,0.01017629,0.0004700596,0.001723263,0.5945572,0.1612603,0.2143804],"study_design_scores_gemma":[0.00007689372,0.0001421141,0.003713841,0.0004721209,0.00009185884,0.0008935208,0.02547785,0.001318468,0.0009277848,0.5629154,0.4038858,0.00008422683],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.1054515,0.0158316,0.02893413,0.6113318,0.009194486,0.0001829581,0.0003736644,0.0005777515,0.2281221],"genre_scores_gemma":[0.876436,0.007672679,0.01489071,0.0331641,0.00234825,0.0002638027,0.0002654438,0.0003591197,0.06459992],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9973668,"threshold_uncertainty_score":0.09337568,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2150897866","doi":"10.1016/j.joi.2011.06.002","title":"Comparing the expert survey and citation impact journal ranking methods: Example from the field of Artificial Intelligence","year":2011,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":81,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University; Lakehead University","funders":"","keywords":"Ranking (information retrieval); Journal ranking; Consistency (knowledge bases); Citation; Computer science; Information retrieval; Quality (philosophy); Field (mathematics); Data science; Publication; Artificial intelligence; World Wide Web; Mathematics; Advertising; Business","authors":[{"name":"Alexander Serenko","is_ca":true},{"name":"Michael S. Dohan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.8625684102688479,"gpt":0.6369410142134144,"spread":0.2256273960554335,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.04629992,0.0007961862,0.001447521,0.008720539,0.0008690643,0.002192702,0.001248617,0.002075299,0.003282637],"category_scores_gemma":[0.1667566,0.0001826083,0.001216166,0.00942368,0.0009681996,0.002577121,0.001432685,0.001072313,0.000677996],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001095291,"about_ca_system_score_gemma":0.001239122,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007647441,"about_ca_topic_score_gemma":0.01178769,"domain_scores_codex":[0.9582388,0.03064423,0.001427245,0.001280224,0.007908639,0.0005008976],"domain_scores_gemma":[0.5815275,0.379699,0.00438351,0.007923705,0.02560654,0.0008597302],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.005972251,0.003178933,0.3295632,0.003975405,0.00512503,0.0003572265,0.003422352,0.02492562,0.003170968,0.02188931,0.0147871,0.5836326],"study_design_scores_gemma":[0.001728479,0.009155566,0.598398,0.001507957,0.00534794,0.001628958,0.0113549,0.2540161,0.01235844,0.06413069,0.0397123,0.00066071],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8280435,0.02287525,0.1096682,0.004345784,0.0004233553,0.0005820994,0.002211497,0.000328966,0.03152132],"genre_scores_gemma":[0.9256704,0.003788102,0.06563744,0.0006158964,0.0002415109,0.0001731178,0.001298825,0.00007283297,0.00250169],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9912795,"threshold_uncertainty_score":0.2448603,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2967837635","doi":"10.1016/j.joi.2019.07.004","title":"Analyzing linguistic complexity and scientific impact","year":2019,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":74,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"Qinglan Project of Jiangsu Province of China; Major Program of National Fund of Philosophy and Social Science of China","keywords":"Computer science; Linguistics; Linguistic sequence complexity; Natural language processing; Data science; Philosophy","authors":[{"name":"Chao Lu","is_ca":false},{"name":"Yi Bu","is_ca":false},{"name":"Xianlei Dong","is_ca":false},{"name":"Jie Wang","is_ca":false},{"name":"Ying Ding","is_ca":false},{"name":"Vincent Larivière","is_ca":true},{"name":"Cassidy R. Sugimoto","is_ca":false},{"name":"Logan Paul","is_ca":false},{"name":"Chengzhi Zhang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02438825478114541,"gpt":0.3143265421653923,"spread":0.2899382873842469,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.008062338,0.0003974176,0.0007093158,0.0190471,0.001760548,0.007188539,0.0008551857,0.0009950238,0.005903327],"category_scores_gemma":[0.1176973,0.0004126641,0.0007490084,0.01245527,0.003630242,0.008554831,0.004021529,0.001945665,0.0005009656],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002354996,"about_ca_system_score_gemma":0.002025949,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004475081,"about_ca_topic_score_gemma":0.004596584,"domain_scores_codex":[0.9923825,0.002645095,0.0005596299,0.0005521025,0.003431752,0.0004288009],"domain_scores_gemma":[0.7888609,0.1719094,0.01567568,0.006105574,0.01380342,0.00364516],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.001654333,0.0008975413,0.6350241,0.0008065751,0.001038794,0.001053878,0.005890341,0.02886854,0.006414833,0.1197326,0.005208951,0.1934095],"study_design_scores_gemma":[0.0001505417,0.0004020917,0.4101758,0.0002544411,0.0007799708,0.0006374432,0.01131309,0.140659,0.004838499,0.41904,0.01151425,0.0002347041],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9539765,0.001635235,0.0147743,0.003132779,0.00009560063,0.00008867105,0.0009902067,0.0001432069,0.02516345],"genre_scores_gemma":[0.9947211,0.0003065239,0.003753889,0.00006731485,0.0001385776,0.00002867105,0.0003596662,0.0000345308,0.0005896469],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9809529,"threshold_uncertainty_score":0.04263818,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3115533035","doi":"10.1016/j.joi.2020.101124","title":"Exploring the interdisciplinarity patterns of highly cited papers","year":2020,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":56,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"National Office for Philosophy and Social Sciences","keywords":"Computer science; Data science; Bibliometrics; Information retrieval; World Wide Web","authors":[{"name":"Shiji