{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":199,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":199,"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":"517981b04501","filters":{"venue":"Journal of Cultural Analytics"}},"results":[{"id":"W3085332162","doi":"10.22148/001c.17212","title":"Can GPT-3 Pass a Writer’s Turing Test?","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computability, Logic, AI Algorithms","field":"Computer Science","cited_by":205,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Computer science; Turing test; Heuristic; Field (mathematics); Grammar; Test (biology); Artificial intelligence; Scale (ratio); Language model; Natural language processing; Natural language; Turing; Language identification; Programming language; Linguistics; Mathematics","authors":[{"name":"Katherine Elkins","is_ca":false},{"name":"Jon Chun","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04380876233099833,"gpt":0.2608145662498129,"spread":0.2170058039188145,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02406939,0.001593007,0.001454692,0.001354577,0.004158891,0.007674352,0.003408726,0.004879539,0.02701957],"category_scores_gemma":[0.1270535,0.000697776,0.001777639,0.001220152,0.007660642,0.01798639,0.005734114,0.008590098,0.03745231],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002433242,"about_ca_system_score_gemma":0.005002933,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002795217,"about_ca_topic_score_gemma":0.002339252,"domain_scores_codex":[0.9858576,0.00573923,0.0007556164,0.002436464,0.003993981,0.001217148],"domain_scores_gemma":[0.9410293,0.03246045,0.001184489,0.01137457,0.01095422,0.002997007],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0003141781,0.0001318508,0.00160174,0.0003354937,0.00009748532,0.0002603903,0.0007503344,0.00217276,0.0005613537,0.2166682,0.5968908,0.1802154],"study_design_scores_gemma":[0.0001388323,0.0002251583,0.0008551439,0.0004568344,0.00005579696,0.0004934738,0.0004926056,0.009401583,0.002623856,0.4408433,0.5442974,0.0001161065],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.02145346,0.01379599,0.2102687,0.5001205,0.02971779,0.0005170859,0.003859173,0.01559705,0.2046701],"genre_scores_gemma":[0.3295205,0.01292814,0.3311094,0.1180474,0.01346921,0.002467877,0.009113978,0.01431151,0.169032],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02701957,"threshold_uncertainty_score":0.1272926,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2401667852","doi":"10.22148/16.004","title":"The Science of Culture? Social Computing, Digital Humanities and Cultural Analytics","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":110,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Cultural analytics; Analytics; Data science; Digital humanities; Digital culture; Visualization; Cultural analysis; Computer science; Cultural studies; Software; Sociology; World Wide Web; Media studies; Social science; Semantic analytics; Anthropology; The Internet; Artificial intelligence","authors":[{"name":"Лев Манович","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07116406461840949,"gpt":0.2735665621756626,"spread":0.2024024975572531,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.004368312,0.0009094948,0.0009105717,0.004528767,0.00463132,0.0196482,0.001098318,0.003177535,0.008426353],"category_scores_gemma":[0.01111274,0.0003839383,0.0004860508,0.006704397,0.04008276,0.03195919,0.005671844,0.005651542,0.001437653],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004510246,"about_ca_system_score_gemma":0.005165994,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007160171,"about_ca_topic_score_gemma":0.005422634,"domain_scores_codex":[0.9963504,0.002336074,0.0001137234,0.0003905484,0.0004915623,0.0003176933],"domain_scores_gemma":[0.9915946,0.004979649,0.0005553883,0.0008024415,0.0009669357,0.001100949],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00002397585,0.00003141729,0.002465065,0.0003076815,0.00003035755,0.00009139819,0.01070693,0.0002463408,0.0001411518,0.9004937,0.02790548,0.05755652],"study_design_scores_gemma":[0.000007119792,0.00001965189,0.001199645,0.0008365964,0.00001077773,0.00008957741,0.01643085,0.0005837617,0.000194788,0.8438148,0.1367854,0.00002699724],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.03723991,0.1360797,0.03952969,0.5709464,0.009926395,0.0000719388,0.0003574123,0.0002700004,0.2055785],"genre_scores_gemma":[0.8328787,0.09009933,0.01228276,0.0316761,0.009347059,0.0002017311,0.0002424341,0.0002597148,0.02301212],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9953687,"threshold_uncertainty_score":0.0327242,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2786709099","doi":"10.22148/16.019","title":"The Transformation of Gender in English-Language Fiction","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Literature: history, themes, analysis","field":"Arts and Humanities","cited_by":95,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"History; Literature; English language; Linguistics; Psychology; Sociology; Gender studies; Art; Philosophy","authors":[{"name":"Ted Underwood","is_ca":false},{"name":"David Bamman","is_ca":false},{"name":"Sabrina Lee","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02661860338798629,"gpt":0.2437616383731717,"spread":0.2171430349851854,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00224424,0.0002321747,0.0001965811,0.001616772,0.004532309,0.007356136,0.0005452933,0.0008796285,0.004440624],"category_scores_gemma":[0.008677143,0.0001130367,0.0001545319,0.001199556,0.01520042,0.005426534,0.00325073,0.00135788,0.0004492319],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002985787,"about_ca_system_score_gemma":0.0009795481,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002304891,"about_ca_topic_score_gemma":0.003777534,"domain_scores_codex":[0.9973876,0.001728114,0.00005620206,0.0001753044,0.0004198241,0.0002329071],"domain_scores_gemma":[0.9960172,0.002851075,0.000311693,0.0002548627,0.000384633,0.0001806346],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.00007241009,0.00002032538,0.002797241,0.0001876211,0.000007310161,0.0006772814,0.5510625,0.00009923891,0.001363351,0.402281,0.004368297,0.03706354],"study_design_scores_gemma":[0.000009664172,0.00008086853,0.009420509,0.0006479198,0.00001434811,0.001446382,0.2628715,0.0003418939,0.002919143,0.07036085,0.6518527,0.00003426459],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4944825,0.01216735,0.009435805,0.01668744,0.001472417,0.00003782787,0.0001617185,0.0000737294,0.4654812],"genre_scores_gemma":[0.9856541,0.001008308,0.0005004575,0.0005084791,0.0002239929,0.000008875504,0.00002901968,0.0000415183,0.01202538],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.007356136,"threshold_uncertainty_score":0.02166355,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3153865529","doi":"10.22148/001c.22221","title":"The Goodreads “Classics”: A Computational Study of Readers, Amazon, and Crowdsourced Amateur Criticism","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Games and Media","field":"Social Sciences","cited_by":55,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Amateur; Publishing; Criticism; Literary criticism; Reading (process); Point (geometry); Literature; Computer science; History; Sociology; Art; Linguistics; Philosophy","authors":[{"name":"Melanie Walsh","is_ca":false},{"name":"Maria Antoniak","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03654861629295647,"gpt":0.3283822056129931,"spread":0.2918335893200366,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009294368,0.0003714842,0.0006705479,0.004478297,0.004665364,0.008563478,0.001962522,0.002099192,0.003407767],"category_scores_gemma":[0.06348027,0.0004707513,0.0005161085,0.007653392,0.006228301,0.01055941,0.003396057,0.001550079,0.0006687543],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004511116,"about_ca_system_score_gemma":0.001443621,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01725219,"about_ca_topic_score_gemma":0.02316911,"domain_scores_codex":[0.9900323,0.007310573,0.0002498949,0.0008263422,0.00125957,0.0003214135],"domain_scores_gemma":[0.9015802,0.08332919,0.00665171,0.003065346,0.004035909,0.001337698],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"observational","study_design_scores_codex":[0.0004049429,0.0004173347,0.1438791,0.000705248,0.0001760516,0.001468098,0.7094734,0.002087173,0.001518691,0.05199159,0.02080817,0.06707019],"study_design_scores_gemma":[0.00009832567,0.0002745229,0.1140499,0.0006439635,0.0001646166,0.001249497,0.626546,0.03128183,0.00208979,0.04502601,0.1783537,0.0002218974],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9656935,0.001821374,0.003733305,0.005199348,0.00006471622,0.00009962147,0.0004641082,0.00009521253,0.02282881],"genre_scores_gemma":[0.9930264,0.0003982296,0.002583643,0.0005396773,0.00005884462,0.00007856457,0.0002766486,0.00006999029,0.002968131],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01725219,"threshold_uncertainty_score":0.04915392,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2942659792","doi":"10.22148/16.036","title":"Data Is the New What? Popular Metaphors & Professional Ethics in Emerging Data Culture","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":50,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Big data; Data science; Field (mathematics); Data collection; Computer science; Conjunction (astronomy); Engineering ethics; Knowledge management; Sociology; Social science; Data mining; Engineering","authors":[{"name":"Luke Stark","is_ca":false},{"name":"Anna Lauren Hoffmann","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2958110940572425,"gpt":0.4945049306874361,"spread":0.1986938366301936,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.01686513,0.000505157,0.0004096233,0.002130238,0.008873138,0.01275202,0.001271858,0.006346005,0.003544425],"category_scores_gemma":[0.02795758,0.0003405227,0.0003319327,0.003156112,0.08029877,0.01478009,0.005069643,0.009738375,0.001009953],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00507718,"about_ca_system_score_gemma":0.004673246,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003211223,"about_ca_topic_score_gemma":0.002345779,"domain_scores_codex":[0.9744061,0.02075091,0.0007111587,0.001073965,0.002377523,0.0006804114],"domain_scores_gemma":[0.9792604,0.01443448,0.001294806,0.001502306,0.002341345,0.001166544],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000006823542,0.00000697486,0.00013525,0.00003485947,0.000001596335,0.00004660675,0.02605265,0.00003282029,0.00006130819,0.958564,0.01136562,0.00369148],"study_design_scores_gemma":[0.00001082938,0.00001400183,0.0002947486,0.0003186185,0.000003266605,0.0003565323,0.03007888,0.0004509114,0.0001631432,0.621304,0.3469795,0.00002549559],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.02653954,0.01854074,0.04926878,0.6194336,0.004589453,0.00007009048,0.0001566398,0.0001044459,0.2812967],"genre_scores_gemma":[0.870588,0.01314371,0.01921292,0.06782064,0.003111672,0.0003993278,0.00009532683,0.0002613043,0.02536709],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9911268,"threshold_uncertainty_score":0.08919245,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2587326830","doi":"10.22148/16.006","title":"There Will Be Numbers","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":40,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Realm; TRACE (psycholinguistics); Mythology; The Internet; Point (geometry); Social media; Natural (archaeology); Computer science; Sociology; Epistemology; Data science; History; World Wide Web; Literature; Art; Philosophy; Mathematics; Linguistics; Archaeology","authors":[{"name":"Andrew Piper","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06962991266336825,"gpt":0.2568047155385784,"spread":0.1871748028752102,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005445322,0.001024469,0.0009978936,0.001334511,0.005629227,0.01428645,0.001670301,0.004007847,0.1691594],"category_scores_gemma":[0.02335223,0.0005147544,0.0007053014,0.001313422,0.007051592,0.01823346,0.006185936,0.008903798,0.09684685],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002642271,"about_ca_system_score_gemma":0.003169374,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001279588,"about_ca_topic_score_gemma":0.001896376,"domain_scores_codex":[0.995213,0.001180702,0.0002293369,0.001250303,0.001632504,0.0004941377],"domain_scores_gemma":[0.9887504,0.003327971,0.0006019514,0.002082065,0.002982147,0.00225546],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001378399,0.00005525414,0.0009906617,0.0001992087,0.00001482655,0.000122814,0.0008202581,0.00009965028,0.0004077891,0.5882481,0.3234496,0.08545411],"study_design_scores_gemma":[0.000007814464,0.00002013165,0.0001846393,0.0001101519,0.000003405007,0.00007045975,0.0003155022,0.00003367403,0.0001046661,0.06390943,0.9352267,0.00001339377],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.004955914,0.01800595,0.01595139,0.2106316,0.04145603,0.0001780678,0.001371987,0.0007108869,0.7067382],"genre_scores_gemma":[0.0668666,0.01801089,0.02098082,0.06134608,0.01377393,0.0006001357,0.001309389,0.001087016,0.8160251],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1691594,"threshold_uncertainty_score":0.5658946,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2979474259","doi":"10.22148/001c.13147","title":"NovelTM