{"meta":{"query_hash":"517981b04501","filters":{"venue":"Journal of Cultural Analytics"},"cohort_total":199,"direct_labels_cover":0,"predictions_cover":199,"exported":199,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/517981b04501","api":"https://metacan.xera.ac/api/v1/cohort?venue=Journal+of+Cultural+Analytics"},"results":[{"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.07116406461840949,"score_gpt":0.27356656217566255,"score_spread":0.20240249755725306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2401667852","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93317354,0.0006360142,0.000015266927,0.0009417039,0.00054120045,0.00007050475,0.00006287686,0.000017649207,0.06454127],"genre_scores_gemma":[0.9799966,0.00016751912,0.000020287294,0.00008042079,0.0010417164,2.0068892e-7,0.0000016087018,0.000009571347,0.01868205],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99803317,0.00002541162,0.000728008,0.00013592861,0.0007334893,0.0003439759],"domain_scores_gemma":[0.9970212,0.00014187311,0.00074335164,0.00011050451,0.0018960481,0.00008703383],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":["sts"],"category_scores_codex":[0.00045969995,0.00020705682,0.00035987535,0.000108761014,0.0013120066,0.0024924453,0.00056557707,0.000040586572,0.00006139894],"category_scores_gemma":[0.00026718434,0.000085940505,0.0002592265,0.00009698617,0.0036352372,0.0026095877,0.00014659911,0.00021859739,0.0000053888343],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001072078,0.00011445645,0.00053784234,0.000049968687,0.00034601448,0.000019450546,0.061970923,0.000009158838,0.0006709402,0.89822805,0.026374754,0.011571219],"study_design_scores_gemma":[0.0019669605,0.0013462561,0.0024659764,0.00048588958,0.00044401633,0.0002702865,0.20255984,0.00019465841,0.0009751098,0.030097961,0.7582834,0.00090968277],"about_ca_topic_score_codex":0.0000079371175,"about_ca_topic_score_gemma":0.00005976284,"teacher_disagreement_score":0.8681301,"about_ca_system_score_codex":0.0000863317,"about_ca_system_score_gemma":0.00007765137,"threshold_uncertainty_score":0.99998814},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.07790262750403741,"score_gpt":0.26037095621257267,"score_spread":0.18246832870853524,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2500585446","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.934434,0.0014127382,0.000008448503,0.0029836206,0.0005438721,0.00003194161,0.000019633686,0.0000069282746,0.060558792],"genre_scores_gemma":[0.9771803,0.0002970724,0.000011025319,0.00011434756,0.0007890319,1.7490808e-7,2.4612754e-7,0.00000494516,0.021602845],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99911183,0.00002268568,0.00043195253,0.000038496957,0.00027113795,0.00012392145],"domain_scores_gemma":[0.99875027,0.00016425748,0.000383065,0.00008329774,0.00054981513,0.00006928417],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022634132,0.00007676261,0.00016658376,0.000033676166,0.0001939569,0.00028798266,0.00026260733,0.00001730618,0.00032909456],"category_scores_gemma":[0.00020791987,0.000026902473,0.00020440052,0.000016671958,0.00033026643,0.00057224405,0.00002672882,0.00008296302,0.000013030521],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011675455,0.00006832477,0.00063796644,0.0000206173,0.00040001515,0.000013098664,0.008380999,0.000010280229,0.00035899135,0.91548246,0.055013306,0.019497212],"study_design_scores_gemma":[0.00047390972,0.00034324816,0.0012201802,0.00015496797,0.00013630427,0.00002544415,0.027466148,0.000009756922,0.00042307982,0.01973919,0.9498503,0.00015745153],"about_ca_topic_score_codex":0.0000066314246,"about_ca_topic_score_gemma":0.000114773255,"teacher_disagreement_score":0.89574325,"about_ca_system_score_codex":0.0000149679645,"about_ca_system_score_gemma":0.000030906067,"threshold_uncertainty_score":0.36033547},"labels":[],"label_agreement":null},{"id":"W2586163947","doi":"10.22148/16.008","title":"Genre Theory and Historicism","year":2016,"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Poetics; Literature; New Historicism; Historicism; Reflexivity; Literary criticism; Criticism; Literary theory; Philosophy; Quarter (Canadian coin); Literary science; History; Art; Sociology; Poetry; Social