Chen","is_ca":false},{"name":"Junping Qiu","is_ca":false},{"name":"Clément Arsenault","is_ca":true},{"name":"Vincent Larivière","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.7915538673463653,"gpt":0.540354756721656,"spread":0.2511991106247093,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.003612443,0.0002334315,0.0004420954,0.01789994,0.001063877,0.004853471,0.0008072689,0.0007593944,0.003908975],"category_scores_gemma":[0.04324429,0.0001771643,0.0005274108,0.02736738,0.0004844472,0.002572994,0.001660686,0.0006234801,0.00100397],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005151518,"about_ca_system_score_gemma":0.0007553541,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001878001,"about_ca_topic_score_gemma":0.003428087,"domain_scores_codex":[0.996595,0.0007385743,0.000638815,0.000616551,0.001176985,0.0002341535],"domain_scores_gemma":[0.9445987,0.03336598,0.009000393,0.002812237,0.008634342,0.001588321],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0002924378,0.0001545183,0.8815221,0.0005476995,0.0006582657,0.0005189207,0.004890432,0.0006721693,0.007615804,0.004430407,0.001729852,0.09696742],"study_design_scores_gemma":[0.00002821042,0.0001387784,0.9659827,0.0001260802,0.0003002876,0.0008349448,0.007134762,0.004870483,0.002953258,0.01007536,0.007510495,0.0000446983],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9879591,0.001689257,0.002203598,0.0003654894,0.00003675398,0.00002849552,0.00134053,0.00004962532,0.006327087],"genre_scores_gemma":[0.9953138,0.0005520018,0.001743992,0.00002764969,0.000063782,0.00001956932,0.001111283,0.00002406769,0.001143797],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9963875,"threshold_uncertainty_score":0.01910466,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2134386090","doi":"10.1016/j.joi.2010.04.001","title":"The development of an AI journal ranking based on the revealed preference approach","year":2010,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":56,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Lakehead University","funders":"","keywords":"Ranking (information retrieval); Journal ranking; Impact factor; Citation; Preference; Computer science; Index (typography); Information retrieval; Field (mathematics); Citation analysis; Domain (mathematical analysis); Library science; Data science; Operations research; Statistics; World Wide Web; Mathematics; Political science","authors":[{"name":"Alexander Serenko","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.578649466269294,"gpt":0.5307756849711357,"spread":0.04787378129815834,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.004283806,0.000547544,0.001162821,0.006422334,0.0009469999,0.003437522,0.001382657,0.0009782545,0.004150451],"category_scores_gemma":[0.01625585,0.0004183847,0.001141792,0.004766143,0.0004455532,0.002879516,0.001181947,0.001054356,0.001497503],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001167035,"about_ca_system_score_gemma":0.003742847,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005447386,"about_ca_topic_score_gemma":0.007666938,"domain_scores_codex":[0.9953284,0.001853013,0.000301074,0.0002996229,0.002002292,0.000215641],"domain_scores_gemma":[0.9906216,0.003632065,0.0003544143,0.0005809075,0.004514496,0.0002965357],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002287216,0.0005904216,0.008458835,0.0005699478,0.0003786969,0.0002412015,0.0003265437,0.05731479,0.008139842,0.07184172,0.007730769,0.8441784],"study_design_scores_gemma":[0.00006400125,0.0004542155,0.004202996,0.0000695565,0.000175362,0.0002979835,0.0002575161,0.942409,0.006769102,0.03548041,0.00971019,0.0001095883],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02828164,0.0003332624,0.9608253,0.0006227859,0.0001244145,0.0003367393,0.000309752,0.0007352454,0.008430893],"genre_scores_gemma":[0.2370347,0.0003751109,0.7583946,0.00008949042,0.00008378537,0.0002389652,0.0003604652,0.00006779842,0.003355118],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9957162,"threshold_uncertainty_score":0.02265519,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2781044009","doi":"10.1016/j.joi.2017.12.005","title":"On the relationships between bibliographic characteristics of scientific documents and citation and Mendeley readership counts: A large-scale analysis of Web of Science publications","year":2018,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":56,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Audience measurement; Citation; Citation analysis; Altmetrics; Discipline; Web of science; Set (abstract data type); Citation impact","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.4491389199918624,"gpt":0.4977714261719171,"spread":0.04863250618005466,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.004119085,0.0003984624,0.000816269,0.01418869,0.0009909152,0.002885436,0.0007423745,0.0007846872,0.002420498],"category_scores_gemma":[0.04427778,0.0002709471,0.001260747,0.0169783,0.0008806969,0.002933854,0.001554981,0.0008261875,0.0007799296],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005857043,"about_ca_system_score_gemma":0.0009073071,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005694335,"about_ca_topic_score_gemma":0.007813164,"domain_scores_codex":[0.9975377,0.0008902885,0.0003266971,0.0004752171,0.0006376266,0.0001325301],"domain_scores_gemma":[0.8900722,0.09421062,0.008805728,0.003069211,0.002638785,0.001203372],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0001484918,0.00008320886,0.9800943,0.00008833065,0.0006319938,0.0001269766,0.0004544997,0.001002458,0.001021229,0.0003519287,0.0002943549,0.0157023],"study_design_scores_gemma":[0.00001239331,0.0000539363,0.9885896,0.00001587351,0.0002562196,0.0002448299,0.0005224732,0.008622396,0.00052949,0.0006400804,0.000486543,0.00002621405],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9962829,0.00048937,0.001034971,0.0001206607,0.000007688135,0.00001023321,0.001101074,0.00004837828,0.0009047416],"genre_scores_gemma":[0.9969985,0.0001961925,0.001185238,0.00001156597,0.00003134449,0.00001339986,0.001254237,0.00001933677,0.0002902373],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9958809,"threshold_uncertainty_score":0.02178407,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2041260947","doi":"10.1016/j.joi.2010.07.005","title":"What's