Datasets for English-Language Fiction, 1700-2009","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":37,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"English language; Computer science; Non-fiction; Resilience (materials science); Fragility; History; Linguistics; Literature; Art; Philosophy","authors":[{"name":"Ted Underwood","is_ca":false},{"name":"Patrick Kimutis","is_ca":false},{"name":"Jessica Witte","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07261985779052788,"gpt":0.2707456352499021,"spread":0.1981257774593742,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001082663,0.001854397,0.001063992,0.007544381,0.001751915,0.003404704,0.002586758,0.002631814,0.03489392],"category_scores_gemma":[0.006334946,0.0003854427,0.001343588,0.004971207,0.000735118,0.002309595,0.002867523,0.001647198,0.05911956],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001573884,"about_ca_system_score_gemma":0.001673127,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01305211,"about_ca_topic_score_gemma":0.03920003,"domain_scores_codex":[0.9989505,0.0001721737,0.0001450552,0.0002490881,0.0003351029,0.0001481003],"domain_scores_gemma":[0.997799,0.0005188849,0.0001980023,0.0005229645,0.0006991478,0.0002619585],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001250034,0.00008236212,0.001570753,0.001079916,0.00004547193,0.0001955581,0.0001435878,0.0004782292,0.0004629069,0.0008404189,0.9788994,0.01607651],"study_design_scores_gemma":[0.0001155763,0.00005495976,0.007310877,0.0004472501,0.00004741715,0.0003951133,0.0004009719,0.001959451,0.001146839,0.001085632,0.986987,0.00004875675],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.003263062,0.001352997,0.0004398104,0.0003991445,0.0004168937,0.0001017118,0.9877356,0.001400343,0.004890524],"genre_scores_gemma":[0.002547784,0.0002982047,0.001243696,0.00007949785,0.00009991197,0.0001466183,0.9927004,0.00008853593,0.002795345],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.03489392,"threshold_uncertainty_score":0.1167318,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2968217872","doi":"10.22148/16.044","title":"Difficult Heritage and the Complexities of Indigenous Data","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Cultural Heritage Management and Preservation","field":"Arts and Humanities","cited_by":31,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Indigenous; Geography; History; Data science; Computer science; Biology","authors":[{"name":"Jennifer Guiliano","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1595948908674861,"gpt":0.2662263728683407,"spread":0.1066314820008546,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.1004846,0.0004816665,0.0008874443,0.006103018,0.01030791,0.0168954,0.003859585,0.002827982,0.006185575],"category_scores_gemma":[0.2772132,0.00123877,0.0007772208,0.009790859,0.03156365,0.02069365,0.01377689,0.007868889,0.001226244],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005060899,"about_ca_system_score_gemma":0.009382963,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01738834,"about_ca_topic_score_gemma":0.02265887,"domain_scores_codex":[0.8954415,0.07238266,0.005038069,0.008201421,0.01682054,0.002115868],"domain_scores_gemma":[0.6514534,0.2737394,0.01517393,0.04157202,0.01501296,0.003048305],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.0001674038,0.00009088022,0.02510556,0.001143861,0.0001108016,0.00110798,0.1424645,0.001152627,0.0009160169,0.7514976,0.02144465,0.05479817],"study_design_scores_gemma":[0.00003485012,0.00004203746,0.01427508,0.001884433,0.00005837443,0.001661249,0.121825,0.00218867,0.001418163,0.5312231,0.3252639,0.0001250603],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1933215,0.01165135,0.3571228,0.2855203,0.004031814,0.001530938,0.009980615,0.0004513932,0.1363893],"genre_scores_gemma":[0.7780029,0.004098375,0.1698396,0.02445434,0.00300517,0.002680908,0.005139269,0.0005382952,0.01224122],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.1004846,"threshold_uncertainty_score":0.5314197,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2587380683","doi":"10.22148/16.010","title":"Understanding Gender and Character Agency in the 19th Century Novel","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Folklore, Mythology, and Literature Studies","field":"Arts and Humanities","cited_by":28,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Character (mathematics); Agency (philosophy); Late 19th century; Art; Literature; History; Aesthetics; Sociology; Social science; Period (music); Mathematics","authors":[{"name":"Matthew L. Jockers","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1580922718877203,"gpt":0.2791555401300548,"spread":0.1210632682423346,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002045096,0.0003046197,0.0003068051,0.001228654,0.00573819,0.01038796,0.0005983348,0.001088146,0.004843497],"category_scores_gemma":[0.002672687,0.000225726,0.0002474203,0.0008847184,0.02245136,0.007515459,0.003433487,0.001645637,0.0004183703],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004935584,"about_ca_system_score_gemma":0.001092793,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003163169,"about_ca_topic_score_gemma":0.006232449,"domain_scores_codex":[0.9984174,0.0009713569,0.00004350918,0.0001962172,0.0001506795,0.0002208489],"domain_scores_gemma":[0.9986641,0.0006997019,0.0002494897,0.000117977,0.0001085403,0.0001600701],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.00006152484,0.00003443702,0.001410641,0.00004829071,0.000008092568,0.0003963588,0.3350092,0.00008403981,0.0003461411,0.65484,0.001503022,0.006258205],"study_design_scores_gemma":[0.00003974968,0.0000744829,0.00646623,0.0003884707,0.00002329818,0.0009557968,0.4387838,0.0004728044,0.001165365,0.2270299,0.3245594,0.00004066396],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.4106264,0.004380171,0.004075895,0.007346784,0.0003282254,0.00002355168,0.00007624983,0.00002878428,0.573114],"genre_scores_gemma":[0.9879981,0.0005824767,0.0003404382,0.0001823105,0.0000544271,0.00001179185,0.00001450196,0.00002488572,0.01079094],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01038796,"threshold_uncertainty_score":0.03581029,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2500585446","doi":"10.22148/16.005","title":"The Life Cycles of Genres","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":27,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"University of Illinois at Urbana-Champaign","keywords":"Literature; Haven; Literary genre; Safe haven; Art; Aesthetics; History; Sociology; Linguistics; Philosophy; Mathematics; Economics","authors":[{"name":"Ted Underwood","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07790262750403741,"gpt":0.2603709562125727,"spread":0.1824683287085352,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006681395,0.0004626582,0.0004440332,0.006986684,0.005952666,0.02031223,0.001520366,0.001864939,0.0139282],"category_scores_gemma":[0.03393526,0.0007210929,0.000570238,0.004235027,0.01602458,0.02346989,0.005799675,0.002252348,0.003690344],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006026666,"about_ca_system_score_gemma":0.002835175,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002998513,"about_ca_topic_score_gemma":0.002339517,"domain_scores_codex":[0.9909276,0.004760717,0.0003820968,0.001188845,0.002069795,0.0006708954],"domain_scores_gemma":[0.9790568,0.009187173,0.001814309,0.003209424,0.004158628,0.002573648],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004465408,0.00002134283,0.002050328,0.00007681745,0.000009468954,0.0001334799,0.02755705,0.0002194281,0.0002958348,0.9311665,0.00746057,0.03096459],"study_design_scores_gemma":[0.00001432186,0.00002232003,0.002537711,0.0001744256,0.000007105567,0.0002248786,0.01689478,0.0007771013,0.0002665686,0.7071232,0.2719285,0.00002898569],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1524135,0.01010001,0.0508788,0.02283351,0.001050322,0.0003899269,0.001316946,0.0003788474,0.7606381],"genre_scores_gemma":[0.8990671,0.003835594,0.02000088,0.001340935,0.0005923111,0.0004743001,0.001027527,0.0006517678,0.07300975],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02031223,"threshold_uncertainty_score":0.0465945,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3117881903","doi":"10.22148/001c.18509","title":"Representing Race and Ethnicity in American Fiction, 1789-1920","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Culinary Culture and Tourism","field":"Agricultural and Biological Sciences","cited_by":23,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Racialization; Ethnic group; Writ; Race (biology); Gender studies; Representation (politics); Sociology; The Imaginary; Historicity (philosophy); Psychology; Political science; Anthropology","authors":[{"name":"Mark Algee‐Hewitt","is_ca":false},{"name":"Judith D. R. Porter","is_ca":false},{"name":"Hannah Walser","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04070182394424551,"gpt":0.2745896654440442,"spread":0.2338878414997987,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002788135,0.0003201642,0.0002368237,0.005313626,0.009177702,0.007186732,0.0005362045,0.0009996331,0.002820148],"category_scores_gemma":[0.005656759,0.0001944194,0.0001354772,0.004106909,0.01012614,0.004169916,0.002508269,0.001477245,0.0004973742],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007065444,"about_ca_system_score_gemma":0.002013574,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01766041,"about_ca_topic_score_gemma":0.03868734,"domain_scores_codex":[0.9988084,0.0005995027,0.00004739402,0.0001486027,0.0002754962,0.0001205958],"domain_scores_gemma":[0.996064,0.002675656,0.0002935941,0.0002055636,0.0005929807,0.0001681808],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.00008776444,0.00002636297,0.003318949,0.0002074002,0.00000962497,0.0006154689,0.8489372,0.00008240429,0.001162497,0.09732262,0.02027176,0.02795796],"study_design_scores_gemma":[0.000005758602,0.00001681274,0.009910179,0.0003999927,0.00001353688,0.0004420084,0.2574281,0.0001656609,0.001159605,0.00765697,0.7227774,0.00002410426],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7081835,0.01519512,0.002396639,0.01772111,0.00233409,0.00004803448,0.001641072,0.0001180872,0.2523623],"genre_scores_gemma":[0.9712632,0.004134051,0.001275448,0.001219654,0.0007072986,0.00004243531,0.0005636687,0.0001252707,0.02066903],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01766041,"threshold_uncertainty_score":0.05126363,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3183176260","doi":"10.22148/001c.25943","title":"The Measure of the Archive: The Robustness of Network Analysis in Early Modern Correspondence","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social Capital and Networks","field":"Social Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"Arts and Humanities Research Council","keywords":"Robustness (evolution); Computer science; Network analysis; Missing data; Transitive relation; Centrality; Data mining; Data loss; Data science; Statistics; Machine learning; Mathematics","authors":[{"name":"Yann Ryan","is_ca":false},{"name":"Sebastian E. Ahnert","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02610744377729758,"gpt":0.2901900606362046,"spread":0.264082616858907,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.02452921,0.001026121,0.0009988962,0.006193304,0.002440575,0.005038984,0.002057727,0.001969016,0.002467423],"category_scores_gemma":[0.2030231,0.0006170718,0.001213613,0.005022339,0.005170016,0.01002576,0.003302447,0.003066391,0.0007569556],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002945951,"about_ca_system_score_gemma":0.001231628,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006567186,"about_ca_topic_score_gemma":0.004241724,"domain_scores_codex":[0.9852043,0.007916361,0.0007543215,0.002881698,0.002779394,0.0004639592],"domain_scores_gemma":[0.798039,0.1443181,0.01517003,0.03024225,0.01060208,0.001628598],"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":[0.002021241,0.0002834217,0.1634563,0.001451304,0.002150613,0.0008400448,0.009530736,0.4426099,0.008876717,0.1220831,0.01327565,0.233421],"study_design_scores_gemma":[0.0000947957,0.0006136869,0.06447154,0.0007507964,0.0005079629,0.001354913,0.003743155,0.5970305,0.01648701,0.2917602,0.02286473,0.0003207515],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5884904,0.003826068,0.3773448,0.005004711,0.0005175893,0.0002923067,0.00293636,0.001705975,0.01988184],"genre_scores_gemma":[0.9500989,0.0004792188,0.04581818,0.0002915512,0.0001731708,0.000126651,0.001383455,0.0003616714,0.001267269],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9754708,"threshold_uncertainty_score":0.1297244,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2586621921","doi":"10.22148/16.007","title":"Linguistic Markers of Status in Food Culture: Bourdieu’s Distinction in a Menu Corpus","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Culinary Culture and Tourism","field":"Agricultural and Biological Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Realm; TRACE (psycholinguistics); Mythology; Point (geometry); Sociology; The Internet; Social media; Linguistics; Epistemology; Literature; Computer science; History; Art; Philosophy; World Wide Web; Mathematics; Archaeology","authors":[{"name":"Dan Jurafsky","is_ca":false},{"name":"Victor Chahuneau","is_ca":false},{"name":"Bryan