science","score_opus":0.0330594690223462,"score_gpt":0.28266785351531826,"score_spread":0.24960838449297207,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2586163947","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08484387,0.0014182987,0.9058395,0.0066020642,0.0006291767,0.000024157029,0.000001111456,0.000020538475,0.000621266],"genre_scores_gemma":[0.99062365,0.00019579417,0.0060669403,0.00024071947,0.0001719273,8.9022926e-8,7.395503e-8,0.000002073386,0.002698724],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99933577,0.000081470185,0.00022926672,0.000068764995,0.0001808698,0.000103889135],"domain_scores_gemma":[0.9992302,0.00012779966,0.0002157269,0.000088307286,0.00023572672,0.00010222648],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00072404125,0.000061206905,0.0001218358,0.000049880684,0.000060744707,0.00004599836,0.0002608678,0.000037056976,0.000016432115],"category_scores_gemma":[0.00020997997,0.000028252918,0.00006927954,0.00011731503,0.00003553997,0.00045579203,0.000052508836,0.000091622846,0.000009730374],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000049580776,0.000032028336,0.0017594857,0.000012350324,0.00010032123,0.000086059896,0.0021895166,0.000013031613,0.014402265,0.85739,0.007156454,0.116808936],"study_design_scores_gemma":[0.004362877,0.0013051134,0.02238611,0.0005677598,0.00034404462,0.0042248424,0.0016354638,0.005635925,0.03547796,0.49145088,0.43115696,0.0014520675],"about_ca_topic_score_codex":3.6999035e-7,"about_ca_topic_score_gemma":2.4902948e-7,"teacher_disagreement_score":0.9057798,"about_ca_system_score_codex":0.000060849026,"about_ca_system_score_gemma":0.000029443157,"threshold_uncertainty_score":0.11521206},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.019912633270328373,"score_gpt":0.24674370388935754,"score_spread":0.22683107061902916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2586621921","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99689955,0.00066440686,0.000009500491,0.0010692334,0.00023682347,0.000085602376,0.0000240323,0.000006212825,0.0010046159],"genre_scores_gemma":[0.99813646,0.00062493427,0.00011661795,0.000040889045,0.00039285579,0.0000010512074,0.00000712963,0.0000011284386,0.0006789144],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99842036,0.00008667302,0.00072038354,0.00014968312,0.00033088567,0.0002920165],"domain_scores_gemma":[0.9988835,0.00012179645,0.00053226546,0.000043249962,0.00029022817,0.00012897578],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00032483123,0.00016025064,0.00038750487,0.000044567976,0.00004077346,0.000026936543,0.00019681829,0.00011279816,0.00008867102],"category_scores_gemma":[0.00040543743,0.000046201258,0.00020500975,0.0004969817,0.00007345817,0.00020718349,0.000037754664,0.00020817203,0.0000033893127],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031482666,0.0018680536,0.30550846,0.00013342006,0.0004969877,0.0007740775,0.0074631874,0.00031902123,0.4991169,0.0027425224,0.010322651,0.16810647],"study_design_scores_gemma":[0.0040737013,0.006990119,0.781557,0.0011772605,0.00019676372,0.000363198,0.003943873,0.00018925828,0.005364778,0.008361603,0.1869363,0.000846132],"about_ca_topic_score_codex":0.000114698436,"about_ca_topic_score_gemma":0.0019031053,"teacher_disagreement_score":0.49375212,"about_ca_system_score_codex":0.00012444731,"about_ca_system_score_gemma":0.000021647627,"threshold_uncertainty_score":0.18840326},"labels":[],"label_agreement":null},{"id":"W2587002626","doi":"10.22148/16.012","title":"The Tell-Tale Hat: Surfacing the Uncertainty in Folklore Classification","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Folklore, Mythology, and Literature Studies","field":"Arts and Humanities","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Folklore; Folkloristics; Legend; Ballad; sort; History; Literature; Art; Computer science; Information retrieval","score_opus":0.06173786260510841,"score_gpt":0.295444449210424,"score_spread":0.2337065866053156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2587002626","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8662591,0.0045490814,0.000008977726,0.047923837,0.002706044,0.00017695523,0.000015868569,0.000013377605,0.07834676],"genre_scores_gemma":[0.99193007,0.00055622344,0.000016319273,0.00018081772,0.0010837977,0.0000011784521,0.0000016457275,0.0000051506654,0.0062247873],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99905455,0.00006462673,0.0004012977,0.00008503054,0.00021071131,0.0001838087],"domain_scores_gemma":[0.9985533,0.00014463194,0.00058630883,0.00025475005,0.00042989207,0.000031153144],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.00065656373,0.00011695826,0.00020562919,0.000033085867,0.0018843667,0.0009106354,0.000526461,0.000038505103,0.000028492443],"category_scores_gemma":[0.00034356807,0.00004512055,0.00016391651,0.000025954501,0.00072870083,0.00037056982,0.0000525083,0.00037676477,0.000007577529],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000339092,0.0001658583,0.01414673,0.00005968065,0.0007751339,0.00010482661,0.25248572,0.001123336,0.0004378747,0.58770716,0.12779367,0.014860916],"study_design_scores_gemma":[0.00056760537,0.00013530855,0.041696157,0.00011873109,0.00014346557,0.00005291723,0.10037059,0.0016623077,0.000022362306,0.0069337557,0.84810764,0.00018917596],"about_ca_topic_score_codex":0.0000748179,"about_ca_topic_score_gemma":0.0028177656,"teacher_disagreement_score":0.72031397,"about_ca_system_score_codex":0.000041994164,"about_ca_system_score_gemma":0.000024908215,"threshold_uncertainty_score":0.99941504},"labels":[],"label_agreement":null},{"id":"W2587195211","doi":"10.22148/16.002","title":"Measured