familiar is excellent: The impact of exposure effect on perceived journal quality","year":2010,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Knowledge Management and Sharing","field":"Social Sciences","cited_by":55,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University; Lakehead University","funders":"","keywords":"Ranking (information retrieval); Journal ranking; Preference; Psychology; Interpretation (philosophy); Quality (philosophy); Perception; Test (biology); Field (mathematics); Statistics; Computer science; Information retrieval; Library science; Epistemology; Mathematics; Citation; Philosophy","authors":[{"name":"Alexander Serenko","is_ca":true},{"name":"Nick Bontis","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03543923403875162,"gpt":0.3812089441243721,"spread":0.3457697100856205,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.006839896,0.0002239395,0.0007085969,0.0007243445,0.0008279076,0.003236275,0.0007437199,0.001427446,0.01522413],"category_scores_gemma":[0.06929068,0.000356598,0.0007832253,0.0007738649,0.001351983,0.001941792,0.002186452,0.001824881,0.0006349434],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005111471,"about_ca_system_score_gemma":0.0008327565,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00176994,"about_ca_topic_score_gemma":0.001533209,"domain_scores_codex":[0.9947039,0.002709848,0.0004208231,0.0006101856,0.001034133,0.0005210444],"domain_scores_gemma":[0.7868529,0.1832851,0.01077712,0.004034249,0.003114871,0.01193577],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.02387118,0.01665205,0.8768318,0.000452392,0.001309997,0.0007338569,0.01348407,0.001054558,0.01569709,0.001063407,0.0009403411,0.04790929],"study_design_scores_gemma":[0.0003387599,0.003410381,0.9875869,0.00004230566,0.0004878571,0.0001865709,0.004049275,0.0007749103,0.001725794,0.0006616188,0.0006785898,0.00005714765],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9976011,0.00008274544,0.0001955536,0.0002482175,0.00002042789,0.00001764794,0.0000532787,0.00001225794,0.001768883],"genre_scores_gemma":[0.9990172,0.00004410853,0.0002193132,0.0000757369,0.00002734997,0.00001507542,0.00004242167,0.00001193076,0.0005467043],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9992757,"threshold_uncertainty_score":0.05092978,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W423015428","doi":"10.1016/j.joi.2015.05.001","title":"Modelling count response variables in informetric studies: Comparison among count, linear, and lognormal regression models","year":2015,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":54,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Western University","funders":"Capital Medical University","keywords":"Count data; Statistics; Negative binomial distribution; Poisson regression; Akaike information criterion; Mathematics; Regression analysis; Linear regression; Poisson distribution; Overdispersion; Log-normal distribution; Generalized linear model; Binomial regression; Regression diagnostic; Econometrics; Polynomial regression; Population","authors":[{"name":"Isola Ajiferuke","is_ca":true},{"name":"Felix Famoye","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.103223804081119,"gpt":0.3570945740796192,"spread":0.2538707699985002,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.07055525,0.001770683,0.00243211,0.004490251,0.0008447333,0.005542858,0.005063356,0.003785598,0.005513296],"category_scores_gemma":[0.2788167,0.0008557212,0.003543701,0.006461505,0.002147052,0.008207985,0.002640764,0.003516645,0.001320098],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003072211,"about_ca_system_score_gemma":0.003096512,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009292494,"about_ca_topic_score_gemma":0.007711653,"domain_scores_codex":[0.9397513,0.05364316,0.001621268,0.002237685,0.002118354,0.0006283176],"domain_scores_gemma":[0.454904,0.5239224,0.01007757,0.005106936,0.005247912,0.0007411545],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.007497935,0.001512008,0.1572095,0.004940731,0.004181415,0.0008308645,0.005764341,0.4057179,0.001091361,0.1348407,0.004204678,0.2722085],"study_design_scores_gemma":[0.0003264513,0.001086855,0.01134864,0.0006959466,0.001186179,0.0003466572,0.00188746,0.9054806,0.0006569601,0.07434648,0.002496466,0.0001413796],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.27918,0.005025704,0.7071362,0.003002206,0.0003149204,0.000931716,0.001085617,0.0004438759,0.002879741],"genre_scores_gemma":[0.8191155,0.003813606,0.1694294,0.0005021491,0.0002705375,0.00168751,0.001081518,0.000221146,0.003878489],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9955097,"threshold_uncertainty_score":0.3731363,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2062524032","doi":"10.1016/j.joi.2013.04.005","title":"Web traffic and organization performance measures: Relationships and data sources examined","year":2013,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":52,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Western University","funders":"","keywords":"Webometrics; Web traffic; Hyperlink; Computer science; World Wide Web; The Internet; Web page","authors":[{"name":"Liwen Vaughan","is_ca":true},{"name":"Rongbin