Routledge","is_ca":false},{"name":"Noah A. Smith","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01991263327032837,"gpt":0.2467437038893575,"spread":0.2268310706190292,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002624847,0.0001911798,0.0002803531,0.002973108,0.003353721,0.003101233,0.0005456564,0.0005543951,0.004371238],"category_scores_gemma":[0.007038263,0.0002555982,0.0001362209,0.003404445,0.004903262,0.003426356,0.00259664,0.001290936,0.00044456],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001851148,"about_ca_system_score_gemma":0.0006360296,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007094752,"about_ca_topic_score_gemma":0.01321222,"domain_scores_codex":[0.9986975,0.0009492809,0.0000560939,0.0001340469,0.00009615531,0.00006701331],"domain_scores_gemma":[0.9951014,0.003913038,0.0001362738,0.000416674,0.0003268018,0.0001057895],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"observational","study_design_scores_codex":[0.0004214823,0.00006612385,0.01467456,0.0001997849,0.00002478332,0.0008435287,0.462999,0.0002590464,0.003555006,0.4123175,0.006334984,0.09830424],"study_design_scores_gemma":[0.00007612743,0.0001028813,0.07120907,0.0008142896,0.0000641426,0.00179167,0.4028215,0.003764285,0.006603994,0.2481622,0.264464,0.0001259312],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8390846,0.001733224,0.02112032,0.004490919,0.000180442,0.00004373536,0.0007021612,0.00009252367,0.132552],"genre_scores_gemma":[0.9932028,0.0001549123,0.003934069,0.00008902374,0.00001797992,0.00002704222,0.0001710532,0.00006272778,0.002340361],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.007094752,"threshold_uncertainty_score":0.01462322,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3008187175","doi":"10.22148/001c.12049","title":"The Social Lives of Books: Reading Victorian Literature on Goodreads","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Games and Media","field":"Social Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Popularity; Reading (process); Syllabus; Victorian literature; Sociology; Plot (graphics); Order (exchange); Media studies; History; Psychology; Literature; Pedagogy; Art; Social psychology; Political science; Law","authors":[{"name":"Karen Bourrier","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03402950775756183,"gpt":0.3202783749835676,"spread":0.2862488672260058,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00177712,0.0002516082,0.0005554291,0.006017361,0.005636622,0.007273829,0.0008778957,0.0009852983,0.005159241],"category_scores_gemma":[0.01789792,0.0004890675,0.0001802521,0.006671917,0.004521009,0.004708201,0.004254621,0.001557429,0.001247925],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004396999,"about_ca_system_score_gemma":0.001829026,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05564287,"about_ca_topic_score_gemma":0.1425021,"domain_scores_codex":[0.9970031,0.001193477,0.0001345191,0.0003384123,0.001011563,0.0003189074],"domain_scores_gemma":[0.9823673,0.009462341,0.003114733,0.0008298135,0.00286058,0.001365254],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"observational","study_design_scores_codex":[0.0000287532,0.00001167603,0.01608734,0.0001424618,0.000008479949,0.000479444,0.9676012,0.0000133317,0.0008804686,0.001132132,0.001426405,0.01218836],"study_design_scores_gemma":[0.000001878442,0.00002756141,0.1035928,0.0003758785,0.000009376977,0.000438238,0.8064708,0.00007954415,0.0005457136,0.0005755585,0.08783907,0.00004369826],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9632881,0.001098736,0.000625464,0.001013836,0.0000471364,0.00002636737,0.0004861879,0.00003113826,0.03338303],"genre_scores_gemma":[0.9812194,0.0007956077,0.0005976507,0.0003095207,0.00004985349,0.00003365403,0.0003962205,0.0001206229,0.0164775],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.05564287,"threshold_uncertainty_score":0.110638,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2913116882","doi":"10.22148/16.032","title":"Social Characters: The Hierarchy of Gender in Contemporary English-Language Fiction","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Media, Gender, and Advertising","field":"Social Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Hollywood; Hierarchy; Variety (cybernetics); Decentralization; Sociology; History; Political science; Computer science","authors":[{"name":"Eve Kraicer","is_ca":false},{"name":"Andrew Piper","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08365479870504956,"gpt":0.3525469657274379,"spread":0.2688921670223884,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001908939,0.0002492981,0.0002270685,0.00311448,0.006207846,0.00599523,0.0004503971,0.0006101455,0.004858888],"category_scores_gemma":[0.006089574,0.0001737905,0.0001134669,0.001515053,0.01171712,0.003006768,0.003166643,0.001101263,0.0004957315],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002531945,"about_ca_system_score_gemma":0.0009494885,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01134167,"about_ca_topic_score_gemma":0.01993322,"domain_scores_codex":[0.9976403,0.00131327,0.00007422165,0.0002542716,0.0004365974,0.0002814675],"domain_scores_gemma":[0.9966257,0.001915764,0.0004830126,0.0001299231,0.0004051064,0.0004404308],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"observational","study_design_scores_codex":[0.00007731101,0.00001802684,0.01524526,0.00007966447,0.0000071144,0.000421197,0.8902111,0.00002692177,0.001019142,0.06750359,0.002338994,0.02305166],"study_design_scores_gemma":[0.00001186621,0.00007597791,0.07770878,0.0003843157,0.0000156309,0.001105099,0.7379732,0.0002803067,0.0007114189,0.021124,0.1605731,0.00003636387],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7850571,0.002420856,0.00212408,0.004353389,0.0001717206,0.00003261593,0.0001021372,0.00003421561,0.2057039],"genre_scores_gemma":[0.9944348,0.0003285873,0.0001973376,0.0002337515,0.00004450297,0.000007961119,0.00002388494,0.00001927985,0.00471008],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01134167,"threshold_uncertainty_score":0.0225513,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2963370929","doi":"10.22148/16.039","title":"The Carceral Archive: Documentary Records, Narrative Construction, and Predictive Risk Assessment","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Law in Society and Culture","field":"Social Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Scholarship; Narrative; Race (biology); Criminology; Sociology; Oral history; Law; Political science; History; Gender studies; Anthropology; Art; Literature","authors":[{"name":"Tonia Sutherland","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.007586517233021723,"gpt":0.302421839749186,"spread":0.2948353225161642,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01293697,0.000468899,0.0003316752,0.02164748,0.003488427,0.01406632,0.002481455,0.001400527,0.01770418],"category_scores_gemma":[0.0910303,0.0004897565,0.0002567481,0.02448234,0.003062801,0.01290369,0.004783249,0.0015365,0.005428738],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002905583,"about_ca_system_score_gemma":0.007964688,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0305122,"about_ca_topic_score_gemma":0.03851943,"domain_scores_codex":[0.992918,0.003406589,0.0007603326,0.0006052325,0.002104031,0.0002057136],"domain_scores_gemma":[0.9088196,0.0391037,0.009358634,0.02355074,0.01723722,0.001930238],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"qualitative","study_design_scores_codex":[0.0002506972,0.000154425,0.02954007,0.001175062,0.0000576538,0.0004176814,0.05138902,0.001129252,0.0004561812,0.1294564,0.2177223,0.5682513],"study_design_scores_gemma":[0.00002617868,0.00004504805,0.01831145,0.002224588,0.00004862943,0.0004027874,0.04491599,0.003605672,0.001045226,0.03289403,0.896374,0.0001063598],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.173723,0.02140701,0.1329932,0.03068995,0.002002281,0.002701661,0.1486224,0.007881095,0.4799794],"genre_scores_gemma":[0.6487433,0.01495076,0.1996128,0.001005462,0.001234808,0.001819831,0.06856242,0.002409584,0.06166101],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.0305122,"threshold_uncertainty_score":0.06841803,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2905726793","doi":"10.22148/16.030","title":"Critical Search: A Procedure for Guided Reading in Large-Scale Textual Corpora","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Reading (process); Computer science; Scale (ratio); Information retrieval; Natural language processing; Artificial intelligence; Linguistics; Geography; Philosophy; Cartography","authors":[{"name":"Jo Guldi","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0522741677462495,"gpt":0.3675912135698836,"spread":0.3153170458236341,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02680897,0.003155011,0.002089575,0.01796972,0.00532086,0.006743511,0.003541521,0.002251031,0.06849349],"category_scores_gemma":[0.1220755,0.00238242,0.002421336,0.0134338,0.004461799,0.00823401,0.01011567,0.00359151,0.02723692],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002258291,"about_ca_system_score_gemma":0.009621345,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004828549,"about_ca_topic_score_gemma":0.01235764,"domain_scores_codex":[0.980667,0.009337437,0.003499361,0.003136174,0.002910244,0.000449863],"domain_scores_gemma":[0.9090465,0.06456757,0.003455837,0.01029953,0.01144255,0.001188052],"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":[0.0007503851,0.0002468888,0.001907921,0.005095237,0.0003074666,0.001212249,0.02700516,0.001767977,0.01591217,0.08144737,0.2851336,0.5792136],"study_design_scores_gemma":[0.0005189066,0.0002040883,0.003441037,0.001802304,0.0002422044,0.001276067,0.01696797,0.03772385,0.02307173,0.2710851,0.6430191,0.0006477128],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004506723,0.0003631263,0.9176248,0.001276471,0.0004470203,0.005114101,0.01255363,0.0433385,0.01477557],"genre_scores_gemma":[0.01513982,0.0001850091,0.9639943,0.0002735906,0.0001293761,0.004572228,0.006637223,0.005000161,0.004068202],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.06849349,"threshold_uncertainty_score":0.2291335,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3123995752","doi":"10.22148/001c.18841","title":"Abolitionist Networks: Modeling Language Change in Nineteenth-Century Activist Newspapers","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social and Cultural Dynamics","field":"Social Sciences","cited_by":19,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"National Science Foundation","keywords":"Newspaper; Scholarship; Vanguard; Credence; Politics; History; Argument (complex analysis); Sociology; Media studies; Political science; Law; Computer science","authors":[{"name":"Sandeep Soni","is_ca":false},{"name":"Lauren Klein","is_ca":false},{"name":"Jacob Eisenstein","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04176687371627995,"gpt":0.339984818055101,"spread":0.298217944338821,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002277491,0.000558677,0.0004121209,0.003018766,0.0009113954,0.002115598,0.001089274,0.001607607,0.002814312],"category_scores_gemma":[0.01659913,0.0004502411,0.0009479225,0.003146248,0.001361424,0.003646361,0.001392737,0.001710688,0.0006226894],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00193062,"about_ca_system_score_gemma":0.0005011031,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03239597,"about_ca_topic_score_gemma":0.03343501,"domain_scores_codex":[0.9989948,0.0005282716,0.00004473894,0.0002708539,0.00007584781,0.00008543827],"domain_scores_gemma":[0.9885754,0.009886351,0.0007341509,0.0003090999,0.0003123399,0.0001827076],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","study_design_scores_codex":[0.0007757897,0.0006277427,0.1827641,0.0003838954,0.0004129859,0.001041592,0.01024333,0.659839,0.002529081,0.05351093,0.006503901,0.08136769],"study_design_scores_gemma":[0.00002163496,0.00002841081,0.008597782,0.00002769773,0.00003433281,0.00006192047,0.0008143916,0.9706166,0.000239041,0.01681869,0.002722479,0.00001706639],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9367682,0.0009196744,0.05118422,0.001811134,0.00008049641,0.000115078,0.002521254,0.0003520857,0.00624789],"genre_scores_gemma":[0.9794302,0.0002949437,0.01552082,0.0001538962,0.00005196914,0.000175301,0.002309255,0.00006906006,0.001994609],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.03239597,"threshold_uncertainty_score":0.0644148,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4319604386","doi":"10.22148/001c.55592","title":"Should we really ‘hermeneutise’ the Digital Humanities? A plea for the epistemic productivity of a ‘cultural technique of flattening’ in the Humanities.","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":17,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Digital humanities; Hermeneutics; Humanities; Epistemology; Sociology; Philosophy","authors":[{"name":"Sybille