{"meta":{"query_hash":"09653c74ed9a","filters":{"venue":"Journal of Informetrics"},"cohort_total":24,"direct_labels_cover":7,"predictions_cover":24,"exported":24,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/09653c74ed9a","api":"https://metacan.xera.ac/api/v1/cohort?venue=Journal+of+Informetrics"},"results":[{"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.02861058399649574,"score_gpt":0.29994983463309877,"score_spread":0.271339250636603,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1221429143","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.105451524,0.0158316,0.028934134,0.61133176,0.009194486,0.00018295812,0.00037366443,0.00057775155,0.22812206],"genre_scores_gemma":[0.876436,0.007672679,0.014890707,0.0331641,0.0023482505,0.00026380268,0.00026544378,0.00035911973,0.06459992],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9909757,0.004483064,0.00024250729,0.0012625193,0.0015094326,0.0015268335],"domain_scores_gemma":[0.9296727,0.015887616,0.0043038116,0.003267101,0.00834432,0.038524497],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.014524034,0.0007058191,0.0007795714,0.0026331889,0.007956987,0.01840775,0.002585858,0.008296563,0.027912222],"category_scores_gemma":[0.05559898,0.0006646373,0.00059133006,0.002243471,0.009345828,0.048673987,0.013032458,0.007153327,0.010393468],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003438628,0.0004843895,0.014930694,0.00058240845,0.00014459185,0.00094651914,0.010176292,0.00047005963,0.0017232628,0.5945572,0.16126032,0.21438035],"study_design_scores_gemma":[0.000076893724,0.00014211415,0.0037138406,0.00047212085,0.00009185884,0.00089352083,0.025477849,0.0013184682,0.0009277848,0.56291544,0.4038858,0.000084226835],"about_ca_topic_score_codex":0.0016222347,"about_ca_topic_score_gemma":0.002650933,"teacher_disagreement_score":0.9973668,"about_ca_system_score_codex":0.0026617139,"about_ca_system_score_gemma":0.007135647,"threshold_uncertainty_score":0.09337568},"labels":[],"label_agreement":null},{"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.6028694960870139,"score_gpt":0.5223753468198439,"score_spread":0.08049414926717002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993562201","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82343864,0.0026959335,0.15181252,0.002049294,0.00048493745,0.0003699197,0.0020146694,0.000772404,0.016361805],"genre_scores_gemma":[0.96932393,0.00037716323,0.027537659,0.00016164636,0.00020347536,0.00015941351,0.0012333891,0.00013969181,0.0008636177],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.8835139,0.056547225,0.012367026,0.008940422,0.03472953,0.003901874],"domain_scores_gemma":[0.5112373,0.3149936,0.07425589,0.049726464,0.04757944,0.0022073635],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.083061896,0.001047659,0.0018792527,0.010497112,0.0013133517,0.006231014,0.0012245085,0.001022225,0.0017980738],"category_scores_gemma":[0.42331833,0.00040764772,0.0016320593,0.020230401,0.0029405206,0.006302005,0.003755005,0.0012221378,0.0007558945],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001723094,0.00015180308,0.6531113,0.0013883543,0.0024078414,0.00031333865,0.003306932,0.01696395,0.0051259375,0.02414061,0.0034766684,0.28789017],"study_design_scores_gemma":[0.0002310656,0.0016221495,0.78309256,0.0006344041,0.0023104248,0.00085431803,0.0029936794,0.056046333,0.02307864,0.10741124,0.021199297,0.0005258246],"about_ca_topic_score_codex":0.002124764,"about_ca_topic_score_gemma":0.0020864706,"teacher_disagreement_score":0.9895029,"about_ca_system_score_codex":0.0024096647,"about_ca_system_score_gemma":0.0029137114,"threshold_uncertainty_score":0.43927854},"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,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.16258717243686877,"score_gpt":0.4348576260224379,"score_spread":0.2722704535855691,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2004655250","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.090019286,0.00162809,0.87497324,0.0015285661,0.00042615077,0.0005312178,0.005887014,0.003003528,0.022003047],"genre_scores_gemma":[0.20482163,0.0004258672,0.7828836,0.000361911,0.00021160684,0.0010468209,0.0013238187,0.00030518748,0.008619528],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9905647,0.0032429884,0.00079956354,0.0017225682,0.0031533935,0.00051679905],"domain_scores_gemma":[0.95542264,0.030671095,0.004844402,0.0043155737,0.0040113404,0.00073491817],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.007955847,0.0006238543,0.0010203826,0.0154070575,0.0010936387,0.0050268085,0.0013174741,0.0019281702,0.00882774],"category_scores_gemma":[0.029162161,0.0006080119,0.0011984543,0.011092688,0.0015809444,0.0044430695,0.00233358,0.001626333,0.0032249375],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010743542,0.00031514236,0.08768631,0.0012485322,0.0007228118,0.00040095535,0.0033535056,0.0032279654,0.08270569,0.04882643,0.013367641,0.75707066],"study_design_scores_gemma":[0.00072315364,0.0011060303,0.26446226,0.00050843466,0.0017903225,0.007564133,0.005885545,0.2044431,0.13434106,0.18005207,0.19810632,0.0010175465],"about_ca_topic_score_codex":0.002311523,"about_ca_topic_score_gemma":0.0048909117,"teacher_disagreement_score":0.9845929,"about_ca_system_score_codex":0.001056563,"about_ca_system_score_gemma":0.001527674,"threshold_uncertainty_score":0.042075038},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.03543923403875162,"score_gpt":0.38120894412437206,"score_spread":0.34576971008562046,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2041260947","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9976011,0.00008274544,0.00019555356,0.0002482175,0.000020427893,0.00001764794,0.000053278698,0.000012257945,0.0017688827],"genre_scores_gemma":[0.99901724,0.000044108532,0.00021931318,0.0000757369,0.000027349974,0.000015075423,0.000042421674,0.00001193076,0.00054670434],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99470395,0.0027098483,0.00042082314,0.0006101856,0.0010341335,0.00052104436],"domain_scores_gemma":[0.7868529,0.1832851,0.010777123,0.004034249,0.0031148705,0.011935771],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.006839896,0.00022393951,0.0007085969,0.00072434446,0.0008279076,0.0032362754,0.00074371987,0.0014274463,0.015224133],"category_scores_gemma":[0.069290675,0.00035659803,0.0007832253,0.00077386486,0.0013519826,0.001941792,0.0021864523,0.0018248813,0.0006349434],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.023871176,0.016652053,0.8768318,0.00045239198,0.0013099974,0.0007338569,0.01348407,0.0010545582,0.015697088,0.0010634072,0.0009403411,0.047909286],"study_design_scores_gemma":[0.00033875994,0.0034103808,0.98758686,0.00004230566,0.0004878571,0.00018657092,0.004049275,0.00077491027,0.0017257943,0.00066161883,0.0006785898,0.00005714765],"about_ca_topic_score_codex":0.00176994,"about_ca_topic_score_gemma":0.001533209,"teacher_disagreement_score":0.9992757,"about_ca_system_score_codex":0.0005111471,"about_ca_system_score_gemma":0.00083275646,"threshold_uncertainty_score":0.050929785},"labels":[],"label_agreement":null},{"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Webometrics; Web traffic; Hyperlink; Computer science; World Wide Web; The Internet; Web page","score_opus":0.05613253156085697,"score_gpt":0.23673782459212364,"score_spread":0.18060529303126666,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2062524032","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98575693,0.00058483443,0.0034073147,0.0002610343,0.000027537646,0.0000754521,0.0024872236,0.0000683906,0.007331421],"genre_scores_gemma":[0.99763584,0.00014693243,0.0008054785,0.0000206553,0.000026214442,0.00004162526,0.00095325743,0.000018596766,0.0003514609],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9886169,0.005165057,0.0014809415,0.000823367,0.0032055522,0.0007081532],"domain_scores_gemma":[0.67043406,0.2597639,0.03714517,0.008761257,0.021688972,0.0022066878],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.011826733,0.00049065484,0.0006435804,0.011283126,0.0006365163,0.004355433,0.0010276797,0.000866293,0.0027140884],"category_scores_gemma":[0.13797793,0.00032224413,0.00075604115,0.01933949,0.0007717622,0.0033015746,0.0016469478,0.0011991811,0.0005036168],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017215402,0.00009650834,0.9891823,0.000059806353,0.00018794485,0.000052741587,0.00045682123,0.0005110489,0.000119767996,0.0008438428,0.00019529663,0.008121909],"study_design_scores_gemma":[0.000015442796,0.00016388469,0.986987,0.000105752784,0.00040138717,0.00024864674,0.0020762347,0.005559009,0.00085688446,0.0020114358,0.001548787,0.000025605776],"about_ca_topic_score_codex":0.0069690933,"about_ca_topic_score_gemma":0.0040976848,"teacher_disagreement_score":0.9887169,"about_ca_system_score_codex":0.0012074417,"about_ca_system_score_gemma":0.002222805,"threshold_uncertainty_score":0.06254649},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"","keywords":"Citation; Period (music); History; Computer science; Library science; Art","score_opus":0.4451528031489057,"score_gpt":0.4940995388492365,"score_spread":0.04894673570033081,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2084394153","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.87612337,0.0041457736,0.08624568,0.005274974,0.0002883484,0.00016515418,0.0018218368,0.00041026215,0.025524618],"genre_scores_gemma":[0.98334056,0.0011877327,0.0064116986,0.00006860437,0.00014670799,0.000052473435,0.0006901997,0.00006837276,0.008033647],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9980963,0.00051247684,0.00015254517,0.0002831489,0.0005041363,0.00045147946],"domain_scores_gemma":[0.94206846,0.035938468,0.007546068,0.004236143,0.007558767,0.002652041],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.008430132,0.0004413705,0.0009966637,0.0039376603,0.0011906486,0.0051335706,0.0018595814,0.0017818564,0.011776122],"category_scores_gemma":[0.078601345,0.0003834555,0.0009834596,0.0039749197,0.001759962,0.0076457975,0.0021161898,0.0017770766,0.0013706974],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011814986,0.0005453447,0.06883934,0.0004622753,0.00013603708,0.0010609315,0.0010956764,0.0873054,0.0052081966,0.64459467,0.00901459,0.1805561],"study_design_scores_gemma":[0.00013410508,0.00040474106,0.029713953,0.000178113,0.00018569565,0.00056806614,0.0010574803,0.5447616,0.005479702,0.40771797,0.