Yang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05613253156085697,"gpt":0.2367378245921236,"spread":0.1806052930312667,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.01182673,0.0004906548,0.0006435804,0.01128313,0.0006365163,0.004355433,0.00102768,0.000866293,0.002714088],"category_scores_gemma":[0.1379779,0.0003222441,0.0007560411,0.01933949,0.0007717622,0.003301575,0.001646948,0.001199181,0.0005036168],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001207442,"about_ca_system_score_gemma":0.002222805,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006969093,"about_ca_topic_score_gemma":0.004097685,"domain_scores_codex":[0.9886169,0.005165057,0.001480942,0.000823367,0.003205552,0.0007081532],"domain_scores_gemma":[0.6704341,0.2597639,0.03714517,0.008761257,0.02168897,0.002206688],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.000172154,0.00009650834,0.9891823,0.00005980635,0.0001879449,0.00005274159,0.0004568212,0.0005110489,0.000119768,0.0008438428,0.0001952966,0.008121909],"study_design_scores_gemma":[0.0000154428,0.0001638847,0.986987,0.0001057528,0.0004013872,0.0002486467,0.002076235,0.005559009,0.0008568845,0.002011436,0.001548787,0.00002560578],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9857569,0.0005848344,0.003407315,0.0002610343,0.00002753765,0.0000754521,0.002487224,0.0000683906,0.007331421],"genre_scores_gemma":[0.9976358,0.0001469324,0.0008054785,0.0000206553,0.00002621444,0.00004162526,0.0009532574,0.00001859677,0.0003514609],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9887169,"threshold_uncertainty_score":0.06254649,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2084394153","doi":"10.1016/j.joi.2014.08.002","title":"On the citation lifecycle of papers with delayed recognition","year":2014,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":36,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"","keywords":"Citation; Period (music); History; Computer science; Library science; Art","authors":[{"name":"Christian Lachance","is_ca":true},{"name":"Vincent Larivière","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.4451528031489057,"gpt":0.4940995388492365,"spread":0.04894673570033081,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.008430132,0.0004413705,0.0009966637,0.00393766,0.001190649,0.005133571,0.001859581,0.001781856,0.01177612],"category_scores_gemma":[0.07860135,0.0003834555,0.0009834596,0.00397492,0.001759962,0.007645797,0.00211619,0.001777077,0.001370697],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002030764,"about_ca_system_score_gemma":0.00269264,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005272181,"about_ca_topic_score_gemma":0.002279653,"domain_scores_codex":[0.9980963,0.0005124768,0.0001525452,0.0002831489,0.0005041363,0.0004514795],"domain_scores_gemma":[0.9420685,0.03593847,0.007546068,0.004236143,0.007558767,0.002652041],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.001181499,0.0005453447,0.06883934,0.0004622753,0.0001360371,0.001060931,0.001095676,0.0873054,0.005208197,0.6445947,0.00901459,0.1805561],"study_design_scores_gemma":[0.0001341051,0.0004047411,0.02971395,0.000178113,0.0001856957,0.0005680661,0.00105748,0.5447616,0.005479702,0.407718,0.009645452,0.0001531749],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8761234,0.004145774,0.08624568,0.005274974,0.0002883484,0.0001651542,0.001821837,0.0004102621,0.02552462],"genre_scores_gemma":[0.9833406,0.001187733,0.006411699,0.00006860437,0.000146708,0.00005247344,0.0006901997,0.00006837276,0.008033647],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9960623,"threshold_uncertainty_score":0.04458338,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4308297793","doi":"10.1016/j.joi.2022.101344","title":"Detecting emerging technologies and their evolution using deep learning and weak signal analysis","year":2022,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Innovation Diffusion and Forecasting","field":"Decision Sciences","cited_by":36,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Defence Research and Development Canada; National Research Council Canada","funders":"","keywords":"Emerging technologies; Prosperity; Computer science; Data science; Field (mathematics); Identification (biology); Emerging markets; Domain (mathematical analysis); Risk analysis (engineering); Management science; Artificial intelligence; Business; Engineering; Political science; Mathematics","authors":[{"name":"Ashkan Ebadi","is_ca":true},{"name":"Alain Auger","is_ca":true},{"name":"Yvan Gauthier","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06226207505265736,"gpt":0.3336898294886794,"spread":0.2714277544360221,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.001269452,0.0006201261,0.0005291636,0.002820249,0.0002713336,0.001324701,0.0006055113,0.001058537,0.0008277865],"category_scores_gemma":[0.005344723,0.0003039394,0.0005271419,0.001441947,0.0003441633,0.001704092,0.000747209,0.001331355,0.0003800622],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004412107,"about_ca_system_score_gemma":0.0004293705,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002222653,"about_ca_topic_score_gemma":0.002147798,"domain_scores_codex":[0.9996582,0.00006292191,0.00002699845,0.00007628961,0.00009575379,0.00007970622],"domain_scores_gemma":[0.9970095,0.001619209,0.0005326099,0.0001909999,0.0004900306,0.0001576641],"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":[0.000703984,0.0006271693,0.1844518,0.0002941766,0.0004079765,0.00062131,0.0002158752,0.2960711,0.03701228,0.01860985,0.005186701,0.4557977],"study_design_scores_gemma":[0.000004964484,0.0000391642,0.008565815,0.000009064191,0.00002148759,0.00005088395,0.00002897599,0.9824166,0.003248077,0.005228368,0.0003764296,0.00001010285],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6530832,0.001130216,0.3412815,0.0009054877,0.0001058507,0.00003823726,0.0005984395,0.0004282529,0.002428817],"genre_scores_gemma":[0.9789432,0.0002580183,0.01902426,0.00005160979,0.00006111363,0.00001395688,0.0004047206,0.00001929768,0.001223812],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9971797,"threshold_uncertainty_score":0.006713569,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W1993562201","doi":"10.1016/j.joi.2014.01.003","title":"Counting