Applause: Toward a Cultural Analysis of Audio Collections","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Diverse Musicological Studies","field":"Arts and Humanities","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Poetry; Meaning (existential); Affect (linguistics); Mode (computer interface); Psychology; Literature; Aesthetics; Art; Linguistics; Communication; Computer science; Philosophy","score_opus":0.18352083345506376,"score_gpt":0.2753292732505344,"score_spread":0.09180843979547065,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2587195211","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88653135,0.0011645679,0.0004925503,0.009068712,0.0017745686,0.0002712559,0.00030553,0.000070871894,0.100320615],"genre_scores_gemma":[0.9786636,0.00017456377,0.00011414332,0.00010791446,0.00039137568,0.0000013963923,0.0000021970268,0.0000044241874,0.02054034],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985715,0.000051008057,0.0006047435,0.000106534746,0.0004783941,0.00018778772],"domain_scores_gemma":[0.9974349,0.00011661106,0.00062187994,0.00008610155,0.0016461485,0.00009436844],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00020772264,0.00015090253,0.000635,0.00026811488,0.00026317372,0.00019217681,0.00022234338,0.00004587309,0.0019075557],"category_scores_gemma":[0.000292198,0.00006528275,0.0006555418,0.0004884096,0.00039525545,0.00034098403,0.000052886262,0.00011416174,0.000013512764],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00076234114,0.0012679695,0.009078185,0.000093624156,0.054081574,0.00012659056,0.102067344,0.0015396994,0.010274219,0.1746533,0.6404566,0.0055985535],"study_design_scores_gemma":[0.0014931603,0.0007292741,0.0045618666,0.000105127845,0.009920979,0.000018847317,0.02794868,0.00013881954,0.0004105544,0.0010751964,0.9531898,0.0004076835],"about_ca_topic_score_codex":0.00004991318,"about_ca_topic_score_gemma":0.0003330768,"teacher_disagreement_score":0.3127332,"about_ca_system_score_codex":0.00013172737,"about_ca_system_score_gemma":0.000029765453,"threshold_uncertainty_score":0.99900484},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.06962991266336825,"score_gpt":0.2568047155385784,"score_spread":0.18717480287521016,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2587326830","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6639833,0.00029865734,0.000015625119,0.005528182,0.0007525655,0.000036149868,0.000032286353,0.000019036946,0.32933423],"genre_scores_gemma":[0.89622325,0.00008611551,0.000021707596,0.00046155907,0.0010876821,1.9025175e-7,8.404369e-7,0.000008511455,0.10211017],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99910563,0.000018827577,0.00035016527,0.000062063955,0.00029636116,0.00016696379],"domain_scores_gemma":[0.99903935,0.00005634878,0.0002623926,0.0000834101,0.00046204796,0.00009642711],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00014544553,0.0001118227,0.00019584263,0.00004746917,0.00013466914,0.0004037806,0.00022330976,0.000027542483,0.0036148336],"category_scores_gemma":[0.00006138227,0.000049216516,0.0002336105,0.000020364612,0.00019363675,0.0016397253,0.000023188133,0.0001229751,0.000043702803],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000088159955,0.00012001083,0.0007632457,0.000022619004,0.0003583076,0.000120718476,0.016639454,0.000006902297,0.0002130852,0.8136779,0.15887381,0.00911576],"study_design_scores_gemma":[0.0003599781,0.00017810235,0.00012789115,0.00009544968,0.000059207352,0.00004977688,0.009419564,0.0000025097731,0.000062180385,0.010244137,0.9792648,0.00013643301],"about_ca_topic_score_codex":0.0000101890655,"about_ca_topic_score_gemma":0.00016439597,"teacher_disagreement_score":0.82039094,"about_ca_system_score_codex":0.000049563027,"about_ca_system_score_gemma":0.000023423723,"threshold_uncertainty_score":0.997296},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.03885955791862655,"score_gpt":0.2399455145696156,"score_spread":0.20108595665098905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2587367367","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9283512,0.003417362,0.00044754