009645452,0.00015317487],"about_ca_topic_score_codex":0.005272181,"about_ca_topic_score_gemma":0.0022796525,"teacher_disagreement_score":0.99606234,"about_ca_system_score_codex":0.0020307638,"about_ca_system_score_gemma":0.0026926398,"threshold_uncertainty_score":0.04458338},"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,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.7928641188145871,"score_gpt":0.6879661796861452,"score_spread":0.10489793912844192,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2092739591","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1548882,0.012344549,0.7431931,0.004229932,0.0015284965,0.0014592151,0.01696963,0.010612605,0.05477426],"genre_scores_gemma":[0.2770422,0.004511342,0.6992574,0.0003683114,0.00101354,0.0011215485,0.012080366,0.0009995018,0.003605757],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98083043,0.0070375293,0.0039006528,0.0012819996,0.006422659,0.0005267057],"domain_scores_gemma":[0.92068017,0.045749873,0.0067725736,0.008402897,0.016895879,0.0014986092],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.016169151,0.0014019267,0.0021489267,0.032994222,0.0016661016,0.017714828,0.0019270703,0.0015913026,0.006488825],"category_scores_gemma":[0.107048124,0.0007579778,0.0014840051,0.062308047,0.0008758044,0.012214735,0.005862308,0.0016244774,0.003467326],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00070514187,0.00084216753,0.03278134,0.0028230837,0.0009383909,0.0002305812,0.001622565,0.011404479,0.0112012895,0.06335532,0.02170214,0.8523935],"study_design_scores_gemma":[0.00087839004,0.0009381868,0.07150097,0.001996163,0.0037352375,0.0008800629,0.006330225,0.37413716,0.046241224,0.28200725,0.21052028,0.0008348899],"about_ca_topic_score_codex":0.004348782,"about_ca_topic_score_gemma":0.010735943,"teacher_disagreement_score":0.98383087,"about_ca_system_score_codex":0.0019286668,"about_ca_system_score_gemma":0.004457721,"threshold_uncertainty_score":0.085511625},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.03318613078372981,"score_gpt":0.22461751359315724,"score_spread":0.19143138280942743,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2129840456","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20716912,0.014159102,0.7197979,0.035193626,0.0011625641,0.000045316323,0.0010941228,0.0006501411,0.02072816],"genre_scores_gemma":[0.92085713,0.0074281828,0.05809332,0.0010892767,0.0012076584,0.00009871828,0.0003620575,0.00026864524,0.010595068],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99847,0.0009229802,0.0000715208,0.0002448423,0.00019021131,0.00010045146],"domain_scores_gemma":[0.9814813,0.014297371,0.00092473417,0.0017659678,0.000932797,0.0005977563],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.004713895,0.00048910617,0.0009685613,0.001953257,0.00084083405,0.0041839224,0.0022271476,0.0033356007,0.003860443],"category_scores_gemma":[0.039685,0.0009284079,0.0008134763,0.0021102128,0.002979392,0.009792574,0.0021375937,0.0035989678,0.00046822228],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006113195,0.00009979922,0.0053419922,0.00007418973,0.00012545986,0.00012243337,0.0005884637,0.2866986,0.00019427322,0.659683,0.006349077,0.0406616],"study_design_scores_gemma":[0.000009976581,0.000008127442,0.00048759932,0.00003096166,0.000015351548,0.0000589192,0.00008510101,0.5196727,0.00007048108,0.47406596,0.0054773744,0.000017499962],"about_ca_topic_score_codex":0.0077008484,"about_ca_topic_score_gemma":0.006858651,"teacher_disagreement_score":0.99804676,"about_ca_system_score_codex":0.0022373942,"about_ca_system_score_gemma":0.0012099151,"threshold_uncertainty_score":0.024929762},"labels":[],"label_agreement":null},{"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.578649466269294,"score_gpt":0.5307756849711357,"score_spread":0.047873781298158335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134386090","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028281638,0.00033326237,0.96082526,0.00062278594,0.0001244145,0.00033673926,0.00030975195,0.00073524535,0.008430893],"genre_scores_gemma":[0.23703468,0.00037511092,0.7583946,0.000089490415,0.000083785366,0.00023896518,0.00036046517,0.00006779842,0.003355118],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9953284,0.0018530131,0.00030107403,0.00029962295,0.0020022918,0.000215641],"domain_scores_gemma":[0.99062157,0.0036320654,0.00035441428,0.00058090745,0.0045144963,0.0002965357],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0042838063,0.000547544,0.0011628213,0.006422334,0.00094699993,0.0034375223,0.0013826565,0.0009782545,0.004150451],"category_scores_gemma":[0.016255852,0.0004183847,0.0011417918,0.0047661434,0.0004455532,0.0028795158,0.0011819466,0.001054356,0.0014975029],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022872158,0.00059042155,0.008458835,0.00056994776,0.00037869686,0.00024120153,0.00032654367,0.057314787,0.008139842,0.07184172,0.0077307695,0.8441784],"study_design_scores_gemma":[0.00006400125,0.00045421548,0.0042029964,0.000069556496,0.000175362,0.0002979835,0.00025751608,0.94240904,0.006769102,0.035480414,0.00971019,0.00010958835],"about_ca_topic_score_codex":0.005447386,"about_ca_topic_score_gemma":0.0076669375,"