and comparing publication output with and without equalizing and inflationary bias","year":2014,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":35,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Ranking (information retrieval); Accreditation; Econometrics; Publication bias; Statistics; Differential (mechanical device); Computer science; Psychology; Actuarial science; Economics; Information retrieval; Mathematics; Political science; Law; Physics","authors":[{"name":"Nils T. Hagen","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.6028694960870139,"gpt":0.5223753468198439,"spread":0.08049414926717002,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0830619,0.001047659,0.001879253,0.01049711,0.001313352,0.006231014,0.001224509,0.001022225,0.001798074],"category_scores_gemma":[0.4233183,0.0004076477,0.001632059,0.0202304,0.002940521,0.006302005,0.003755005,0.001222138,0.0007558945],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002409665,"about_ca_system_score_gemma":0.002913711,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002124764,"about_ca_topic_score_gemma":0.002086471,"domain_scores_codex":[0.8835139,0.05654722,0.01236703,0.008940422,0.03472953,0.003901874],"domain_scores_gemma":[0.5112373,0.3149936,0.07425589,0.04972646,0.04757944,0.002207363],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.001723094,0.0001518031,0.6531113,0.001388354,0.002407841,0.0003133386,0.003306932,0.01696395,0.005125938,0.02414061,0.003476668,0.2878902],"study_design_scores_gemma":[0.0002310656,0.001622149,0.7830926,0.0006344041,0.002310425,0.000854318,0.002993679,0.05604633,0.02307864,0.1074112,0.0211993,0.0005258246],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.8234386,0.002695933,0.1518125,0.002049294,0.0004849375,0.0003699197,0.002014669,0.000772404,0.0163618],"genre_scores_gemma":[0.9693239,0.0003771632,0.02753766,0.0001616464,0.0002034754,0.0001594135,0.001233389,0.0001396918,0.0008636177],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9895029,"threshold_uncertainty_score":0.4392785,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2736387275","doi":"10.1016/j.joi.2017.07.003","title":"DataCite as a novel bibliometric source: Coverage, strengths and limitations","year":2017,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Research Data Management Practices","field":"Computer Science","cited_by":34,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"Ministerio de Economía y Competitividad; European Commission","keywords":"Bibliometrics; Open data; Value (mathematics); Scientometrics; Production (economics); Open source","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.1482204385072352,"gpt":0.4057818156424827,"spread":0.2575613771352475,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.1109295,0.0007558643,0.002580575,0.05204681,0.002986145,0.01971485,0.004722516,0.002717972,0.01121622],"category_scores_gemma":[0.5060846,0.0008482817,0.001430838,0.09090099,0.003169403,0.01558222,0.007952167,0.003268613,0.003127907],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004592418,"about_ca_system_score_gemma":0.01473222,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01503976,"about_ca_topic_score_gemma":0.02691872,"domain_scores_codex":[0.8582805,0.05515123,0.02397857,0.006931694,0.05401505,0.001642998],"domain_scores_gemma":[0.2770288,0.4895791,0.03742309,0.04816721,0.1412702,0.006531633],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0008797212,0.0003789647,0.1565151,0.0249197,0.001741031,0.0002632173,0.00648188,0.001123373,0.001589932,0.04498041,0.1458569,0.6152698],"study_design_scores_gemma":[0.0002762715,0.0002294169,0.07783289,0.02468524,0.002964237,0.001522803,0.007114196,0.006902667,0.004546217,0.02418077,0.8493326,0.0004126847],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.1925867,0.08541535,0.1110287,0.2057737,0.009473724,0.003835781,0.1997832,0.003392684,0.1887102],"genre_scores_gemma":[0.6261361,0.04588661,0.2010135,0.02056577,0.007129082,0.009180649,0.07558449,0.002713778,0.01179002],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9479532,"threshold_uncertainty_score":0.5866583,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2624878181","doi":"10.1016/j.joi.2017.05.002","title":"Predicting the age of researchers using bibliometric data","year":2017,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":34,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Predictive power; Proxy (statistics); Linear regression; Statistics; Regression analysis; Linear model; Predictive modelling; Regression; Computer science; Econometrics; Demography; Mathematics","authors":[{"name":"Gabriela F. Nane","is_ca":false},{"name":"Vincent Larivière","is_ca":true},{"name":"Rodrigo