107,0.003226228,0.0014706405,0.0001208727,0.000031693922,0.000014778211,0.06291967],"genre_scores_gemma":[0.99278855,0.0006548262,0.00003248025,0.0001022214,0.00038074944,2.574677e-7,0.0000011368238,0.000004170447,0.0060355784],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99921674,0.0000379877,0.0003882308,0.00004479254,0.00023076496,0.000081494145],"domain_scores_gemma":[0.9987994,0.00011900297,0.00038692364,0.00004647179,0.0006141876,0.00003404528],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00027548766,0.00007603391,0.00018990845,0.000044636585,0.0001494925,0.000246578,0.00009274875,0.00002005828,0.0001625047],"category_scores_gemma":[0.00006348315,0.000029711997,0.00013240252,0.000021054635,0.0005007416,0.0006450417,0.000023826027,0.00008763529,0.0000022519434],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002634715,0.00006408877,0.00037749836,0.000044855544,0.00030972253,0.000004483296,0.03024066,0.00004488773,0.00006756649,0.93606246,0.022654824,0.009865467],"study_design_scores_gemma":[0.005678855,0.0007428109,0.0037046184,0.00036903555,0.00063849485,0.00015186319,0.07261026,0.0003430451,0.00016596813,0.12876171,0.7864199,0.00041342792],"about_ca_topic_score_codex":0.000011422758,"about_ca_topic_score_gemma":0.000019826106,"teacher_disagreement_score":0.80730075,"about_ca_system_score_codex":0.000015745261,"about_ca_system_score_gemma":0.000014111226,"threshold_uncertainty_score":0.23777574},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Character (mathematics); Agency (philosophy); Late 19th century; Art; Literature; History; Aesthetics; Sociology; Social science; Period (music); Mathematics","score_opus":0.15809227188772026,"score_gpt":0.27915554013005484,"score_spread":0.12106326824233457,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2587380683","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8554792,0.007884832,0.00071336277,0.023725096,0.003287487,0.00026014863,0.00007732506,0.000023427934,0.10854913],"genre_scores_gemma":[0.99645185,0.00085464166,0.0000448312,0.000662625,0.0009843694,3.8136798e-7,7.4142986e-7,0.0000043551863,0.000996227],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99929434,0.000034033837,0.00028180485,0.00007531548,0.00016332026,0.0001511691],"domain_scores_gemma":[0.99953496,0.000079098296,0.00017530526,0.000061014216,0.000118498894,0.00003113386],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00028926574,0.000103607694,0.0001812943,0.00007696476,0.00016124551,0.00013744964,0.00012409342,0.00003103705,0.000105869454],"category_scores_gemma":[0.000047340644,0.000037940885,0.000088088185,0.00003534726,0.00022012625,0.0003276658,0.00002135095,0.00016903356,0.0000029099217],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008124495,0.000110434936,0.002238284,0.000039901275,0.000272806,0.00008841674,0.27000973,0.0000015675075,0.0010342692,0.7072186,0.018188111,0.0007166696],"study_design_scores_gemma":[0.00225291,0.0003560841,0.04369714,0.00031565755,0.00032116193,0.0004997908,0.16513814,0.000027113783,0.00002126198,0.019370176,0.7675559,0.00044464917],"about_ca_topic_score_codex":0.0000026786918,"about_ca_topic_score_gemma":0.000040519386,"teacher_disagreement_score":0.7493678,"about_ca_system_score_codex":0.00004033956,"about_ca_system_score_gemma":0.000009650196,"threshold_uncertainty_score":0.15471844},"labels":[],"label_agreement":null},{"id":"W2593276484","doi":"10.22148/16.016","title":"Topic Modeling the Hàn diăn Ancient Classics","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Pinyin; Topic model; Pronunciation; Linguistics; Computer science; History; Natural language processing; Artificial intelligence; Chinese characters; Philosophy","score_opus":0.1623321213294615,"score_gpt":0.43606651150148973,"score_spread":0.2737343901720282,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2593276484","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8986989,0.0006660022,0.038817737,0.036756724,0.0012533127,0.00007800573,0.0000015230357,0.000011568507,0.023716176],"genre_scores_gemma":[0.9932456,0.00020203642,0.0021886458,0.00022637892,0.0011571976,2.2096225e-7,3.2689152e-7,0.0000024544408,0.0029771118],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99874073,0.00012427605,0.0003165121,0.00006792362,0.00059513916,0.00015541517],"domain_scores_gemma":[0.9984744,0.00011083971,0.00049104786,0.00014547615,0.0006783881,0.00009985106],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.001319069,0.000068159585,0.00017616461,0.000034660163,0.0014350741,0.00042549134,0.0006044455,0.000040850777,0.000050047245],"category_scores_gemma":[0.0007738468,0.000034962,0.0002705012,0.00011124664,0.00018