Krämer","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1598366603311808,"gpt":0.3027010683654781,"spread":0.1428644080342973,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","sts"],"consensus_categories":[],"category_scores_codex":[0.02479535,0.001123899,0.001110098,0.003894401,0.008896286,0.03150945,0.002548892,0.00928714,0.01191641],"category_scores_gemma":[0.06326448,0.0007312577,0.0008449898,0.003885327,0.120134,0.06669638,0.01551073,0.01612875,0.0038852],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006722381,"about_ca_system_score_gemma":0.007407139,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005971476,"about_ca_topic_score_gemma":0.00456676,"domain_scores_codex":[0.9746756,0.01647847,0.0007779187,0.002716259,0.003951298,0.00140047],"domain_scores_gemma":[0.9461583,0.03248921,0.003177119,0.01170325,0.004289113,0.002182933],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00000980346,0.000009510461,0.0001566401,0.0000647975,0.000005348588,0.00002122211,0.006121176,0.000106839,0.00008238475,0.9804103,0.006598,0.006413969],"study_design_scores_gemma":[0.00001014246,0.000008746372,0.000151466,0.0002196571,0.000004133559,0.00006458412,0.00583494,0.0002737754,0.0002127606,0.8112347,0.181956,0.00002912281],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"commentary","genre_gemma":"commentary","genre_scores_codex":[0.01202959,0.02409665,0.1292667,0.6455571,0.005915417,0.00011232,0.0001632721,0.0004497284,0.1824092],"genre_scores_gemma":[0.692957,0.01931445,0.09412449,0.1181039,0.008919007,0.000615449,0.0002443145,0.00124215,0.06447931],"genre_candidate":"commentary","genre_consensus":"commentary","teacher_disagreement_score":0.9911037,"threshold_uncertainty_score":0.1311318,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2962933257","doi":"10.22148/16.043","title":"Visual Style in Two Network Era Sitcoms","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Media Studies and Communication","field":"Social Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Scholarship; Framing (construction); Style (visual arts); Computer science; Sociology; Political science; Geography; Visual arts; Art","authors":[{"name":"Taylor Arnold","is_ca":false},{"name":"Lauren Tilton","is_ca":false},{"name":"Annie Berke","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03048623578256494,"gpt":0.3704559229220717,"spread":0.3399696871395068,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006203989,0.0002994638,0.000192363,0.001749895,0.001201222,0.005134469,0.0004354059,0.0006320974,0.02058627],"category_scores_gemma":[0.005517426,0.0001482869,0.0001688468,0.001500006,0.001235646,0.002663026,0.001966883,0.0006329215,0.001831963],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001044856,"about_ca_system_score_gemma":0.0003895357,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003812313,"about_ca_topic_score_gemma":0.00655011,"domain_scores_codex":[0.9995584,0.0001523309,0.00001618705,0.00008013283,0.0001195928,0.0000732648],"domain_scores_gemma":[0.9979741,0.0006943463,0.0002022552,0.000301771,0.0004237344,0.000403793],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.002847653,0.0005016166,0.1051093,0.0004796827,0.0001313624,0.002051275,0.1057048,0.009537425,0.01446701,0.3749462,0.05949596,0.3247277],"study_design_scores_gemma":[0.0002603065,0.0006157696,0.2398709,0.0006082936,0.0001710625,0.00146836,0.1307357,0.1054113,0.005740133,0.1682213,0.3466869,0.0002101135],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7146246,0.0003891077,0.009521519,0.001057756,0.0003177655,0.00006784758,0.000722808,0.0004140158,0.2728845],"genre_scores_gemma":[0.9809878,0.00009917781,0.002585071,0.0000941492,0.00004850363,0.00002686965,0.000450788,0.0001472831,0.01556044],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02058627,"threshold_uncertainty_score":0.06886798,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4247740782","doi":"10.22148/16.011","title":"Fictionality","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Narrative Theory and Analysis","field":"Arts and Humanities","cited_by":15,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Narrative; Literature; Philosophy; Art","authors":[{"name":"Andrew Piper","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05804555801263898,"gpt":0.2682156559321219,"spread":0.2101700979194829,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002385056,0.0005606881,0.0003805693,0.003142171,0.003967381,0.01010921,0.0008753847,0.001746449,0.03487546],"category_scores_gemma":[0.01661249,0.0002022087,0.0003600911,0.001802651,0.01372145,0.00850075,0.004146175,0.002188026,0.005180015],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002791211,"about_ca_system_score_gemma":0.001446945,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001230444,"about_ca_topic_score_gemma":0.001052685,"domain_scores_codex":[0.9960008,0.001800605,0.0001866436,0.0005963809,0.001145547,0.0002699256],"domain_scores_gemma":[0.9917373,0.004814436,0.0005836374,0.001350841,0.00113627,0.0003775464],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001329852,0.00000898862,0.0003198299,0.0002142364,0.000007697148,0.0001478841,0.01094075,0.00008273589,0.0002547643,0.9482661,0.01611799,0.02362565],"study_design_scores_gemma":[0.000004715616,0.00001673496,0.0005503802,0.0003248468,0.000007825729,0.0005934474,0.003175128,0.0001213432,0.000399642,0.1635179,0.8312746,0.00001344251],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.01337788,0.007829034,0.01397258,0.01387248,0.00185917,0.00008953374,0.0005405857,0.0001747944,0.948284],"genre_scores_gemma":[0.8087724,0.008213574,0.009983752,0.006484411,0.002869779,0.000201845,0.001377505,0.0003635751,0.1617333],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.03487546,"threshold_uncertainty_score":0.11667,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2985141934","doi":"10.22148/16.048","title":"A Shared Task for the Digital Humanities Chapter 1: Introduction to Annotation, Narrative Levels and Shared Tasks","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":14,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Digital humanities; Annotation; Narrative; Task (project management); Field (mathematics); Computer science; Humanities; World Wide Web; Artificial intelligence; Literature; Philosophy; Art; Engineering; Mathematics","authors":[{"name":"Nils Reiter","is_ca":false},{"name":"Marcus Willand","is_ca":false},{"name":"Evelyn Gius","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0752353978742625,"gpt":0.2643746749599136,"spread":0.1891392770856511,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01427153,0.001300139,0.001143779,0.004185178,0.007729395,0.01333561,0.003892546,0.004763374,0.02718707],"category_scores_gemma":[0.02596805,0.001434968,0.001351419,0.005831302,0.006988809,0.02309825,0.01732752,0.005201131,0.01586731],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003730508,"about_ca_system_score_gemma":0.007881817,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005378689,"about_ca_topic_score_gemma":0.006868643,"domain_scores_codex":[0.9862877,0.006445365,0.001363784,0.002258447,0.002999013,0.0006457079],"domain_scores_gemma":[0.9834418,0.006359776,0.0007049288,0.005344437,0.002881585,0.001267492],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0002198687,0.0001513211,0.001146266,0.001374801,0.00003319828,0.0002695666,0.03092879,0.002150332,0.005140747,0.6494197,0.08582208,0.2233433],"study_design_scores_gemma":[0.00002203003,0.00003965121,0.0008163316,0.0009940466,0.00001873889,0.000323585,0.004819789,0.005799655,0.00327868,0.2511884,0.7326143,0.00008487636],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.005705918,0.002282458,0.8781458,0.01201048,0.001141209,0.001386558,0.002537522,0.004386022,0.092404],"genre_scores_gemma":[0.08217233,0.001962863,0.8294266,0.001631519,0.0007328334,0.003012022,0.005930121,0.003885868,0.07124583],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02718707,"threshold_uncertainty_score":0.09094983,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3004295495","doi":"10.22148/001c.11829","title":"On the perceived complexity of literature. A response to Nan Z. Da","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Reductionism; Variance (accounting); Literary criticism; Epistemology; Sociology; Linguistics; Philosophy; Economics","authors":[{"name":"Fotis Jannidis","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07676143944045041,"gpt":0.3253034171408474,"spread":0.248541977700397,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01376581,0.00132163,0.001977023,0.002116005,0.01068772,0.01497194,0.003298161,0.02553937,0.006618226],"category_scores_gemma":[0.06551142,0.000915264,0.000903907,0.002908917,0.019528,0.0237937,0.01136221,0.05071031,0.002509311],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006815816,"about_ca_system_score_gemma":0.0061077,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009142978,"about_ca_topic_score_gemma":0.0118873,"domain_scores_codex":[0.9876373,0.005286698,0.0009586891,0.002041126,0.003280166,0.0007959881],"domain_scores_gemma":[0.8916463,0.08153684,0.004357274,0.002757768,0.01353036,0.006171548],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00003167514,0.00001587195,0.0003880365,0.0001345061,0.00001565631,0.0001913446,0.007828004,0.00004157794,0.0001324799,0.01744245,0.96646,0.007318384],"study_design_scores_gemma":[0.0000242783,0.00002288742,0.001133278,0.0008109937,0.00001188896,0.0006843568,0.01881252,0.0001620094,0.0001791888,0.02143895,0.9566278,0.0000917725],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"commentary","genre_scores_codex":[0.0003940121,0.003800369,0.000133015,0.9899424,0.004919623,0.000003761992,0.00002377548,0.00001033741,0.0007725911],"genre_scores_gemma":[0.01209759,0.004577415,0.0005123787,0.9684254,0.01065646,0.00004203373,0.00003017971,0.0000686082,0.00359008],"genre_candidate":"commentary","genre_consensus":"commentary","teacher_disagreement_score":0.02553937,"threshold_uncertainty_score":0.07280147,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2931020970","doi":"10.22148/16.028","title":"Historiography’s Two Voices: Data Infrastructure and History at Scale in the Oxford Dictionary of National Biography (ODNB)","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Philippine History and Culture","field":"Social Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Historiography; Biography; Scale (ratio); History; Classics; Literature; Media studies; Art history; Sociology; Art; Geography; Cartography; Archaeology","authors":[{"name":"Christopher Warren","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04836581654379901,"gpt":0.3157815727627241,"spread":0.2674157562189251,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.008667947,0.0002877114,0.0003251933,0.003732789,0.002647544,0.0148904,0.0007611075,0.001218607,0.01189008],"category_scores_gemma":[0.02931227,0.0004580688,0.0001440633,0.01215618,0.00934084,0.02464372,0.004648978,0.003320439,0.001742566],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004821695,"about_ca_system_score_gemma":0.005005967,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01243224,"about_ca_topic_score_gemma":0.01703307,"domain_scores_codex":[0.9952839,0.002905726,0.0004185554,0.0004985108,0.0006669172,0.0002263436],"domain_scores_gemma":[0.97939,0.01530692,0.001116708,0.002132514,0.001288949,0.0007649849],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.00004197173,0.000007504348,0.003388913,0.0003223907,0.000009504687,0.0001519522,0.07608505,0.0001421387,0.0001414899,0.7160186,0.1209472,0.08274333],"study_design_scores_gemma":[0.000004587283,0.000006156039,0.002203688,0.001013231,0.000004951613,0.00009347697,0.02627714,0.000158838,0.0001524362,0.06833294,0.9017346,0.0000179678],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.06128519,0.07260926,0.09148709,0.2651959,0.01006342,0.0002354657,0.01582465,0.00116119,0.4821379],"genre_scores_gemma":[0.7783541,0.04475364,0.048661,0.02230053,0.003554987,0.0005654411,0.00867063,0.001969166,0.09117047],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9973525,"threshold_uncertainty_score":0.04584104,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2587367367","doi":"10.22148/16.009","title":"Genre, Computation, and the Varieties of Twentieth-Century U.S. Fiction","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":13,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Literary fiction; Literature; Literary genre; Fiction theory; Popular fiction; Techno-thriller; History; World War II; Art; Literary criticism; Archaeology","authors":[{"name":"Matthew Wilkens","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03885955791862655,"gpt":0.2399455145696156,"spread":0.2010859566509891,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001478831,0.0002080784,0.0001681324,0.004093278,0.004188154,0.008494351,0.0003940788,0.0005062533,0.006341181],"category_scores_gemma":[0.006392492,0.000130576,0.0001684829,0.003785139,0.00782301,0.004167661,0.002126204,0.001385508,0.0005636361],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004467332,"about_ca_system_score_gemma":0.001016956,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01054464,"about_ca_topic_score_gemma":0.02285491,"domain_scores_codex":[0.9991666,0.0003259915,0.00005294436,0.0001091433,0.0002254753,0.0001199346],"domain_scores_gemma":[0.9970552,0.001601631,0.0003861203,0.0002574969,0.0003947048,0.0003048024],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.0001156204,0.00003597352,0.008393425,0.0001125797,0.00001107918,0.0002903198,0.09982475,0.0001407533,0.0003993427,0.7931846,0.0233842,0.07410748],"study_design_scores_gemma":[0.00002379959,0.00006456659,0.04695582,0.0007527726,0.00001394936,0.0006053303,0.05672237,0.0004383759,0.0006851883,0.156935,0.73677,0.00003281779],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.3152326,0.0223246,0.001601221,0.01612964,0.0010259,0.00001868576,0.0003481174,0.00008851164,0.6432307],"genre_scores_gemma":[0.9730357,0.004899678,0.0007510665,0.0007024357,0.0004347762,0.00001797444,0.0001916381,0.00007074105,0.01989608],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01054464,"threshold_uncertainty_score":0.03241295,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2898247218","doi":"10.22148/16.026","title":"Image