Costas","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.9265886468305123,"gpt":0.6788366096396239,"spread":0.2477520371908885,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.01142854,0.0006362158,0.001059179,0.01848493,0.0007739851,0.003289012,0.0009077421,0.001146708,0.002606581],"category_scores_gemma":[0.08159205,0.0002846892,0.0009457286,0.0197018,0.0004087793,0.003180695,0.001380048,0.00111242,0.003662437],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003532563,"about_ca_system_score_gemma":0.004034497,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.08895481,"about_ca_topic_score_gemma":0.08573564,"domain_scores_codex":[0.9952643,0.001244765,0.000557708,0.0006881804,0.00174391,0.0005011107],"domain_scores_gemma":[0.9279585,0.03520949,0.01480837,0.004735717,0.01510152,0.002186462],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0001415942,0.00009686103,0.9247236,0.0003291565,0.0002174722,0.0001579276,0.000567747,0.01151465,0.0003223277,0.001668507,0.007532866,0.05272735],"study_design_scores_gemma":[0.00003470672,0.000143993,0.8724988,0.0003958738,0.000197826,0.0002219914,0.001384352,0.07987332,0.00205085,0.005782951,0.03730294,0.000112355],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8722318,0.006464023,0.02891068,0.003196753,0.0002879656,0.0003014939,0.06998356,0.0009625484,0.01766131],"genre_scores_gemma":[0.9543205,0.002066406,0.01176282,0.000149689,0.0002095867,0.0001345666,0.02802239,0.0000630083,0.003271168],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9885715,"threshold_uncertainty_score":0.176874,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2004655250","doi":"10.1016/j.joi.2008.06.005","title":"Biomedical innovation at the laboratory, clinical and commercial interface: A new method for mapping research projects, publications and patents in the field of microarrays","year":2008,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Biomedical and Engineering Education","field":"Engineering","cited_by":30,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal; McGill University","funders":"National Institute of Mental Health; Office of Extramural Research, National Institutes of Health; Canadian Institutes of Health Research; Health Canada; National Institutes of Health","keywords":"Field (mathematics); Biomedicine; Data science; Computer science; Interface (matter); Software; Government (linguistics); Information retrieval; Bioinformatics","authors":[{"name":"Andreï Mogoutov","is_ca":false},{"name":"Alberto Cambrosio","is_ca":true},{"name":"Peter Keating","is_ca":true},{"name":"Philippe Mustar","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1625871724368688,"gpt":0.4348576260224379,"spread":0.2722704535855691,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.007955847,0.0006238543,0.001020383,0.01540706,0.001093639,0.005026808,0.001317474,0.00192817,0.00882774],"category_scores_gemma":[0.02916216,0.0006080119,0.001198454,0.01109269,0.001580944,0.004443069,0.00233358,0.001626333,0.003224937],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001056563,"about_ca_system_score_gemma":0.001527674,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002311523,"about_ca_topic_score_gemma":0.004890912,"domain_scores_codex":[0.9905647,0.003242988,0.0007995635,0.001722568,0.003153394,0.000516799],"domain_scores_gemma":[0.9554226,0.0306711,0.004844402,0.004315574,0.00401134,0.0007349182],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.001074354,0.0003151424,0.08768631,0.001248532,0.0007228118,0.0004009554,0.003353506,0.003227965,0.08270569,0.04882643,0.01336764,0.7570707],"study_design_scores_gemma":[0.0007231536,0.00110603,0.2644623,0.0005084347,0.001790322,0.007564133,0.005885545,0.2044431,0.1343411,0.1800521,0.1981063,0.001017547],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09001929,0.00162809,0.8749732,0.001528566,0.0004261508,0.0005312178,0.005887014,0.003003528,0.02200305],"genre_scores_gemma":[0.2048216,0.0004258672,0.7828836,0.000361911,0.0002116068,0.001046821,0.001323819,0.0003051875,0.008619528],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9845929,"threshold_uncertainty_score":0.04207504,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2092739591","doi":"10.1016/j.joi.2009.12.001","title":"Combining commercial citation indexes and open-access bibliographic databases to delimit highly interdisciplinary research fields for citation analysis","year":2010,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":29,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"Stem Cell Network","keywords":"Citation; Scientometrics; Scopus; Computer science; Field (mathematics); Data science; Citation analysis; Bibliometrics; Information retrieval; Citation database; World Wide Web; MEDLINE; Political science; Mathematics","authors":[{"name":"Andreas Strotmann","is_ca":true},{"name":"Dangzhi Zhao","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.7928641188145871,"gpt":0.6879661796861452,"spread":0.1048979391284419,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.01616915,0.001401927,0.002148927,0.03299422,0.001666102,0.01771483,0.00192707,0.001591303,0.006488825],"category_scores_gemma":[0.1070481,0.0007579778,0.001484005,0.06230805,0.0008758044,0.01221474,0.005862308,0.001624477,0.003467326],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001928667,"about_ca_system_score_gemma":0.004457721,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004348782,"about_ca_topic_score_gemma":0.01073594,"domain_scores_codex":[0.9808304,0.007037529,0.003900653,0.001282,0.006422659,0.0005267057],"domain_scores_gemma":[0.9206802,0.04574987,0.006772574,0.008402897,0.01689588,0.001498609],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0007051419,0.0008421675,0.03278134,0.002823084,0.0009383909,0.0002305812,0.001622565,0.01140448,0.01120129,0.06335532,0.02170214,0.8523935],"study_design_scores_gemma":[0.00087839,0.0009381868,0.07150097,0.001996163,0.003735238,0.0008800629,0.006330225,0.3741372,0.04624122,0.2820072,0.2105203,0.0008348899],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1548882,0.01234455,0.7431931,0.004229932,0.001528497,0.001459215,0.01696963,0.01061261,0.05477426],"genre_scores_gemma":[0.2770422,0.004511342,0.6992574,0.0003683114,0.00101354,0.001121548,0.01208037,0.0009995018,0.003605757],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9838309,"threshold_uncertainty_score":0.08551162,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3032915873","doi":"10.1016/j.joi.2020.101041","title":"Early-stage reciprocity in sustainable scientific collaboration","year":2020,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":20,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Institute