536732,0.00036054276,0.000054539003,0.00020350901,0.0000058571286],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000090047346,0.00033725426,0.026418123,0.000030026362,0.0012888991,0.00012677557,0.055819426,0.32111603,0.0005980571,0.32453054,0.020528538,0.24911627],"study_design_scores_gemma":[0.0006736052,0.00010613438,0.017851096,0.000094720956,0.00069223077,0.000033310454,0.018466927,0.5675534,0.00005883458,0.09311102,0.30096957,0.00038915718],"about_ca_topic_score_codex":0.00016915903,"about_ca_topic_score_gemma":0.00037500766,"teacher_disagreement_score":0.28044105,"about_ca_system_score_codex":0.0000764216,"about_ca_system_score_gemma":0.00013304592,"threshold_uncertainty_score":0.99986494},"labels":[],"label_agreement":null},{"id":"W2601984134","doi":"10.22148/16.013","title":"The Migrant Letter Digitised: Visualising Metadata","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social Media and Politics","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Arts and Humanities Research Council","keywords":"Metadata; Computer science; World Wide Web; Range (aeronautics); Big data; Digital era; Data science; The Internet; Data mining; Engineering","score_opus":0.08763915689161367,"score_gpt":0.4044566234577365,"score_spread":0.3168174665661228,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2601984134","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8976845,0.0007329877,0.00014193823,0.08809258,0.003792689,0.00009539876,0.000010980393,0.000012966226,0.009435958],"genre_scores_gemma":[0.9905639,0.0006006485,0.00011783498,0.0007410672,0.0040444895,2.7210507e-7,6.4941855e-7,0.0000051513334,0.0039259354],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986589,0.00010989529,0.00031468982,0.000054725984,0.00058411784,0.0002776731],"domain_scores_gemma":[0.9983744,0.00028093765,0.0005954054,0.00018104003,0.00040262172,0.00016560487],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0005833843,0.0000772777,0.0001844353,0.00002117918,0.002318543,0.001633262,0.00075809704,0.00006047753,0.000015637099],"category_scores_gemma":[0.0022393356,0.00004123097,0.00020509363,0.000054660617,0.00060239685,0.0010371768,0.00004461314,0.00022002017,0.000008272735],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018986991,0.00021126041,0.055708405,0.00003811158,0.0015838555,0.0005359303,0.1512448,0.000029046054,0.0016863882,0.1875321,0.5460454,0.05519481],"study_design_scores_gemma":[0.00018005946,0.000032014086,0.0019049929,0.000031611828,0.00012959332,0.00000985516,0.012532764,0.000015254454,0.00019890285,0.0031766312,0.98169947,0.00008884056],"about_ca_topic_score_codex":0.00062486215,"about_ca_topic_score_gemma":0.0007351836,"teacher_disagreement_score":0.43565407,"about_ca_system_score_codex":0.000069136426,"about_ca_system_score_gemma":0.00013270308,"threshold_uncertainty_score":0.9994031},"labels":[],"label_agreement":null},{"id":"W2751480313","doi":"10.22148/16.015","title":"Towards a Poetics of Strangeness: Experiments in Classifying Language of Technological Novelty","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Space Science and Extraterrestrial Life","field":"Physics and Astronomy","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Arizona State University; National Aeronautics and Space Administration","keywords":"Poetics; Novelty; Strangeness; Literature; Trajectory; Philosophy; Epistemology; Art; Poetry; Physics; Theology; Astronomy; Particle physics","score_opus":0.07629002922789323,"score_gpt":0.369867382042338,"score_spread":0.29357735281444475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2751480313","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9974416,0.000092409435,0.00031592106,0.0002617546,0.00008504072,0.000034778055,0.000009016063,0.0000014106876,0.0017580738],"genre_scores_gemma":[0.9988676,0.000016386512,0.00094283966,0.000004926735,0.000102263824,3.0017475e-7,8.543021e-7,0.000002571757,0.00006229708],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9991318,0.000015312979,0.00041380688,0.00006591705,0.00024539442,0.00012779166],"domain_scores_gemma":[0.9989659,0.00001542847,0.0006982393,0.00014848447,0.00012515095,0.000046784957],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023478399,0.00008161736,0.00029769758,0.000064804546,0.00004541456,0.00004250841,0.00039238078,0.000046899437,0.00004197015],"category_scores_gemma":[0.00006773975,0.000050542345,0.0001414094,0.000093894014,0.00018325131,0.00023905512,0.000051770297,0.00015001037,7.3412735e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023491924,0.0008262679,0.38535622,0.00006758423,0.0003051589,0.000052953896,0.004970386,0.0014017159,0.5652932,0.011109065,0.00022528166,0.030157268],"study_design_scores_gemma":[0.004018676,0.0006626957,0.09578661,0.00062486617,0.0001983275,0.000012483886,0.06990395,0.0021562746,0.8218717,0.0035255088,0.00078490755,0.00045402657],"about_ca_topic_score_codex":0.00016992757,"about_ca_topic_score_gemma":0.000010496178,"teacher_disagreement_score":0.2895696,"about_ca_system_score_codex":0.0000168565,"about_ca_system_score_gemma":0.00006478679,"threshold_uncertainty_score":0.2061057},"labels":[],"label_agreement":null},{"id":"W2768038527","doi":"10.22148/16.017","title":"Scotland’s