Analytics and the Nineteenth-Century Illustrated Newspaper","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Photography and Visual Culture","field":"Arts and Humanities","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Newspaper; Sublime; Flood myth; History; Analytics; Art history; Media studies; Literature; Art; Computer science; Sociology; Data science; Archaeology","authors":[{"name":"Paul Fyfe","is_ca":false},{"name":"Qian Ge","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03466716520478621,"gpt":0.2718072904928382,"spread":0.237140125288052,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002047104,0.0003400468,0.00020658,0.008934707,0.002413372,0.01113186,0.000508557,0.001587323,0.008579212],"category_scores_gemma":[0.008255416,0.0003258463,0.0002226844,0.009646267,0.006164176,0.008509493,0.001631542,0.002125008,0.002247184],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007204555,"about_ca_system_score_gemma":0.001942676,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03299425,"about_ca_topic_score_gemma":0.05207391,"domain_scores_codex":[0.9984512,0.0004662142,0.00008854952,0.0001378311,0.000701531,0.0001546626],"domain_scores_gemma":[0.9904805,0.0064334,0.0006066623,0.000382341,0.00181675,0.0002804163],"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":[0.0001212604,0.00006169742,0.002165634,0.002385854,0.00005156326,0.0007190813,0.02947633,0.0001463793,0.0008066534,0.2344768,0.3075488,0.42204],"study_design_scores_gemma":[0.000005587768,0.000009932871,0.006024399,0.001568529,0.00001584673,0.0003452111,0.008310238,0.00008097689,0.0006182925,0.007720859,0.9752824,0.00001776139],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"review","genre_gemma":"empirical","genre_scores_codex":[0.02757438,0.6634194,0.001721576,0.07731471,0.002902629,0.00002423732,0.0005460998,0.0001995523,0.2262976],"genre_scores_gemma":[0.323803,0.5530174,0.002702091,0.01564153,0.006097092,0.00004854829,0.0008203768,0.0003485118,0.09752139],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.03299425,"threshold_uncertainty_score":0.06560439,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4313409859","doi":"10.22148/001c.57195","title":"From Concepts to Texts and Back: Operationalization as a Core Activity of Digital Humanities","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Operationalization; Realm; Epistemology; Process (computing); Construct (python library); Computer science; Sociology; Political science; Philosophy","authors":[{"name":"Axel Pichler","is_ca":false},{"name":"Nils Reiter","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.203411329081511,"gpt":0.4160787014155996,"spread":0.2126673723340886,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04579665,0.001767043,0.001180499,0.009632679,0.006996685,0.02460787,0.003438674,0.003843159,0.008071251],"category_scores_gemma":[0.07822306,0.001134969,0.001418797,0.01147999,0.1003684,0.06063425,0.01593588,0.008798849,0.001454145],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0102201,"about_ca_system_score_gemma":0.0103194,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006056285,"about_ca_topic_score_gemma":0.002661518,"domain_scores_codex":[0.9475803,0.0386041,0.00323611,0.00367276,0.005593982,0.001312721],"domain_scores_gemma":[0.9320339,0.04940997,0.00378627,0.008660753,0.004712197,0.001396946],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001419152,0.00001330881,0.0002587979,0.0001531329,0.000007647357,0.00007136341,0.03220145,0.0002337924,0.0002039199,0.9530539,0.00071509,0.01307349],"study_design_scores_gemma":[0.00001266576,0.00001877044,0.000294444,0.0004828596,0.000009943012,0.0001356924,0.02494897,0.0007728804,0.0005165152,0.9203677,0.05240854,0.0000309878],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03631973,0.009500409,0.7043635,0.04133306,0.001097969,0.0009919072,0.0003565187,0.0006572762,0.2053797],"genre_scores_gemma":[0.6449718,0.004249705,0.3349949,0.003444008,0.0005930242,0.001831749,0.000517579,0.0008343186,0.008562968],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.04579665,"threshold_uncertainty_score":0.2421987,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2895759845","doi":"10.22148/16.025","title":"Stable Random Projection: Lightweight, General-Purpose Dimensionality Reduction for Digitized Libraries","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"sort; Computer science; Dimensionality reduction; Digital library; Projection (relational algebra); Random projection; Reduction (mathematics); Computer graphics (images); World Wide Web; Multimedia; Information retrieval; Artificial intelligence; Mathematics; Algorithm; Art","authors":[{"name":"Benjamin M. Schmidt","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03697347895973706,"gpt":0.2909724707646574,"spread":0.2539989918049203,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001027125,0.001561209,0.001521961,0.001443528,0.0009465024,0.00213478,0.002163875,0.0009262979,0.007293351],"category_scores_gemma":[0.003302937,0.0006626668,0.001089506,0.00289494,0.0006142898,0.002947487,0.004411119,0.001531627,0.009864712],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004507294,"about_ca_system_score_gemma":0.001879141,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003115568,"about_ca_topic_score_gemma":0.006551013,"domain_scores_codex":[0.9983045,0.0003823849,0.00008499399,0.0002355917,0.0008234456,0.0001691217],"domain_scores_gemma":[0.9990871,0.0001355756,0.00006321789,0.0004058503,0.00023099,0.0000772503],"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":[0.0006427833,0.0004130719,0.001567728,0.0003205497,0.0001858522,0.0002864471,0.0002183508,0.03468489,0.02024862,0.01060384,0.06342649,0.8674014],"study_design_scores_gemma":[0.000133888,0.0001977356,0.001593996,0.00003826446,0.0000470796,0.0003879499,0.00020449,0.9261075,0.0202107,0.02744813,0.02354821,0.00008203308],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01318255,0.0005308755,0.9681756,0.0002802275,0.0001063739,0.0001783824,0.0009106837,0.01449612,0.002139163],"genre_scores_gemma":[0.1189873,0.0008409322,0.8637557,0.0002364884,0.000154016,0.0005222562,0.005203797,0.0009954777,0.00930399],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007293351,"threshold_uncertainty_score":0.02439868,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3003312070","doi":"10.22148/001c.11830","title":"Is there a text in my data? (Part 1): on counting words","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Computer science; Word (group theory); Context (archaeology); Simple (philosophy); Normalization (sociology); Natural language processing; Linguistics; Artificial intelligence; Epistemology; Philosophy; History; Sociology","authors":[{"name":"Michael Gavin","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1950699218308501,"gpt":0.292525711556765,"spread":0.09745578972591484,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010927,0.0008859095,0.0009406736,0.004421032,0.003118183,0.009427847,0.001344073,0.001987066,0.005289522],"category_scores_gemma":[0.0827159,0.0005810066,0.0008505905,0.007251428,0.01525944,0.02580109,0.004953692,0.00444024,0.002030061],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003323568,"about_ca_system_score_gemma":0.002321137,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003513196,"about_ca_topic_score_gemma":0.003369981,"domain_scores_codex":[0.9877512,0.007611741,0.0007156303,0.00177107,0.001925415,0.0002247883],"domain_scores_gemma":[0.9485883,0.04190818,0.002998591,0.002897517,0.003159814,0.0004476695],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00007956885,0.00004606479,0.006719489,0.0005687373,0.00004483612,0.0001351786,0.0151212,0.001007598,0.0006265532,0.773562,0.04968791,0.1524008],"study_design_scores_gemma":[0.00001291892,0.00004737208,0.004419804,0.001298776,0.0000290135,0.0004180164,0.01004736,0.006481671,0.001115687,0.7657818,0.2102691,0.00007843394],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.05606665,0.02819243,0.6805737,0.151506,0.006912997,0.0005159006,0.002550229,0.0006706977,0.07301131],"genre_scores_gemma":[0.5224733,0.02382638,0.3723469,0.03624285,0.01616304,0.001875952,0.003028542,0.001365267,0.02267762],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.010927,"threshold_uncertainty_score":0.05778825,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3091866587","doi":"10.22148/001c.17585","title":"Divergence and the Complexity of Difference in Text and Culture","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language and cultural evolution","field":"Social Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"John Templeton Foundation","keywords":"Divergence (linguistics); Operationalization; Metaphor; Rhetorical question; Darwin (ADL); Function (biology); Epistemology; Computer science; Linguistics; Construal level theory; Sociology; Philosophy; Social science; Evolutionary biology","authors":[{"name":"Kent K. Chang","is_ca":false},{"name":"Simon DeDeo","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06567579411856664,"gpt":0.3085290246590696,"spread":0.242853230540503,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01358716,0.0005350969,0.001070108,0.008728098,0.003007241,0.01268083,0.001508408,0.001694415,0.003436351],"category_scores_gemma":[0.1082032,0.0004726411,0.0007615685,0.008130609,0.02316382,0.02645265,0.008156106,0.002571795,0.000410923],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004595399,"about_ca_system_score_gemma":0.001480428,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00168041,"about_ca_topic_score_gemma":0.0009011543,"domain_scores_codex":[0.9742343,0.0129196,0.002245841,0.003784694,0.006101069,0.0007144568],"domain_scores_gemma":[0.8920442,0.08340787,0.008636674,0.007953231,0.006344718,0.001613339],"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":[0.000264131,0.00004789501,0.01616916,0.000845373,0.0002212918,0.0005814412,0.07855598,0.00322231,0.002304279,0.8136673,0.001516473,0.0826044],"study_design_scores_gemma":[0.00002862354,0.00005821852,0.01195184,0.0002350852,0.00004592841,0.0004134527,0.0168883,0.003731487,0.0008139197,0.9520684,0.01367209,0.00009255364],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6406915,0.01098568,0.2270398,0.01565089,0.0007057819,0.0002440078,0.001352901,0.0004168857,0.1029126],"genre_scores_gemma":[0.9787831,0.0008266862,0.01799427,0.0004046886,0.0001936012,0.0001619034,0.0002688133,0.0000973928,0.001269479],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01358716,"threshold_uncertainty_score":0.07185662,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2909609490","doi":"10.22148/16.031","title":"Race, Writing, and Computation: Racial Difference and the US Novel, 1880-2000","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Race (biology); Computation; Racial differences; Sociology; Computer science; Gender studies; Ethnic group; Anthropology; Programming language","authors":[{"name":"Richard Jean So","is_ca":false},{"name":"Hoyt Long","is_ca":false},{"name":"Yuancheng Zhu","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03399302375473869,"gpt":0.3561622399781074,"spread":0.3221692162233687,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002556202,0.0002489955,0.0002570567,0.001565535,0.01018221,0.004293242,0.0003550808,0.001073612,0.005120891],"category_scores_gemma":[0.00663331,0.0001214676,0.0001340698,0.002088304,0.01298365,0.004175162,0.001990287,0.002678544,0.0004187482],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00394919,"about_ca_system_score_gemma":0.00182292,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01439251,"about_ca_topic_score_gemma":0.02570266,"domain_scores_codex":[0.9984301,0.0009720244,0.0000490344,0.0001981056,0.0001682594,0.0001824154],"domain_scores_gemma":[0.9978939,0.001333754,0.000250306,0.0001428874,0.0002213516,0.0001578922],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.00008916726,0.00002560363,0.001406112,0.00004210172,0.000006440372,0.0001943562,0.2152879,0.00005554831,0.0000906589,0.7152286,0.03369215,0.03388144],"study_design_scores_gemma":[0.00002642263,0.00004149786,0.007421412,0.0004080747,0.00001121661,0.0004881743,0.08928259,0.0002023925,0.0005562748,0.1152852,0.7862521,0.00002450233],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.291674,0.02451946,0.001829147,0.1016509,0.002282082,0.00002470591,0.0002893452,0.00004668806,0.5776837],"genre_scores_gemma":[0.9582317,0.002754885,0.0003648631,0.003347124,0.0005082909,0.00002420844,0.00004294182,0.00004666513,0.03467922],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01439251,"threshold_uncertainty_score":0.02865356,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3210937173","doi":"10.22148/001c.29802","title":"Audiobook Stylistics: Comparing print and audio in the bestselling segment","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Stylistics; Depiction; Focus (optics); Point (geometry); Style (visual arts); Set (abstract data type); Plot (graphics); Computer science; Advertising; Art; Literature; Mathematics","authors":[{"name":"Karl Berglund","is_ca":false},{"name":"Mats Dahllöf","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06872344227832917,"gpt":0.3059254956830484,"spread":0.2372020534047192,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001162747,0.0002119638,0.0002101073,0.00308696,0.001036662,0.003581005,0.0002374914,0.0002621273,0.005965314],"category_scores_gemma":[0.0103292,0.0001194276,0.0001757122,0.002702754,0.001545712,0.001901172,0.001399916,0.0004306037,0.001172587],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005756612,"about_ca_system_score_gemma":0.0002794133,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00146783,"about_ca_topic_score_gemma":0.002662599,"domain_scores_codex":[0.9989889,0.0003533507,0.00008847798,0.0001437057,0.0003379116,0.00008771538],"domain_scores_gemma":[0.9919881,0.004831039,0.001579434,0.0005945522,0.0006441296,0.000362774],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.002086675,0.0002169538,0.4735393,0.0007357433,0.0001370703,0.002286848,0.3042111,0.001166993,0.02739864,0.02100465,0.005853611,0.1613623],"study_design_scores_gemma":[0.00003087174,0.0003244181,0.6852593,0.0002211862,0.00009561953,0.003204782,0.2313438,0.003414085,0.0103473,0.007067807,0.0585539,0.0001368437],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9842756,0.0002041959,0.0009784608,0.00009139228,0.0000183187,0.00001582059,0.000295527,0.0000224816,0.01409812],"genre_scores_gemma":[0.996739,0.0001162776,0.0005922049,0.00002035104,0.00001807726,0.00001206309,0.0002989502,0.00003536981,0.00216777],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.005965314,"threshold_uncertainty_score":0.01995599,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2993640618","doi":"10.22148/001c.11826","title":"Do we know what we are doing?","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Mental Health Research Topics","field":"Psychology","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Open science; Value (mathematics); Norm (philosophy); Psychological science; Replication (statistics); Psychology; Space (punctuation); Social psychology; Epistemology; Sociology; Political science; Computer science; Law; Mathematics; Statistics","authors":[{"name":"Andrew Piper","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1495892918248977,"gpt":0.445563802616336,"spread":0.2959745107914384,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04578544,0.0006317475,0.001076794,0.002386702,0.009371823,0.01855665,0.001485774,0.00503462,0.01363649],"category_scores_gemma":[0.169923,0.0005766642,0.0008409058,0.002586027,0.02469349,0.0238597,0.006542509,0.0151104,0.008896349],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006574544,"about_ca_system_score_gemma":0.01580869,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007820679,"about_ca_topic_score_gemma":0.01097845,"domain_scores_codex":[0.9458447,0.03273217,0.002256377,0.004712239,0.01218399,0.002270496],"domain_scores_gemma":[0.8818179,0.0685349,0.008532139,0.009754933,0.02032631,0.01103386],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002014049,0.0001642517,0.02634527,0.001558162,0.000266666,0.0004092544,0.1276923,0.0001363515,0.000485001,0.1552804,0.3509425,0.3365185],"study_design_scores_gemma":[0.00002095224,0.00008010487,0.005450385,0.00256373,0.00007117354,0.0006708493,0.06473634,0.00008924864,0.0004526969,0.08577486,0.8399886,0.0001011059],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"commentary","genre_scores_codex":[0.01477581,0.06130392,0.01220394,0.7819423,0.02261188,0.00009104366,0.0004261864,0.000291628,0.1063533],"genre_scores_gemma":[0.4443466,0.07399358,0.01342969,0.4053006,0.008849445,0.000319164,0.0009004336,0.0008227863,0.0520377],"genre_candidate":"commentary","genre_consensus":"commentary","teacher_disagreement_score":0.04578544,"threshold_uncertainty_score":0.2421395,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2993740281","doi":"10.22148/16.035","title":"Data Cultures, Culture as Data – Special Issue of Cultural Analytics","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Research data; Data science; Computer science; Sociology; Data curation","authors":[{"name":"Tanya Clement","is_ca":false},{"name":"Amelia Acker","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1263592046849277,"gpt":0.3235582065088711,"spread":0.1971990018239434,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01338524,0.001656848,0.00267076,0.01167234,0.006002742,0.02044198,0.002720657,0.007201587,0.03877913],"category_scores_gemma":[0.03685902,0.00142125,0.001610531,0.01246327,0.006302414,0.02652218,0.01154021,0.0130102,0.01252718],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00382705,"about_ca_system_score_gemma":0.006697992,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002124444,"about_ca_topic_score_gemma":0.004351779,"domain_scores_codex":[0.989517,0.002952193,0.001438728,0.001453295,0.003987321,0.0006514694],"domain_scores_gemma":[0.9375276,0.03436476,0.003089763,0.006429594,0.01159173,0.006996519],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001631511,0.00004184012,0.0005316368,0.0006033237,0.00002223585,0.00009074491,0.000615328,0.00007170261,0.0002459815,0.01154637,0.9403657,0.04584878],"study_design_scores_gemma":[0.000002515561,0.00001117757,0.0004882879,0.0006344254,0.000007656083,0.0002044409,0.000499786,0.0001016639,0.00007622822,0.005110503,0.9928449,0.00001839056],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"editorial","genre_gemma":"editorial","genre_scores_codex":[0.0008180216,0.09776075,0.005953442,0.1434879,0.7184384,0.0001566271,0.0008277407,0.0004781999,0.03207881],"genre_scores_gemma":[0.006272408,0.1013578,0.003310976,0.03401759,0.7940872,0.0002186155,0.002129932,0.001646116,0.05695941],"genre_candidate":"editorial","genre_consensus":"editorial","teacher_disagreement_score":0.03877913,"threshold_uncertainty_score":0.1297292,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2802843067","doi":"10.22148/16.022","title":"Self-Repetition and East Asian Literary Modernity, 1900-1930","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Music Technology and Sound Studies","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"BATES; Feeling; Sampling (signal processing); Literature; Aesthetics; Sound (geography); Empirical research; History; Psychology; Art; Social psychology; Computer science; Epistemology; Philosophy; Engineering; Acoustics; Telecommunications","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02039322726241128,"gpt":0.2531978847216531,"spread":0.2328046574592418,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001515405,0.0003366722,0.0003100487,0.00151961,0.004179905,0.002826867,0.0002752114,0.0004415841,0.003421459],"category_scores_gemma":[0.001701216,0.0001694883,0.000144085,0.00180234,0.01001563,0.002583133,0.001667231,0.001454263,0.000391543],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003306062,"about_ca_system_score_gemma":0.001152146,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007109692,"about_ca_topic_score_gemma":0.0152262,"domain_scores_codex":[0.9995091,0.0001956133,0.00004858477,0.00008119318,0.0000691898,0.00009627439],"domain_scores_gemma":[0.998913,0.0004804277,0.0001995894,0.00009274388,0.000184398,0.000129714],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00011943,0.00005361224,0.01001325,0.0002518242,0.00002650421,0.001012856,0.7714717,0.0001150047,0.0009883936,0.1850698,0.002060828,0.02881681],"study_design_scores_gemma":[0.00004749137,0.0003991781,0.1272922,0.001027382,0.00006760463,0.002589447,0.3556892,0.0004365139,0.003518339,0.03466404,0.4741744,0.00009424138],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7896613,0.0153625,0.0003244543,0.002118951,0.0002712967,0.00001888922,0.00004994809,0.00002067123,0.192172],"genre_scores_gemma":[0.9874883,0.001958485,0.00007994581,0.00009985262,0.00007636927,0.000005524163,0.00001166186,0.000009079602,0.01027082],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.007109692,"threshold_uncertainty_score":0.02398729,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3000500424","doi":"10.22148/001c.11652","title":"Fleshing Out Models of Gender in English-Language Novels (1850 – 2000)","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Gender Studies in Language","field":"Social Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Binary opposition; Psychology; Characterization (materials science); Linguistics; Sociology; Gender studies; Philosophy","authors":[{"name":"Jonathan Cheng","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1068315857302793,"gpt":0.3483203648240278,"spread":0.2414887790937486,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003725492,0.0005101349,0.0002488901,0.001899786,0.001854053,0.004068424,0.0008747919,0.0006906866,0.003524378],"category_scores_gemma":[0.01159577,0.0002142141,0.0003835312,0.001419651,0.006192963,0.005342886,0.001201484,0.001339127,0.0005842246],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003330614,"about_ca_system_score_gemma":0.0006569501,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006139969,"about_ca_topic_score_gemma":0.00836512,"domain_scores_codex":[0.998946,0.0006883451,0.00003315719,0.0001507131,0.0001245721,0.00005733045],"domain_scores_gemma":[0.9947357,0.004077033,0.0004546036,0.0003252306,0.0002903877,0.0001170496],"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":[0.0001285428,0.00004066677,0.01111863,0.0001514695,0.00003637959,0.0003908963,0.2501152,0.003077362,0.0003095982,0.6963503,0.006201428,0.0320795],"study_design_scores_gemma":[0.00002786812,0.0001791564,0.04063009,0.0008612444,0.00005588286,0.001120916,0.1727206,0.03198618,0.001506958,0.447251,0.3035617,0.00009849889],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7455669,0.006609496,0.03990894,0.01186267,0.000578842,0.00003814216,0.0007757759,0.0001515693,0.1945076],"genre_scores_gemma":[0.9927314,0.0005957711,0.001038186,0.000108085,0.00008272642,0.000009278408,0.0001209873,0.00005277663,0.005260801],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.006139969,"threshold_uncertainty_score":0.02416545,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2988314392","doi":"10.22148/16.049","title":"A Shared Task for the Digital Humanities Chapter 2: Evaluating Annotation Guidelines","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Guideline; Annotation; Task (project management); Section (typography); Computer science; Data science; Artificial intelligence; Political science; Engineering; Systems engineering","authors":[{"name":"Evelyn Gius","is_ca":false},{"name":"Nils Reiter","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1018420387922206,"gpt":0.3715093349101505,"spread":0.2696672961179299,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.04692203,0.001505454,0.001767347,0.006084652,0.004477749,0.007984232,0.003374399,0.004248032,0.01162917],"category_scores_gemma":[0.1885483,0.0007979593,0.001106921,0.004650766,0.002310216,0.01145882,0.009599663,0.003629172,0.004654831],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00458913,"about_ca_system_score_gemma":0.008193662,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01144683,"about_ca_topic_score_gemma":0.01780963,"domain_scores_codex":[0.9087276,0.06026801,0.006812132,0.005144856,0.0174338,0.001613664],"domain_scores_gemma":[0.8523111,0.08454446,0.005347255,0.01903677,0.03456369,0.004196706],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001678711,0.0029275,0.02198892,0.005334939,0.0004857744,0.0006118375,0.02512304,0.01411222,0.01761865,0.04582936,0.1777117,0.6865774],"study_design_scores_gemma":[0.00121907,0.002410392,0.03278251,0.006029101,0.0006928158,0.000903485,0.0440766,0.1782044,0.05867124,0.1627481,0.5114981,0.0007642117],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.336787,0.004490403,0.455254,0.0152191,0.001557852,0.02218469,0.01641939,0.01601109,0.1320764],"genre_scores_gemma":[0.3657272,0.0008203166,0.5883886,0.001606955,0.0001973393,0.006673206,0.02141812,0.001743746,0.01342464],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.953078,"threshold_uncertainty_score":0.2481503,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4313363893","doi":"10.22148/001c.46662","title":"Are Computational Literary Studies Structuralist?","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Structuralism (philosophy of science); Narrative; Epistemology; Literary criticism; Presentation (obstetrics); Relation (database); Literary theory; Sociology; Philosophy; Computer science; Linguistics","authors":[{"name":"Evelyn Gius","is_ca":false},{"name":"Janina