of Population and Public Health; China Postdoctoral Science Foundation; King Saud University","keywords":"Reciprocity (cultural anthropology); Sustainability; Teamwork; Altruism (biology); Sustainable development; Computer science; Knowledge management; Business; Political science; Sociology; Economics; Psychology; Management; Social psychology; Social science","authors":[{"name":"Wei Wang","is_ca":false},{"name":"Jing Ren","is_ca":false},{"name":"Mubarak Alrashoud","is_ca":false},{"name":"Mengyi Mao","is_ca":false},{"name":"Amr Tolba","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01774139570065081,"gpt":0.2744047769177139,"spread":0.2566633812170631,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.007915727,0.0002997278,0.0007759224,0.001527801,0.001596381,0.003128604,0.001359234,0.001843679,0.009469598],"category_scores_gemma":[0.04693188,0.0006454923,0.0005860207,0.001196409,0.002151946,0.008579529,0.00416395,0.001829684,0.001388906],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001076813,"about_ca_system_score_gemma":0.001659229,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00135402,"about_ca_topic_score_gemma":0.001993315,"domain_scores_codex":[0.9967865,0.001555522,0.0001508409,0.0006364444,0.000437967,0.000432727],"domain_scores_gemma":[0.9547792,0.03067733,0.004248218,0.004551941,0.003024679,0.002718576],"domain_codex":null,"domain_gemma":"incentives","domain_candidate":"incentives","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.0004917713,0.0002892229,0.04787945,0.0002850509,0.0001281311,0.0005868064,0.004995275,0.02490921,0.003429258,0.8629923,0.002526326,0.05148726],"study_design_scores_gemma":[0.00004475228,0.0002212802,0.01929113,0.00009746145,0.000075434,0.000435916,0.003564572,0.08804244,0.001474619,0.8800126,0.006664644,0.00007529845],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7392498,0.0004965854,0.2148362,0.001512665,0.00006565473,0.000121344,0.0003410114,0.0001821335,0.04319458],"genre_scores_gemma":[0.9881224,0.00009053996,0.007535181,0.00002815691,0.00001695292,0.00004916008,0.00006235623,0.00001848147,0.004076768],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9984722,"threshold_uncertainty_score":0.04186285,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4213241381","doi":"10.1016/j.joi.2022.101256","title":"Female inventors over time: Factors affecting female Inventors’ innovation performance","year":2022,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Innovation Policy and R&D","field":"Economics, Econometrics and Finance","cited_by":16,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"HEC Montréal","funders":"HEC Montréal","keywords":"Business; Production (economics); European patent office; Panel data; Demographic economics; Industrial organization; Economics; International trade; Microeconomics","authors":[{"name":"Leila Tahmooresnejad","is_ca":true},{"name":"Ekaterina Turkina","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0561466103949233,"gpt":0.2499877970366727,"spread":0.1938411866417494,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00460925,0.0002041872,0.0003313014,0.002074705,0.0005333279,0.002186924,0.0005636257,0.0009519636,0.006221063],"category_scores_gemma":[0.02067705,0.0002190292,0.0007597187,0.001381523,0.000596997,0.001400909,0.0007203344,0.0009374106,0.0007769631],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003775362,"about_ca_system_score_gemma":0.0006642121,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00358994,"about_ca_topic_score_gemma":0.003922145,"domain_scores_codex":[0.9983922,0.000397374,0.000209703,0.0002050021,0.0002994563,0.0004963108],"domain_scores_gemma":[0.9592743,0.01945676,0.01270141,0.001442564,0.001994563,0.005130344],"domain_codex":null,"domain_gemma":"incentives","domain_candidate":"incentives","domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00009552357,0.00006225258,0.9961665,0.000005964356,0.00003100815,0.00007444693,0.0003081217,0.00004028839,0.00008704466,0.00005114652,0.00009050117,0.002987229],"study_design_scores_gemma":[0.000004927439,0.00009799218,0.9977695,0.00001314553,0.00002842333,0.0002215773,0.001018803,0.0001355377,0.0001054134,0.00007920965,0.0005187148,0.000006795528],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9976539,0.0006013188,0.0001124777,0.0003538464,0.00002263447,0.000007811016,0.0002268711,0.000002264042,0.001019041],"genre_scores_gemma":[0.9985734,0.0001537231,0.00003900335,0.00003764353,0.00003999436,0.000005420945,0.0001148135,0.000001830563,0.001034135],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9979253,"threshold_uncertainty_score":0.02437639,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2762227057","doi":"10.1016/j.joi.2017.08.008","title":"Mapping science using Library of Congress Subject Headings","year":2017,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Subject (documents); Library of congress; Computer science; Information retrieval; Library science; Library of Congress Classification; Data science; Library classification","authors":[{"name":"Fei Shu","is_ca":true},{"name":"Jesse David Dinneen","is_ca":false},{"name":"Banafsheh Asadi","is_ca":true},{"name":"Charles‐Antoine