Poetics of Space: An Experiment in Geospatial Semantics","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Geographic Information Systems Studies","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Geospatial analysis; Poetics; Semantic space; Space (punctuation); Semantics (computer science); Computer science; Distributional semantics; Natural language processing; Information retrieval; Data science; Linguistics; Geography; Remote sensing; Programming language; Poetry; Philosophy","score_opus":0.05408576375419114,"score_gpt":0.3762965498804735,"score_spread":0.32221078612628234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2768038527","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9895591,0.00010819594,0.00011045761,0.0021484813,0.00054243347,0.00008898054,0.00000332078,0.0000041672415,0.00743484],"genre_scores_gemma":[0.99830663,0.0003783583,0.00075073267,0.000019831432,0.00024383362,4.0735452e-7,5.718483e-7,0.0000030026833,0.0002966506],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.9984207,0.000060364826,0.00063498825,0.000052502543,0.00064678816,0.0001846277],"domain_scores_gemma":[0.9975861,0.000028588904,0.0012677081,0.00016595563,0.00085515983,0.00009650958],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000928756,0.000087355715,0.00031791278,0.00012673653,0.000400147,0.00014788951,0.00040444272,0.00007291927,0.000025180208],"category_scores_gemma":[0.00036541052,0.00006135998,0.00012289324,0.00011141143,0.00029795448,0.00072661747,0.000051944553,0.00013945047,0.0000024873652],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011426325,0.00034712392,0.5968453,0.00010499426,0.00035299986,0.00005870981,0.3540106,0.0020612348,0.0006997571,0.041870657,0.0022209892,0.001313315],"study_design_scores_gemma":[0.003713384,0.000937653,0.37728068,0.0006631786,0.00027609334,0.000051851028,0.5698858,0.0030892144,0.0021214294,0.0032695686,0.037977323,0.0007338091],"about_ca_topic_score_codex":0.0012825744,"about_ca_topic_score_gemma":0.0031632523,"teacher_disagreement_score":0.21956466,"about_ca_system_score_codex":0.000062458574,"about_ca_system_score_gemma":0.00008845927,"threshold_uncertainty_score":0.30776477},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"History; Literature; English language; Linguistics; Psychology; Sociology; Gender studies; Art; Philosophy","score_opus":0.026618603387986293,"score_gpt":0.2437616383731717,"score_spread":0.2171430349851854,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2786709099","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9232108,0.002441172,0.00018040516,0.0005103002,0.0015426025,0.00008754307,0.000014087876,0.000014342402,0.07199872],"genre_scores_gemma":[0.99640876,0.000101639474,0.000034265027,0.00005728977,0.0013105155,2.899423e-7,0.0000036293713,0.0000042590277,0.0020793786],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990587,0.00004553299,0.0004991266,0.000043410066,0.00025394405,0.000099249046],"domain_scores_gemma":[0.9987434,0.000033601624,0.00035900215,0.00008418247,0.00075083744,0.00002899056],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00034513493,0.00007549121,0.00016696313,0.00011922033,0.0001501548,0.00010242583,0.0001438517,0.000032801083,0.00021348929],"category_scores_gemma":[0.00006703194,0.00003946801,0.00016910647,0.00008753468,0.00017058536,0.0004824158,0.0000052390824,0.00016286566,0.0000034272528],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008963648,0.00005869952,0.00007781186,0.000028403769,0.00022496296,0.0000037572681,0.961338,0.00012531404,0.00061896036,0.02141788,0.014393609,0.0016229813],"study_design_scores_gemma":[0.000952005,0.00034083417,0.00040748404,0.0000945302,0.00046697728,0.00001939208,0.23368002,0.0032082696,0.001719154,0.0015362257,0.7573453,0.00022979594],"about_ca_topic_score_codex":0.000028227385,"about_ca_topic_score_gemma":0.0005922425,"teacher_disagreement_score":0.7429517,"about_ca_system_score_codex":0.000072110604,"about_ca_system_score_gemma":0.000020253046,"threshold_uncertainty_score":0.2337558},"labels":[],"label_agreement":null},{"id":"W2789904340","doi":"10.22148/16.018","title":"Linked