teacher_disagreement_score":0.9957162,"about_ca_system_score_codex":0.0011670353,"about_ca_system_score_gemma":0.003742847,"threshold_uncertainty_score":0.022655189},"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":"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.8625684102688479,"score_gpt":0.6369410142134144,"score_spread":0.22562739605543347,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150897866","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82804346,0.022875251,0.10966821,0.004345784,0.00042335535,0.0005820994,0.0022114967,0.00032896604,0.031521324],"genre_scores_gemma":[0.92567044,0.0037881024,0.06563744,0.0006158964,0.00024151088,0.00017311783,0.0012988247,0.00007283297,0.0025016896],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9582388,0.03064423,0.0014272452,0.0012802238,0.0079086395,0.00050089764],"domain_scores_gemma":[0.5815275,0.37969896,0.00438351,0.007923705,0.02560654,0.00085973024],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.046299923,0.00079618616,0.0014475206,0.008720539,0.0008690643,0.002192702,0.0012486172,0.0020752994,0.0032826366],"category_scores_gemma":[0.16675662,0.0001826083,0.0012161656,0.00942368,0.00096819963,0.0025771214,0.0014326847,0.0010723129,0.00067799597],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0059722513,0.0031789327,0.3295632,0.0039754054,0.0051250304,0.0003572265,0.0034223516,0.024925621,0.0031709678,0.021889312,0.0147870965,0.5836326],"study_design_scores_gemma":[0.0017284792,0.009155566,0.598398,0.0015079572,0.0053479397,0.0016289582,0.011354903,0.25401607,0.012358437,0.06413069,0.0397123,0.00066071],"about_ca_topic_score_codex":0.0076474412,"about_ca_topic_score_gemma":0.011787687,"teacher_disagreement_score":0.9912795,"about_ca_system_score_codex":0.0010952912,"about_ca_system_score_gemma":0.0012391219,"threshold_uncertainty_score":0.24486029},"labels":[],"label_agreement":null},{"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,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.1625741757646505,"score_gpt":0.461230812010258,"score_spread":0.2986566362456075,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2191720934","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47483855,0.22252984,0.016041644,0.12571026,0.026094142,0.0004510092,0.009209718,0.0007991634,0.12432579],"genre_scores_gemma":[0.9096039,0.037503075,0.0068777935,0.009971501,0.019414727,0.00029701853,0.004139919,0.00042722808,0.011764885],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.93406355,0.017064072,0.0123709785,0.00937392,0.022747425,0.0043801474],"domain_scores_gemma":[0.579525,0.2003029,0.09303771,0.027226234,0.06718995,0.032718204],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.044121943,0.0010958915,0.0037981165,0.034060195,0.0046150275,0.038594507,0.003084013,0.006140408,0.035100643],"category_scores_gemma":[0.28652436,0.001039573,0.0022798511,0.06463293,0.003376351,0.018760609,0.006949789,0.0032383099,0.007043443],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027990926,0.0009395782,0.48818165,0.011489274,0.009849382,0.0015542597,0.007998778,0.0008383203,0.004929084,0.026116986,0.05663872,0.38866484],"study_design_scores_gemma":[0.0009175303,0.0007570809,0.660415,0.005953767,0.005940383,0.0037314547,0.023957875,0.0011674949,0.0039676833,0.097086124,0.19571242,0.00039314543],"about_ca_topic_score_codex":0.00093257736,"about_ca_topic_score_gemma":0.0018273027,"teacher_disagreement_score":0.9659398,"about_ca_system_score_codex":0.002698384,"about_ca_system_score_gemma":0.0036700864,"threshold_uncertainty_score":0.23334187},"labels":[],"label_agreement":null},{"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.9265886468305123,"score_gpt":0.6788366096396239,"score_spread":0.24775203719088845,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2624878181","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8722318,0.0064640227,0.028910678,0.0031967526,0.0002879656,0.00030149386,0.069983564,0.0009625484,0.017661309],"genre_scores_gemma":[0.9543205,0.002066406,0.0117628155,0.00014968902,0.00020958668,0.00013456661,0.028022392,0.0000630083,0.0032711683],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9952643,0.0012447655,0.00055770803,0.00068818044,0.0017439099,0.00050111074],"domain_scores_gemma":[0.92795855,0.03520949,0.01480837,0.0047357166,0.015101518,0.0021864623],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.011428539,0.0006362158,0.0010591787,0.018484926,0.00077398506,0.0032890115,0.00090774207,0.0011467077,0.002606581],"category_scores_gemma":[0.08159205,0.00028468916,0.00094572856,0.019701801,0.00040877933,0.0031806948,0.0013800485,0.0011124203,0.0036624367],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014159424,0.00009686103,0.9247236,0.00032915652,0.00021747225,0.00015792756,0.00056774705,0.011514654,0.00032232772,0.0016685072,0.0075328657,0.052727345],"study_design_scores_gemma":[0.00003470672,0.00014399302,0.87249875,0.00039587377,0.00019782603,0.00022199137,0.0013843516,0.07987332,0.00205085,0.0057829507,0.037302945,0.00011235502],"about_ca_topic_score_codex":0.088954814,"about_ca_topic_score_gemma":0.08573564,"teacher_disagreement_score":0.98857147,"about_ca_system_score_codex":0.0035325629,"about_ca_system_score_gemma":0.004034497,"threshold_uncertainty_score":0.17687404},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.14822043850723518,"score_gpt":0.4057818156424827,"score_spread":0.2575613771352475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2736387275","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19258668,0.08541535,0.111028686,0.20577373,0.009473724,0.003835781,0.19978319,0.0033926836,0.18871024],"genre_scores_gemma":[0.6261361,0.045886613,0.20101354,0.020565767,0.0071290825,0.009180649,0.075584486,0.0027137778,0.01179002],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.85828054,0.05515123,0.023978574,0.006931694,0.05401505,0.0016429976],"domain_scores_gemma":[0.27702877,0.48957905,0.037423093,0.048167206,0.14127022,0.0065316334],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.110929504,0.0007558643,0.0025805752,0.052046813,0.0029861447,0.019714851,0.0047225156,0.0027179723,0.011216218],"category_scores_gemma":[0.50608456,0.00084828166,0.0014308385,0.09090099,0.003169403,0.015582221,0.007952167,0.0032686125,0.003127907],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008797212,0.00037896467,0.1565151,0.0249197,0.0017410311,0.00026321734,0.0064818803,0.0011233728,0.0015899317,0.044980407,0.14585687,0.6152698],"study_design_scores_gemma":[0.00027627154,0.0002294169,0.077832885,0.024685241,0.002964237,0.0015228033,0.0071141957,0.006902667,0.0045462167,0.024180774,0.84933263,0.00041268466],"about_ca_topic_score_codex":0.015039762,"about_ca_topic_score_gemma":0.026918717,"teacher_disagreement_score":0.94795316,"about_ca_system_score_codex":0.004592418,"about_ca_system_score_gemma":0.014732218,"threshold_uncertainty_score":0.5866583},"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":"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.6384368990329121,"score_gpt":0.5833211230820153,"score_spread":0.055115775950896784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2762227057","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.32096753,0.044736806,0.077379405,0.005057373,0.0025712978,0.0013622189,0.29564008,0.0059156404,0.24636964],"genre_scores_gemma":[0.6428392,0.03223612,0.16749221,0.00067804323,0.0015332527,0.0019322608,0.12872742,0.000712313,0.02384918],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9934296,0.0016994326,0.0015913284,0.00076618494,0.0021549263,0.00035864796],"domain_scores_gemma":[0.9709598,0.013340526,0.00643629,0.0024861495,0.0059563206,0.0008209745],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0047026104,0.00090052735,0.0010686402,0.15042222,0.001575557,0.009952385,0.0006598224,0.0008472997,0.017122937],"category_scores_gemma":[0.038123,0.0002990892,0.0013427327,0.17744502,0.00061935704,0.005432299,0.0029545983,0.0006422146,0.005007069],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028149958,0.00015034803,0.11241257,0.009709301,0.0010533544,0.0003867369,0.004901415,0.0020800459,0.0065174964,0.052385725,0.058780026,0.75134146],"study_design_scores_gemma":[0.000089394984,0.0002588963,0.17645277,0.0024883277,0.0011281435,0.00064809783,0.01063823,0.0051614693,0.010571923,0.03920665,0.75320005,0.00015598853],"about_ca_topic_score_codex":0.021073753,"about_ca_topic_score_gemma":0.01630556,"teacher_disagreement_score":0.8495778,"about_ca_system_score_codex":0.0019269327,"about_ca_system_score_gemma":0.0077847005,"threshold_uncertainty_score":0.05728191},"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":"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.4491389199918624,"score_gpt":0.49777142617191705,"score_spread":0.04863250618005466,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2781044009","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9962829,0.00048936997,0.0010349712,0.00012066074,0.000007688135,0.000010233212,0.001101074,0.000048378282,0.0009047416],"genre_scores_gemma":[0.99699855,0.0001961925,0.0011852375,0.000011565973,0.000031344494,0.000013399858,0.0012542367,0.000019336765,0.00029023728],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99753773,0.0008902885,0.0003266971,0.00047521712,0.00063762657,0.00013253013],"domain_scores_gemma":[0.89007217,0.09421062,0.008805728,0.0030692115,0.002638785,0.0012033725],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.004119085,0.00039846238,0.00081626896,0.01418869,0.0009909152,0.0028854362,0.0007423745,0.00078468723,0.0024204983],"category_scores_gemma":[0.044277783,0.00027094712,0.0012607474,0.016978297,0.00088069687,0.002933854,0.0015549805,0.00082618755,0.0007799296],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001484918,0.00008320886,0.98009425,0.00008833065,0.0006319938,0.00012697664,0.00045449974,0.0010024578,0.001021229,0.0003519287,0.00029435492,0.0157023],"study_design_scores_gemma":[0.0000123933105,0.000053936303,0.98858964,0.00001587351,0.00025621965,0.00024482995,0.00052247324,0.008622396,0.00052949,0.0006400804,0.000486543,0.000026214055],"about_ca_topic_score_codex":0.005694335,"about_ca_topic_score_gemma":0.007813164,"teacher_disagreement_score":0.9958809,"about_ca_system_score_codex":0.0005857043,"about_ca_system_score_gemma":0.00090730714,"threshold_uncertainty_score":0.021784067},