Julien","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.6384368990329121,"gpt":0.5833211230820153,"spread":0.05511577595089678,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00470261,0.0009005273,0.00106864,0.1504222,0.001575557,0.009952385,0.0006598224,0.0008472997,0.01712294],"category_scores_gemma":[0.038123,0.0002990892,0.001342733,0.177445,0.000619357,0.005432299,0.002954598,0.0006422146,0.005007069],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001926933,"about_ca_system_score_gemma":0.0077847,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02107375,"about_ca_topic_score_gemma":0.01630556,"domain_scores_codex":[0.9934296,0.001699433,0.001591328,0.0007661849,0.002154926,0.000358648],"domain_scores_gemma":[0.9709598,0.01334053,0.00643629,0.00248615,0.005956321,0.0008209745],"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":[0.0002814996,0.000150348,0.1124126,0.009709301,0.001053354,0.0003867369,0.004901415,0.002080046,0.006517496,0.05238573,0.05878003,0.7513415],"study_design_scores_gemma":[0.00008939498,0.0002588963,0.1764528,0.002488328,0.001128144,0.0006480978,0.01063823,0.005161469,0.01057192,0.03920665,0.7532001,0.0001559885],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.3209675,0.04473681,0.07737941,0.005057373,0.002571298,0.001362219,0.2956401,0.00591564,0.2463696],"genre_scores_gemma":[0.6428392,0.03223612,0.1674922,0.0006780432,0.001533253,0.001932261,0.1287274,0.000712313,0.02384918],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8495778,"threshold_uncertainty_score":0.05728191,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410484490","doi":"10.1016/j.joi.2025.101686","title":"Measuring the university–industry–government relations synthesized by the Triple Helix and the diversity","year":2025,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"University-Industry-Government Innovation Models","field":"Business, Management and Accounting","cited_by":3,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Institute on Governance","funders":"Nanjing University; National Natural Science Foundation of China","keywords":"Triple helix; Diversity (politics); Government (linguistics); Computer science; Political science; Business; Data science; Library science; Engineering management; Engineering; Linguistics; Stereochemistry; Chemistry; Philosophy","authors":[{"name":"Shelia X. Wei","is_ca":false},{"name":"Howell Y. Wang","is_ca":false},{"name":"Sanhong Deng","is_ca":false},{"name":"Wanru Wang","is_ca":false},{"name":"Fred Y. Ye","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02322332636499285,"gpt":0.1913355452925803,"spread":0.1681122189275874,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.01173138,0.0002456984,0.0004868734,0.006617402,0.001669501,0.008347714,0.0006938166,0.001566427,0.00434219],"category_scores_gemma":[0.05491814,0.0002223585,0.0003662502,0.01085164,0.003905571,0.009929478,0.006183301,0.001446357,0.0005877908],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002956314,"about_ca_system_score_gemma":0.001998391,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002057566,"about_ca_topic_score_gemma":0.001986843,"domain_scores_codex":[0.9848183,0.009551618,0.0004594025,0.0009285005,0.003429695,0.0008125303],"domain_scores_gemma":[0.9266696,0.05495286,0.008865925,0.003992152,0.003272784,0.002246693],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0007944594,0.0004571983,0.5582638,0.000245488,0.0003786059,0.0002793686,0.03640219,0.01667585,0.003295538,0.2770364,0.003918384,0.1022527],"study_design_scores_gemma":[0.0001223343,0.0008395049,0.4383436,0.0002721188,0.0002700028,0.0003803678,0.109649,0.09089672,0.004692631,0.3275016,0.02686985,0.0001622996],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9462729,0.000217074,0.009375206,0.001816356,0.00004040523,0.00004215454,0.0001979469,0.00002967872,0.04200824],"genre_scores_gemma":[0.9990416,0.00003369173,0.0006374664,0.00003249465,0.00001319955,0.00001083455,0.00002770066,0.000004091493,0.0001988672],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9933826,"threshold_uncertainty_score":0.06204218,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4403785816","doi":"10.1016/j.joi.2025.101759","title":"WISDOM: An AI-powered framework for emerging research detection using weak signal analysis and advanced topic modelling","year":2025,"lang":"en","type":"preprint","venue":"Journal of Informetrics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Defence Research and Development Canada; National Research Council Canada","funders":"","keywords":"SIGNAL (programming language); Computer science; Detection theory; Data science; Artificial intelligence; Telecommunications","authors":[{"name":"Ashkan Ebadi","is_ca":true},{"name":"Alain Auger","is_ca":true},{"name":"Yvan Gauthier","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2034003105124179,"gpt":0.525670070531794,"spread":0.3222697600193761,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.01350452,0.00165803,0.002223715,0.009126758,0.002040765,0.008276497,0.003364322,0.003362417,0.007728137],"category_scores_gemma":[0.05813561,0.0008936432,0.002769443,0.005914059,0.002845972,0.009846504,0.007758322,0.004420588,0.003264869],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001495071,"about_ca_system_score_gemma":0.003745015,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002335398,"about_ca_topic_score_gemma":0.003312553,"domain_scores_codex":[0.9933537,0.003223697,0.0004928424,0.001187012,0.001466968,0.0002758142],"domain_scores_gemma":[0.9671735,0.02336145,0.001859748,0.003608695,0.002676515,0.00132019],"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":[0.0003333448,0.0001579283,0.003933563,0.000767724,0.00037083,0.0003141904,0.001137685,0.02926953,0.00247832,0.6382527,0.018759,0.3042252],"study_design_scores_gemma":[0.00003491446,0.00004003334,0.0004381434,0.00009019604,0.00006255162,0.0001169481,0.0001447969,0.2361391,0.0008144187,0.7508931,0.01118547,0.00004051637],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.00161037,0.0004229024,0.9931582,0.001127786,0.0001353668,0.00008179925,0.0005204907,0.001066644,0.001876483],"genre_scores_gemma":[0.1174406,0.0009547513,0.8711008,0.0008230525,0.0009398901,0.0006152386,0.001664028,0.0004750792,0.005986622],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9908732,"threshold_uncertainty_score":0.0714196,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}