Reading: Digital Historicism and Early Modern Discourses of Race around Shakespeare’s Othello","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Financial Crisis of the 21st Century","field":"Economics, Econometrics and Finance","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Historicism; Ideology; Race (biology); New Historicism; The Renaissance; Literature; Reading (process); Art; Sexual difference; Aesthetics; Gender studies; Philosophy; Sociology; Psychology; Psychoanalysis; Art history; Linguistics; Politics; Political science","score_opus":0.032727857027036435,"score_gpt":0.2457932959054304,"score_spread":0.21306543887839396,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2789904340","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99017245,0.00326025,0.0009690696,0.0009543661,0.00068302534,0.000057645757,0.00004527623,0.000007363238,0.0038505364],"genre_scores_gemma":[0.9943694,0.00032274207,0.00065331924,0.00004363334,0.00066942675,2.4404238e-7,0.0000012757748,0.000017940825,0.0039220415],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985377,0.000008451666,0.00095400016,0.00017290334,0.00010972085,0.00021720602],"domain_scores_gemma":[0.99802345,0.00004207164,0.0013374989,0.00018637837,0.00028946614,0.00012111084],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00032315974,0.00016289232,0.0006282159,0.00020020634,0.00007825217,0.00015073665,0.0003158336,0.000098665456,0.000024527657],"category_scores_gemma":[0.00033019396,0.00013553072,0.00023678808,0.00018234413,0.00021407427,0.00088545366,0.000067966845,0.00021102918,0.00003127648],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001916331,0.0019211383,0.5925206,0.0004909168,0.0030032385,0.00018994344,0.056493364,0.0011741653,0.005607,0.23598579,0.07780799,0.022889545],"study_design_scores_gemma":[0.005776499,0.004931444,0.23217249,0.00077626295,0.0006198727,0.0005227264,0.0045000953,0.014315764,0.0017264469,0.336046,0.3960682,0.0025442035],"about_ca_topic_score_codex":0.000046136105,"about_ca_topic_score_gemma":0.0000060319694,"teacher_disagreement_score":0.36034808,"about_ca_system_score_codex":0.00014826549,"about_ca_system_score_gemma":0.000023101442,"threshold_uncertainty_score":0.5526782},"labels":[],"label_agreement":null},{"id":"W2790939515","doi":"10.22148/16.021","title":"Crossing Over: Gendered Reading Formations at the Muncie Public Library, 1891-1902","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Library Science and Administration","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Reading (process); Meaning (existential); Artifact (error); Originality; Horizon; Sociology; Linguistics; Computer science; Epistemology; Philosophy; Artificial intelligence; Social science; Mathematics","score_opus":0.08992471200700271,"score_gpt":0.3428549130230063,"score_spread":0.2529302010160036,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2790939515","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89376587,0.00010015893,0.0001822213,0.036361143,0.0005813828,0.000073923366,0.000005929502,0.000026432106,0.068902954],"genre_scores_gemma":[0.9835156,0.00009090428,0.0004635223,0.0009433695,0.0013425896,2.9219638e-7,0.00000430552,0.0000036845324,0.01363569],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99861836,0.00009505223,0.00038032664,0.00007648342,0.0005691045,0.00026068886],"domain_scores_gemma":[0.9989754,0.00008343188,0.00044049945,0.00010054968,0.00022789433,0.00017222113],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0005915544,0.000077180084,0.0001152532,0.00006322566,0.002485282,0.0018168548,0.00043543166,0.000056093115,0.0012753195],"category_scores_gemma":[0.0002012447,0.000042349067,0.0001369206,0.00061288505,0.0007529712,0.0054896604,0.000059256447,0.00014265976,0.000036216024],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007285717,0.00015136963,0.054922063,0.000019742201,0.00019883439,0.000033327266,0.13507566,0.00010239254,0.00440439,0.10231067,0.6889643,0.013744361],"study_design_scores_gemma":[0.0002344909,0.00014652897,0.011830361,0.00003374842,0.00004576049,0.000049615668,0.03390595,0.0012542997,0.0013448382,0.0033396115,0.94764847,0.00016632963],"about_ca_topic_score_codex":0.000039070248,"about_ca_topic_score_gemma":0.00035787918,"teacher_disagreement_score":0.25868413,"about_ca_system_score_codex":0.00007252024,"about_ca_system_score_gemma":0.0005554058,"threshold_uncertainty_score":0.99963766},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.02039322726241128,"score_gpt":0.25319788472165305,"score_spread":0.23280465745924178,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2802843067","