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.7615079244057587,"score_gpt":0.5814149810599263,"score_spread":0.18009294334583237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2908106500","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9689168,0.001378244,0.0028340612,0.00060156925,0.00017056208,0.00007199525,0.0044976505,0.00008423939,0.021444961],"genre_scores_gemma":[0.9907596,0.00038664116,0.0028017007,0.00010113877,0.00012201912,0.00006881448,0.0044824374,0.000041075295,0.0012365842],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99106616,0.0035429364,0.0011802684,0.0007073335,0.0026458246,0.000857416],"domain_scores_gemma":[0.8528819,0.09209782,0.021817682,0.007111813,0.020764709,0.005326133],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.008995476,0.00034860376,0.0007097425,0.029380048,0.000947903,0.007389792,0.00078071596,0.0011996444,0.0095778005],"category_scores_gemma":[0.08237549,0.00018390449,0.0008798636,0.022836592,0.0011416331,0.0038761161,0.0026300012,0.00092213537,0.0018746199],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013289015,0.00017756181,0.94116676,0.00037466863,0.000521725,0.00006331469,0.0017370577,0.0005574893,0.0028122745,0.008738606,0.0031243654,0.03939745],"study_design_scores_gemma":[0.00010581524,0.0003353108,0.968101,0.00022411621,0.00043577855,0.00009912118,0.006453929,0.0027726518,0.0014881463,0.013831525,0.0061062193,0.000046476303],"about_ca_topic_score_codex":0.0028134394,"about_ca_topic_score_gemma":0.0034772886,"teacher_disagreement_score":0.97062,"about_ca_system_score_codex":0.0014718439,"about_ca_system_score_gemma":0.0014477569,"threshold_uncertainty_score":0.04757315},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.024388254781145414,"score_gpt":0.3143265421653923,"score_spread":0.2899382873842469,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2967837635","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95397645,0.0016352346,0.014774297,0.0031327787,0.00009560063,0.000088671055,0.0009902067,0.00014320688,0.025163446],"genre_scores_gemma":[0.99472106,0.0003065239,0.003753889,0.000067314846,0.0001385776,0.000028671055,0.00035966616,0.000034530804,0.0005896469],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99238247,0.0026450949,0.00055962993,0.00055210246,0.003431752,0.00042880094],"domain_scores_gemma":[0.78886086,0.17190936,0.015675677,0.006105574,0.013803421,0.00364516],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.008062338,0.00039741764,0.00070931576,0.019047104,0.0017605476,0.0071885386,0.0008551857,0.0009950238,0.005903327],"category_scores_gemma":[0.117697276,0.00041266414,0.0007490084,0.012455265,0.0036302423,0.008554831,0.0040215286,0.0019456651,0.0005009656],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001654333,0.0008975413,0.6350241,0.0008065751,0.0010387937,0.0010538781,0.0058903415,0.028868537,0.006414833,0.11973263,0.0052089514,0.19340953],"study_design_scores_gemma":[0.00015054172,0.00040209174,0.41017583,0.00025444105,0.0007799708,0.00063744315,0.011313086,0.14065903,0.004838499,0.41904002,0.011514251,0.00023470406],"about_ca_topic_score_codex":0.0044750813,"about_ca_topic_score_gemma":0.0045965835,"teacher_disagreement_score":0.9809529,"about_ca_system_score_codex":0.0023549956,"about_ca_system_score_gemma":0.0020259493,"threshold_uncertainty_score":0.042638183},"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,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.017741395700650812,"score_gpt":0.27440477691771387,"score_spread":0.25666338121706306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3032915873","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.73924977,0.00049658545,0.21483618,0.0015126647,0.000065654734,0.000121343975,0.00034101136,0.00018213352,0.043194577],"genre_scores_gemma":[0.9881224,0.00009053996,0.0075351815,0.000028156914,0.000016952921,0.000049160084,0.00006235623,0.000018481469,0.004076768],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9967865,0.0015555223,0.00015084087,0.00063644437,0.00043796696,0.00043272704],"domain_scores_gemma":[0.9547792,0.030677333,0.0042482177,0.004551941,0.0030246794,0.0027185755],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.007915727,0.00029972778,0.00077592244,0.0015278014,0.0015963811,0.0031286043,0.0013592344,0.0018436786,0.009469598],"category_scores_gemma":[0.04693188,0.00064549234,0.0005860207,0.0011964092,0.0021519463,0.008579529,0.0041639498,0.0018296839,0.0013889062],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004917713,0.00028922292,0.047879446,0.00028505086,0.00012813105,0.0005868064,0.0049952753,0.024909208,0.0034292578,0.8629923,0.0025263259,0.051487256],"study_design_scores_gemma":[0.000044752276,0.00022128024,0.019291133,0.00009746145,0.000075434,0.00043591595,0.0035645722,0.08804244,0.0014746186,0.8800126,0.006664644,0.00007529845],"about_ca_topic_score_codex":0.0013540204,"about_ca_topic_score_gemma":0.0019933148,"teacher_disagreement_score":0.9984722,"about_ca_system_score_codex":0.0010768125,"about_ca_system_score_gemma":0.0016592289,"threshold_uncertainty_score":0.041862845},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.7915538673463653,"score_gpt":0.540354756721656,"score