Jacke","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1411680910892733,"gpt":0.3041771103153833,"spread":0.16300901922611,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03060575,0.0008334725,0.001374638,0.005977337,0.006703847,0.01924643,0.004671792,0.0042473,0.01199166],"category_scores_gemma":[0.117806,0.000917417,0.001332536,0.006119515,0.06208305,0.03904952,0.0105409,0.006625152,0.001809013],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006068312,"about_ca_system_score_gemma":0.00481933,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001268292,"about_ca_topic_score_gemma":0.001708948,"domain_scores_codex":[0.9628887,0.02629367,0.001011695,0.004585797,0.004499658,0.0007204147],"domain_scores_gemma":[0.8378956,0.1248957,0.005241651,0.02435835,0.006029261,0.001579425],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001336345,0.00001194772,0.0004393377,0.0002102024,0.0000209416,0.00002603715,0.005705201,0.000340614,0.00003134715,0.9837945,0.002146318,0.007260163],"study_design_scores_gemma":[0.00001452802,0.000006533714,0.000227627,0.0001480114,0.000007951185,0.00004523785,0.002909803,0.0009353471,0.0000767988,0.9668513,0.02877014,0.000006794034],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05033036,0.01685806,0.3592331,0.2667365,0.002894093,0.0002755098,0.000941904,0.0007470662,0.3019834],"genre_scores_gemma":[0.8870693,0.005823498,0.08523071,0.007089166,0.003000371,0.0009652613,0.0006944154,0.0005956469,0.009531759],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.03060575,"threshold_uncertainty_score":0.1618606,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3005998127","doi":"10.22148/16.034","title":"A BLAST-based, Language-agnostic Text Reuse Algorithm with a MARKUS Implementation and Sequence Alignment Optimized for Large Chinese Corpora","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Computer science; Reuse; Natural language processing; Phrase; Sequence (biology); Artificial intelligence; Engineering","authors":[{"name":"Paul Vierthaler","is_ca":false},{"name":"Meet Gelein","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01158736095800556,"gpt":0.3090124364441694,"spread":0.2974250754861639,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002832057,0.002866157,0.002179322,0.002974736,0.003108259,0.002276474,0.002799159,0.001807993,0.01293735],"category_scores_gemma":[0.007849076,0.001697958,0.00191081,0.005937689,0.0009527617,0.002432511,0.002236833,0.003262169,0.01823928],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001165234,"about_ca_system_score_gemma":0.003592986,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004480249,"about_ca_topic_score_gemma":0.01347598,"domain_scores_codex":[0.998014,0.0003495892,0.0002669621,0.0008323491,0.0003758373,0.0001613102],"domain_scores_gemma":[0.9983144,0.0005388206,0.0001693907,0.0004022489,0.0004556921,0.0001195431],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.004164942,0.0008601502,0.0111502,0.004060322,0.001176163,0.001953809,0.002877934,0.01010929,0.3003986,0.01350539,0.1473284,0.5024148],"study_design_scores_gemma":[0.001425735,0.001782929,0.02188251,0.0004686585,0.001382371,0.00553958,0.00209577,0.3627977,0.3089742,0.02960248,0.2633708,0.0006772167],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07497234,0.001228422,0.7125255,0.0005807053,0.0006907873,0.001498045,0.01887269,0.1793929,0.01023872],"genre_scores_gemma":[0.06024638,0.0003611268,0.8932496,0.0002877522,0.00006942928,0.001260037,0.02999623,0.008391843,0.006137557],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01293735,"threshold_uncertainty_score":0.04327983,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3129389071","doi":"10.22148/001c.21182","title":"Cultural Capitals: Modeling Minor European Literature","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Translation Studies and Practices","field":"Arts and Humanities","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Minor (academic); Hierarchy; German; History; Media studies; Literature; Sociology; Political science; Art; Law","authors":[{"name":"Matt Erlin","is_ca":true},{"name":"Andrew Piper","is_ca":true},{"name":"Douglas Knox","is_ca":true},{"name":"Stephen Pentecost","is_ca":true},{"name":"Michaela Drouillard","is_ca":true},{"name":"Brian Powell","is_ca":true},{"name":"Cienna Townson","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09238203772860684,"gpt":0.2961933228713019,"spread":0.2038112851426951,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001999936,0.0003929686,0.0005604933,0.002198222,0.001194769,0.004177895,0.001232154,0.0008884112,0.003371267],"category_scores_gemma":[0.01456274,0.0003235068,0.0007777113,0.003100127,0.001805185,0.002817651,0.002277699,0.0009042021,0.0003644853],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002509042,"about_ca_system_score_gemma":0.001289737,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04478665,"about_ca_topic_score_gemma":0.03116975,"domain_scores_codex":[0.9993339,0.0003346639,0.00002295171,0.0001670469,0.0000543798,0.00008706856],"domain_scores_gemma":[0.9942696,0.004114611,0.0006218144,0.0003728764,0.0002506805,0.0003704704],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0005077579,0.0005301065,0.2239301,0.0001633453,0.0002694259,0.0006108709,0.01113555,0.5428519,0.0007050338,0.1767368,0.003247909,0.03931124],"study_design_scores_gemma":[0.00006014896,0.00005373459,0.01994621,0.00005446317,0.00005889137,0.0001184003,0.003709432,0.9062852,0.0002036963,0.063377,0.006107517,0.00002536643],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.974568,0.000420287,0.01289,0.0008591935,0.00001485474,0.00004370895,0.0004816184,0.00009984615,0.01062248],"genre_scores_gemma":[0.9940064,0.000161965,0.004360375,0.00005046649,0.00001006398,0.00004487024,0.0003539996,0.00002553535,0.0009861917],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04478665,"threshold_uncertainty_score":0.08905196,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4283727265","doi":"10.22148/001c.36446","title":"Literary value in the era of big data. Operationalizing critical distance in professional and non-professional reviews","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Communication and Language","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Operationalization; Literary criticism; Value (mathematics); Literary science; Sociology; Criticism; Newspaper; Phenomenon; Praxis; Literary theory; Epistemology; Reading (process); Computer science; Literature; Media studies; Political science; Law; Art; Philosophy","authors":[{"name":"Massimo Salgaro","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08848718552736484,"gpt":0.371559383609055,"spread":0.2830721980816902,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03395707,0.0006517525,0.0008213061,0.01462608,0.003399231,0.01773131,0.001535136,0.001808448,0.002410185],"category_scores_gemma":[0.1753516,0.0005148805,0.0005248477,0.0133581,0.02074544,0.02431169,0.007289068,0.002549738,0.000500225],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007015625,"about_ca_system_score_gemma":0.004106636,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001586226,"about_ca_topic_score_gemma":0.002176858,"domain_scores_codex":[0.9500908,0.03191494,0.002328013,0.00345276,0.01170279,0.0005107439],"domain_scores_gemma":[0.7011628,0.233572,0.02211618,0.01479504,0.02358246,0.004771488],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.0002636905,0.00007072195,0.03329464,0.003539011,0.0002468472,0.0004743987,0.08720715,0.001071758,0.0008017636,0.6708349,0.01326041,0.1889348],"study_design_scores_gemma":[0.00003625743,0.0001707649,0.03290887,0.003104006,0.0001030243,0.001167209,0.06980897,0.004837089,0.00153575,0.7266645,0.1595489,0.0001147258],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.3182392,0.09085502,0.2016661,0.09886435,0.005068257,0.001163206,0.002194819,0.0004864231,0.2814627],"genre_scores_gemma":[0.9503495,0.00660822,0.0341818,0.0022194,0.00188624,0.0005479752,0.000289162,0.0001476181,0.003769932],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.03395707,"threshold_uncertainty_score":0.1795843,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4211124067","doi":"10.22148/001c.32551","title":"Shakespeare and Company Project Data Sets","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Shakespeare, Adaptation, and Literary Criticism","field":"Arts and Humanities","cited_by":8,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"University Research Committee, Emory University; Princeton University","keywords":"Bridge (graph theory); Work (physics); Set (abstract data type); Data set; Library science; History; World Wide Web; Computer science; Engineering","authors":[{"name":"Joshua Kotin","is_ca":false},{"name":"Rebecca Sutton Koeser","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1678217392894345,"gpt":0.3192285861083995,"spread":0.151406846818965,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007393952,0.001107799,0.001139704,0.01023913,0.002546333,0.004360908,0.002519851,0.00229236,0.02762804],"category_scores_gemma":[0.04974448,0.0007966734,0.0008233391,0.01805846,0.001300662,0.002288242,0.0038831,0.002884217,0.02880403],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002651378,"about_ca_system_score_gemma":0.007196125,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04153895,"about_ca_topic_score_gemma":0.03969883,"domain_scores_codex":[0.9882691,0.002326169,0.001441021,0.001344334,0.005752621,0.0008667418],"domain_scores_gemma":[0.9670568,0.009968745,0.002147352,0.004375129,0.01466654,0.001785437],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001980502,0.0002933716,0.01241478,0.0005969677,0.00005184532,0.0002213814,0.00102063,0.00093494,0.0001815098,0.008481079,0.9463494,0.02925599],"study_design_scores_gemma":[0.000106791,0.00004305372,0.02662944,0.0003956138,0.00002043904,0.0001555737,0.001950792,0.0006639364,0.0005280591,0.003336826,0.9660713,0.00009821674],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.01474816,0.0005863463,0.003289395,0.001755954,0.0003211387,0.001397771,0.9556482,0.0006135434,0.02163949],"genre_scores_gemma":[0.01286065,0.0003495838,0.006919723,0.0003444021,0.00007948716,0.004474171,0.9676731,0.0002372569,0.007061612],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.04153895,"threshold_uncertainty_score":0.09242499,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3174730804","doi":"10.22148/001c.25273","title":"Too isolated, too insular: American Literature and the World","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Historical Economic and Social Studies","field":"Economics, Econometrics and Finance","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"American literature; History; Identity (music); Politics; National identity; Principal (computer security); Political science; Literature; Law; Aesthetics; Art","authors":[{"name":"Matthew Wilkens","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02049754593415602,"gpt":0.2272385805091287,"spread":0.2067410345749727,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001630723,0.0002002846,0.0002819437,0.00794538,0.006105428,0.01429905,0.0004924271,0.0007230421,0.003757209],"category_scores_gemma":[0.00605776,0.0001358008,0.0001064034,0.01107152,0.008933914,0.006491104,0.002189394,0.001352406,0.0004409502],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002242206,"about_ca_system_score_gemma":0.001670474,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007777507,"about_ca_topic_score_gemma":0.02158973,"domain_scores_codex":[0.9989005,0.0004681832,0.00004510756,0.000141381,0.0002968449,0.0001479491],"domain_scores_gemma":[0.9935982,0.003004098,0.0008309541,0.0005381159,0.001282849,0.0007458125],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001232412,0.00006578715,0.04514714,0.000322861,0.00006461292,0.001036395,0.4776182,0.0001230784,0.001043693,0.260672,0.02895667,0.1848264],"study_design_scores_gemma":[0.000006464638,0.00003703353,0.09533429,0.0008989286,0.00004894939,0.0009478709,0.3947001,0.0001809126,0.0004628976,0.03300415,0.4743393,0.00003914976],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6062086,0.02723507,0.0009648492,0.01736938,0.0008500168,0.000009088138,0.0001664713,0.00005269835,0.3471438],"genre_scores_gemma":[0.9852602,0.006979395,0.0002430425,0.001152828,0.0004181961,0.00000515379,0.00006531876,0.00003424586,0.005841652],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01429905,"threshold_uncertainty_score":0.01626843,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2985974955","doi":"10.22148/16.050","title":"A Shared Task for the Digital Humanities Chapter 3: Description of Submitted Guidelines and Final Evaluation Results","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Task (project management); Annotation; Reflection (computer programming); Narrative; Computer science; Digital humanities; Mathematics education; Information retrieval; Data science; Artificial intelligence; Psychology; World Wide Web; Linguistics; Programming language; Engineering; Philosophy","authors":[{"name":"Marcus Willand","is_ca":false},{"name":"Evelyn Gius","is_ca":false},{"name":"Nils