_spread":0.25119911062470934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3115533035","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98795915,0.0016892574,0.0022035982,0.00036548942,0.000036753976,0.000028495522,0.00134053,0.000049625323,0.006327087],"genre_scores_gemma":[0.99531376,0.00055200176,0.0017439916,0.000027649694,0.000063782005,0.00001956932,0.0011112831,0.00002406769,0.0011437966],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99659497,0.00073857425,0.000638815,0.000616551,0.0011769849,0.00023415346],"domain_scores_gemma":[0.94459873,0.03336598,0.009000393,0.0028122365,0.008634342,0.0015883213],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0036124433,0.00023343146,0.0004420954,0.017899942,0.0010638771,0.0048534707,0.0008072689,0.00075939443,0.0039089746],"category_scores_gemma":[0.043244295,0.00017716427,0.00052741077,0.027367376,0.0004844472,0.002572994,0.0016606862,0.0006234801,0.0010039696],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002924378,0.00015451829,0.88152206,0.0005476995,0.00065826566,0.0005189207,0.0048904316,0.00067216926,0.007615804,0.0044304067,0.0017298522,0.09696742],"study_design_scores_gemma":[0.000028210416,0.00013877844,0.9659827,0.00012608017,0.00030028762,0.00083494483,0.007134762,0.0048704827,0.0029532576,0.010075362,0.007510495,0.0000446983],"about_ca_topic_score_codex":0.0018780008,"about_ca_topic_score_gemma":0.0034280869,"teacher_disagreement_score":0.99638754,"about_ca_system_score_codex":0.00051515177,"about_ca_system_score_gemma":0.0007553541,"threshold_uncertainty_score":0.01910466},"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":"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,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.0561466103949233,"score_gpt":0.24998779703667268,"score_spread":0.19384118664174937,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4213241381","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9976539,0.00060131884,0.000112477726,0.0003538464,0.000022634475,0.000007811016,0.0002268711,0.0000022640422,0.0010190406],"genre_scores_gemma":[0.9985734,0.00015372313,0.000039003346,0.000037643535,0.000039994364,0.0000054209454,0.00011481347,0.0000018305631,0.001034135],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99839216,0.00039737404,0.00020970296,0.00020500214,0.00029945633,0.0004963108],"domain_scores_gemma":[0.9592743,0.019456755,0.012701412,0.0014425638,0.0019945635,0.005130344],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0046092505,0.00020418718,0.00033130136,0.002074705,0.0005333279,0.0021869245,0.00056362565,0.0009519636,0.0062210625],"category_scores_gemma":[0.020677047,0.00021902921,0.0007597187,0.0013815226,0.000596997,0.0014009092,0.00072033436,0.0009374106,0.00077696313],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000095523574,0.00006225258,0.99616647,0.000005964356,0.00003100815,0.00007444693,0.00030812167,0.00004028839,0.00008704466,0.000051146522,0.00009050117,0.0029872293],"study_design_scores_gemma":[0.0000049274386,0.00009799218,0.9977695,0.000013145526,0.000028423327,0.00022157727,0.0010188031,0.00013553766,0.00010541338,0.000079209654,0.00051871483,0.000006795528],"about_ca_topic_score_codex":0.00358994,"about_ca_topic_score_gemma":0.0039221453,"teacher_disagreement_score":0.9979253,"about_ca_system_score_codex":0.00037753623,"about_ca_system_score_gemma":0.0006642121,"threshold_uncertainty_score":0.024376392},"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":"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.103223804081119,"score_gpt":0.3570945740796192,"score_spread":0.2538707699985002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W423015428","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.27917996,0.005025704,0.7071362,0.0030022059,0.00031492038,0.000931716,0.0010856166,0.00044387588,0.0028797411],"genre_scores_gemma":[0.8191155,0.0038136055,0.16942944,0.0005021491,0.00027053748,0.0016875102,0.0010815185,0.00022114595,0.0038784887],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.93975127,0.05364316,0.0016212679,0.0022376846,0.0021183544,0.0006283176],"domain_scores_gemma":[0.454904,0.5239224,0.010077566,0.0051069357,0.005247912,0.0007411545],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.070555255,0.0017706831,0.00243211,0.004490251,0.0008447333,0.005542858,0.005063356,0.0037855976,0.005513296],"category_scores_gemma":[0.27881673,0.00085572124,0.0035437006,0.0064615053,0.0021470515,0.008207985,0.0026407635,0.0035166447,0.001320098],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0074979346,0.0015120083,0.1572095,0.0049407305,0.0041814153,0.00083086453,0.005764341,0.40571794,0.0010913606,0.13484067,0.004204678,0.27220848],"study_design_scores_gemma":[0.00032645126,0.0010868552,0.01134864,0.00069594657,0.0011861789,0.00034665718,0.0018874602,0.90548056,0.0006569601,0.07434648,0.0024964663,0.00014137963],"about_ca_topic_score_codex":0.009292494,"about_ca_topic_score_gemma":0.0077116527,"teacher_disagreement_score":0.99550974,"about_ca_system_score_codex":0.0030722108,"about_ca_system_score_gemma":0.0030965116,"threshold_uncertainty_score":0.37313628},"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":"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.062262075052657365,"score_gpt":0.3336898294886794,"score_spread":0.2714277544360221,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4308297793","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6530832,0.0011302158,0.3412815,0.0009054877,0.00010585068,0.000038237256,0.0005984395,0.0004282529,0.0024288166],"genre_scores_gemma":[0.97894317,0.00025801826,0.019024257,0.000051609793,0.000061113635,0.000013956876,0.00040472057,0.000019297682,0.0012238125],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9996582,0.000062921914,0.00002699845,0.00007628961,0.00009575379,0.000079706224],"domain_scores_gemma":[0.9970095,0.0016192091,0.00053260993,0.00019099988,0.00049003056,0.00015766414],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0012694522,0.00062012614,0.00052916363,0.0028202487,0.00027133359,0.0013247014,0.0006055113,0.0010585373,0.00082778645],"category_scores_gemma":[0.0053447234,0.00030393942,0.0005271419,0.001441947,0.00034416333,0.001704092,0.000747209,0.001331355,0.0003800622],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000703984,0.00062716927,0.18445182,0.00029417663,0.0004079765,0.00062130997,0.00021587525,0.2960711,0.037012283,0.018609853,0.005186701,0.45579773],"study_design_scores_gemma":[0.000004964484,0.0000391642,0.008565815,0.000009064191,0.000021487589,0.000050883948,0.000028975994,0.9824166,0.0032480771,0.005228368,0.00037642958,0.000010102852],"about_ca_topic_score_codex":0.0022226528,"about_ca_topic_score_gemma":0.0021477982,"teacher_disagreement_score":0.99717975,"about_ca_system_score_codex":0.00044121066,"about_ca_system_score_gemma":0.00042937053,"threshold_uncertainty_score":0.006713569},"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":"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.20340031051241791,"score_gpt":0.525670070531794,"score_spread":0.32226976001937613,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403785816","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016103697,0.0004229024,0.99315816,0.0011277865,0.00013536683,0.00008179925,0.0005204907,0.0010666441,0.001876483],"genre_scores_gemma":[0.11744056,0.0009547513,0.8711008,0.00082305254,0.0009398901,0.00061523856,0.0016640284,0.00047507917,0.0059866216],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9933537,0.003223697,0.0004928424,0.0011870124,0.0014669681,0.00027581418],"domain_scores_gemma":[0.96717346,0.023361454,0.001859748,0.0036086948,0.0026765154,0.0013201896],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.013504519,0.0016580304,0.0022237147,0.009126758,0.0020407652,0.008276497,0.0033643222,0.003362417,0.007728137],"category_scores_gemma":[0.058135614,0.0008936432,0.0027694427,0.0059140595,0.0028459718,0.009846504,0.007758322,0.004420588,0.003264869],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033334483,0.00015792834,0.003933563,0.000767724,0.00037083004,0.00031419037,0.001137685,0.029269528,0.0024783204,0.6382527,0.018758995,0.30422524],"study_design_scores_gemma":[0.000034914465,0.00004003334,0.0004381434,0.000090196045,0.00006255162,0.000116948104,0.00014479694,0.23613906,0.0008144187,0.75089306,0.011185471,0.000040516366],"about_ca_topic_score_codex":0.002335398,"about_ca_topic_score_gemma":0.0033125533,"teacher_disagreement_score":0.9908732,"about_ca_system_score_codex":0.0014950709,"about_ca_system_score_gemma":0.0037450148,"threshold_uncertainty_score":0.0714196},"labels":[],"label_agreement":null},{"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.02322332636499285,"score_gpt":0.19133554529258026,"score_spread":0.1681122189275874,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410484490","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9462729,0.00021707403,0.009375206,0.001816356,0.00004040523,0.00004215454,0.00019794694,0.000029678722,0.042008244],"genre_scores_gemma":[0.9990416,0.000033691726,0.00063746644,0.000032494652,0.000013199546,0.000010834552,0.00002770066,0.0000040914933,0.00019886717],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9848183,0.009551618,0.00045940245,0.0009285005,0.0034296955,0.00081253034],"domain_scores_gemma":[0.9266696,0.054952864,0.0088659255,0.003992152,0.0032727837,0.002246693],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.011731375,0.00024569838,0.00048687344,0.0066174017,0.0016695013,0.008347714,0.00069381663,0.0015664271,0.0043421905],"category_scores_gemma":[0.054918136,0.00022235847,0.0003662502,0.010851642,0.0039055713,0.009929478,0.006183301,0.0014463568,0.0005877908],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00079445937,0.00045719833,0.55826384,0.00024548805,0.00037860588,0.00027936863,0.03640219,0.016675847,0.0032955378,0.2770364,0.003918384,0.10225271],"study_design_scores_gemma":[0.00012233428,0.0008395049,0.4383436,0.0002721188,0.0002700028,0.0003803678,0.109649025,0.09089672,0.004692631,0.32750162,0.02686985,0.0001622996],"about_ca_topic_score_codex":0.0020575663,"about_ca_topic_score_gemma":0.0019868426,"teacher_disagreement_score":0.9933826,"about_ca_system_score_codex":0.002956314,"about_ca_system_score_gemma":0.0019983915,"threshold_uncertainty_score":0.062042177},"labels":[],"label_agreement":null}]}