Reiter","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.3084406244997505,"gpt":0.3152306524364105,"spread":0.006790027936659948,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.06616345,0.001742839,0.001418978,0.006245153,0.007908522,0.01014654,0.003210788,0.003592662,0.037553],"category_scores_gemma":[0.1586,0.001396477,0.001427465,0.005442129,0.00301132,0.006405545,0.01000162,0.004202381,0.03526781],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.007636675,"about_ca_system_score_gemma":0.01890993,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0109248,"about_ca_topic_score_gemma":0.01718282,"domain_scores_codex":[0.917559,0.04724182,0.007825905,0.005443599,0.01807242,0.003857173],"domain_scores_gemma":[0.8749779,0.0297655,0.00284312,0.01408079,0.07264806,0.005684523],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001806225,0.001692886,0.00424937,0.009682487,0.000114783,0.0007732506,0.09002178,0.001964442,0.01932767,0.01759154,0.4398637,0.4129119],"study_design_scores_gemma":[0.0003891141,0.0008249132,0.005911415,0.005869525,0.000129742,0.0002466872,0.03556634,0.002354964,0.01781237,0.01148435,0.9190764,0.0003341483],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1331837,0.005519364,0.3706459,0.02138407,0.009855323,0.142858,0.04636386,0.0282648,0.241925],"genre_scores_gemma":[0.1504785,0.002422553,0.4202439,0.005575412,0.0008145286,0.2355574,0.04925694,0.01717796,0.1184729],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9338366,"threshold_uncertainty_score":0.34991,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4242882238","doi":"10.22148/16.014","title":"Topic Modeling, Epistemology, and the English and German Novel","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Contextualization; German; Context (archaeology); Historicism; Meaning (existential); Epistemology; Space (punctuation); Linguistics; Sociology; Philosophy; History; Interpretation (philosophy); Archaeology","authors":[{"name":"Matt Erlin","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07670709901819155,"gpt":0.4049405778109269,"spread":0.3282334787927353,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006584639,0.0005358199,0.0005557696,0.003373942,0.002986054,0.009192334,0.0008490619,0.001345505,0.002713371],"category_scores_gemma":[0.009498481,0.0003616123,0.0005623279,0.003081747,0.01367325,0.009795095,0.002691953,0.002020487,0.0004525953],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005124731,"about_ca_system_score_gemma":0.001166411,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005559473,"about_ca_topic_score_gemma":0.005370311,"domain_scores_codex":[0.9960791,0.002813731,0.0001593878,0.0004495567,0.0002951878,0.000203078],"domain_scores_gemma":[0.9928418,0.005794122,0.0004663424,0.0004546664,0.0002578066,0.0001852341],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0000211426,0.00001301622,0.0006736104,0.00005447377,0.00001490109,0.0000881853,0.008371454,0.0005306892,0.00005144234,0.9787232,0.003670926,0.007786988],"study_design_scores_gemma":[0.00001813434,0.00001353457,0.001438069,0.0001966596,0.00001834958,0.0002666632,0.009823291,0.004269834,0.0002796041,0.9035063,0.08014103,0.00002839204],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.2289388,0.04642818,0.2379523,0.1068935,0.001848075,0.00009055002,0.001191681,0.000576172,0.3760807],"genre_scores_gemma":[0.9758194,0.00325989,0.01005758,0.0009640074,0.0006236574,0.0000659782,0.0002278327,0.0001464621,0.008835305],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009192334,"threshold_uncertainty_score":0.03718269,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3153523203","doi":"10.22148/001c.22335","title":"Introduction: Contemporary Culture After Data Science","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Affordance; Analytics; Sociology; Cultural analytics; Data science; Epistemology; Social science; Computer science; Human–computer interaction; World Wide Web; Philosophy; The Internet","authors":[{"name":"Richard Jean So","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1232672934213013,"gpt":0.2890828548198972,"spread":0.1658155613985959,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003921762,0.0006066363,0.000537261,0.00190668,0.00506505,0.01176881,0.0007550393,0.002998454,0.01932479],"category_scores_gemma":[0.01572859,0.0003136108,0.0004556891,0.003277916,0.01195266,0.01233397,0.003365304,0.01080055,0.005187333],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004260417,"about_ca_system_score_gemma":0.003534991,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003984781,"about_ca_topic_score_gemma":0.002686659,"domain_scores_codex":[0.9963568,0.001714876,0.0001767615,0.00062523,0.0008984004,0.0002279028],"domain_scores_gemma":[0.9884014,0.00776577,0.0004261783,0.0008648266,0.001922123,0.000619715],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00002761943,0.00001894239,0.0005288466,0.0003700997,0.0000108254,0.00006079419,0.008872197,0.0001208225,0.0001944863,0.6408973,0.295499,0.05339909],"study_design_scores_gemma":[0.00000259579,0.0000118444,0.0003355323,0.0008272558,0.000003012778,0.00007713523,0.002896948,0.00006792634,0.0001135317,0.07149272,0.9241597,0.00001177813],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"commentary","genre_gemma":"commentary","genre_scores_codex":[0.004864044,0.1432447,0.01562554,0.538807,0.08797035,0.00005115546,0.0006655316,0.0002174403,0.2085543],"genre_scores_gemma":[0.245756,0.1859428,0.01108981,0.2499173,0.1310199,0.0003744177,0.0009816467,0.001508987,0.1734091],"genre_candidate":"commentary","genre_consensus":"commentary","teacher_disagreement_score":0.01932479,"threshold_uncertainty_score":0.06464785,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3040858410","doi":"10.22148/001c.13680","title":"Like Two Pis in a Pod: Author Similarity Across Time in the Ancient Greek Corpus","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"Alfred P. Sloan Foundation; National Science Foundation","keywords":"Similarity (geometry); Divergence (linguistics); Focus (optics); Imitation; Range (aeronautics); Natural language processing; Reading (process); Feature (linguistics); Linguistics; Computer science; Artificial intelligence; History; Psychology","authors":[{"name":"Grant Storey","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08545780393490782,"gpt":0.3457816242500205,"spread":0.2603238203151127,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.001212506,0.0001863318,0.000256377,0.004634674,0.001038511,0.001374664,0.0002837037,0.0004617821,0.001776638],"category_scores_gemma":[0.01389705,0.0001074094,0.0001611713,0.006446868,0.001033856,0.001421867,0.001289411,0.0004712033,0.0006740511],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004381333,"about_ca_system_score_gemma":0.0003418144,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001839852,"about_ca_topic_score_gemma":0.004118424,"domain_scores_codex":[0.9990045,0.0003492129,0.0001625501,0.0002174942,0.0002164542,0.00004972272],"domain_scores_gemma":[0.9896474,0.00706922,0.00105378,0.001083123,0.0008756287,0.0002708988],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.001358796,0.0004450501,0.4313358,0.002174082,0.0002827454,0.004049053,0.1048628,0.005400187,0.02023062,0.02029331,0.03867621,0.3708914],"study_design_scores_gemma":[0.0001050549,0.0001436849,0.7394413,0.0004829129,0.0001251911,0.005006166,0.03519248,0.02031833,0.008373646,0.01243742,0.1782539,0.0001199632],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9879459,0.0008040916,0.002079884,0.0002170102,0.00006762374,0.00002784063,0.003603871,0.00008194748,0.005171844],"genre_scores_gemma":[0.9811764,0.0003968239,0.005724411,0.00005115079,0.00004904283,0.00007327032,0.0100503,0.00007694586,0.002401638],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9953653,"threshold_uncertainty_score":0.006412446,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4402197367","doi":"10.22148/001c.121866","title":"Digital Humanities and Distributed Cognition: From a Lack of Theory to its Visual Augmentation","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language and cultural evolution","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Cognition; Cognitive science; Digital humanities; Conceptual framework; Visualization; Psychology; Sociology; Computer science; Epistemology; Cognitive psychology; Artificial intelligence; Social science","authors":[{"name":"Florian Windhager","is_ca":false},{"name":"Eva Mayr","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06055083490596529,"gpt":0.3632976391978363,"spread":0.302746804291871,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005932253,0.0006245134,0.0008617571,0.004851622,0.002372029,0.0140939,0.002194885,0.003557249,0.006038801],"category_scores_gemma":[0.01308269,0.000475526,0.0004891467,0.003268456,0.04977537,0.03275851,0.007927246,0.005246979,0.0007024736],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00443686,"about_ca_system_score_gemma":0.003017074,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002725282,"about_ca_topic_score_gemma":0.001950123,"domain_scores_codex":[0.9971995,0.00155413,0.0001207889,0.0004342475,0.0005505128,0.0001407844],"domain_scores_gemma":[0.9828343,0.01333642,0.000567427,0.002091776,0.0007311477,0.000438996],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000009053206,0.00001272116,0.0002646537,0.0003113777,0.000008287891,0.00003819319,0.003026125,0.0003274573,0.00006613252,0.9712337,0.001880798,0.02282146],"study_design_scores_gemma":[0.000006904706,0.00001324169,0.0003173564,0.0006912297,0.000007150601,0.0001122271,0.004590427,0.0006637532,0.00009685763,0.9432636,0.05022534,0.00001178522],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.04270505,0.2211595,0.1507757,0.3028774,0.003190047,0.00008190128,0.0003474982,0.0004904687,0.2783725],"genre_scores_gemma":[0.8801666,0.07221643,0.02355966,0.01270264,0.003178208,0.0002287145,0.0001518332,0.0002217493,0.007574203],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.0140939,"threshold_uncertainty_score":0.03219187,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2978801148","doi":"10.22148/16.024","title":"Nation, Ethnicity, and the Geography of British Fiction, 1880-1940","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Travel Writing and Literature","field":"Arts and Humanities","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Unconscious mind; Ethnic group; Empire; History; First world war; Literature; Sociology; Geography; Anthropology; Ancient history; Art; Psychology; Psychoanalysis","authors":[{"name":"Elizabeth F. Wilkens Evans","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02283525077622644,"gpt":0.2388033116570712,"spread":0.2159680608808448,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000983351,0.0001742061,0.0002035688,0.002485963,0.005619115,0.004671839,0.0003592012,0.0004838578,0.005031242],"category_scores_gemma":[0.00384456,0.0001537159,0.00007203556,0.003888946,0.006543303,0.001868786,0.00184253,0.0008955431,0.000713148],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.01016387,"about_ca_system_score_gemma":0.002347834,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.276879,"about_ca_topic_score_gemma":0.4483694,"domain_scores_codex":[0.9992806,0.0003301645,0.00003584365,0.00007916969,0.0001366101,0.00013761],"domain_scores_gemma":[0.9988253,0.0004882669,0.0001471897,0.00006049301,0.000304744,0.0001740967],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00009669462,0.00001937167,0.01682338,0.0001654948,0.00001172079,0.0005789761,0.8782093,0.00007803039,0.000472895,0.06474159,0.01290563,0.02589686],"study_design_scores_gemma":[0.000008163858,0.00003534189,0.109337,0.0004295502,0.00001658153,0.0003805242,0.37566,0.00006870165,0.0003433501,0.004010174,0.5096824,0.00002815668],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.713851,0.008714741,0.0002238671,0.00407945,0.0002740856,0.00001263061,0.0003031273,0.00001967609,0.2725213],"genre_scores_gemma":[0.971818,0.00218693,0.00009439648,0.0001790786,0.0000420231,0.000007237129,0.0001265764,0.00003543121,0.02551029],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.276879,"threshold_uncertainty_score":0.5505347,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3151065226","doi":"10.22148/001c.21993","title":"Imaginative Networks: Tracing Connections Among Early Modern Book Dedications","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Politics; Popularity; Centrality; Literature; History; Classics; Law; Art; Political science","authors":[{"name":"John Ladd","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03109520831080825,"gpt":0.295098267797732,"spread":0.2640030594869237,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.001215874,0.0002577409,0.0001671458,0.005709999,0.001149457,0.002602482,0.0006293782,0.0005477408,0.002328591],"category_scores_gemma":[0.01400933,0.0002272342,0.0001763028,0.005380285,0.001582243,0.005635465,0.002016222,0.0007261533,0.0004790698],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001390904,"about_ca_system_score_gemma":0.0004089861,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004700362,"about_ca_topic_score_gemma":0.0084667,"domain_scores_codex":[0.9990563,0.0004605429,0.00003469797,0.0001986834,0.0001768873,0.00007295698],"domain_scores_gemma":[0.990807,0.005666559,0.001667871,0.0007305156,0.0007358253,0.0003922476],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.000279242,0.0001208565,0.3016784,0.0004081519,0.0001399127,0.0009949063,0.152444,0.008270001,0.005643525,0.283599,0.01190855,0.2345135],"study_design_scores_gemma":[0.000019669,0.00009715411,0.3040045,0.0006073256,0.0001207415,0.001834039,0.1015736,0.08661968,0.005850316,0.2605259,0.2386171,0.0001298752],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.889075,0.002032795,0.06593759,0.001681926,0.00007315283,0.00006272531,0.00160383,0.0002066979,0.03932645],"genre_scores_gemma":[0.9817472,0.0004963222,0.01440128,0.00005269622,0.00002867627,0.00004709423,0.0007354745,0.00006168926,0.002429521],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.99429,"threshold_uncertainty_score":0.01009178,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}