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8290507,0.0019882033,0.12593351,0.02650866,0.0012851955,0.0001127999,0.0000039002034,0.00020817676,0.01490882],"genre_scores_gemma":[0.9785387,0.00014204398,0.020301564,0.0003122868,0.0003618153,2.6093144e-7,3.0571346e-7,0.0000023177818,0.00034069255],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9992736,0.00003016411,0.00025744154,0.00011003151,0.00019389452,0.00013488994],"domain_scores_gemma":[0.9991573,0.000014120008,0.00020336352,0.00013714802,0.00042108516,0.00006694418],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00018500422,0.0000936621,0.0001786493,0.00009024431,0.00019849726,0.0001278552,0.00032942384,0.000070958864,0.000007790939],"category_scores_gemma":[0.00004749621,0.000060765018,0.00006194598,0.00020562501,0.00013183564,0.0005844456,0.00015308117,0.0002024027,0.000008484346],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018519184,0.0010955596,0.037107196,0.00032318407,0.0029190613,0.00090423104,0.21112299,0.00005486522,0.0019533853,0.3906214,0.11165788,0.24205506],"study_design_scores_gemma":[0.0034534438,0.003762035,0.05750791,0.0005233066,0.0005879861,0.007250868,0.004967227,0.12760997,0.0012621055,0.44834286,0.34322444,0.001507844],"about_ca_topic_score_codex":0.0000011592177,"about_ca_topic_score_gemma":0.000011211314,"teacher_disagreement_score":0.24054721,"about_ca_system_score_codex":0.000025614501,"about_ca_system_score_gemma":0.000027123475,"threshold_uncertainty_score":0.24779256},"labels":[],"label_agreement":null},{"id":"W2885197710","doi":"10.22148/001c.19102","title":"Reassembling the English Novel, 1789–1919","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Historical Economic and Social Studies","field":"Economics, Econometrics and Finance","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"History; Period (music); Population; Literature; Genealogy; Classics; Demography; Sociology; Art; Aesthetics","score_opus":0.06666583389974345,"score_gpt":0.2473028550808614,"score_spread":0.18063702118111793,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2885197710","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71302176,0.034273237,0.007657364,0.01873839,0.007144548,0.00013772072,0.00007936349,0.000037449143,0.21891017],"genre_scores_gemma":[0.9913029,0.0014581518,0.00047855065,0.0005312065,0.0013505046,3.3981004e-7,0.0000015034959,0.000008443629,0.004868373],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99894965,0.000010871018,0.00071723975,0.0001109298,0.000049958267,0.00016132441],"domain_scores_gemma":[0.9987588,0.00008119906,0.0006113603,0.000107296015,0.00036852012,0.000072839386],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041871294,0.000092772294,0.00041116882,0.000038944374,0.00020974547,0.000091548485,0.00019888082,0.00006191851,0.00012875294],"category_scores_gemma":[0.000861031,0.00006489331,0.0003072806,0.00023079905,0.000057225232,0.00021041682,0.000046425852,0.00025541033,0.0000447104],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004769877,0.00042504875,0.021588305,0.000047250836,0.001719372,0.00009301018,0.02589235,0.003436438,0.00021622813,0.84234256,0.10187223,0.0023195168],"study_design_scores_gemma":[0.00047682883,0.000044311557,0.00322294,0.000017006121,0.000048668004,0.000031648004,0.0050721397,0.0005674907,0.000044362565,0.008572821,0.98171395,0.00018783052],"about_ca_topic_score_codex":0.000032430344,"about_ca_topic_score_gemma":0.000020597972,"teacher_disagreement_score":0.87984174,"about_ca_system_score_codex":0.00020928876,"about_ca_system_score_gemma":0.000035914672,"threshold_uncertainty_score":0.26462725},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.036973478959737065,"score_gpt":0.2909724707646574,"score_spread":0.2539989918049203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2895759845","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06972648,0.0005460573,0.92302865,0.0042106067,0.0015338127,0.00031422084,0.000018019107,0.00025126062,0.00037088842],"genre_scores_gemma":[0.23610817,0.00022417783,0.7582468,0.00016048856,0.0018023516,0.000010676279,0.000014875402,0.00001531408,0.0034171487],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9986131,0.000039625163,0.00054387026,0.00021704928,0.00035138082,0.00023498044],"domain_scores_gemma":[0.99785244,0.00008139112,0.00057756825,0.00032834488,0.0010821372,0.00007810834],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00031184038,0.00015891885,0.0003436482,0.00013846565,0.00030596353,0.000310086,0.0006470841,0.0000852128,0.000009039355],"category_scores_gemma":[0.00034792745,0.00009982896,0.0001745555,0.00051415834,0.00022379952,0.0034732008,0.00018270545,0.00017849909,0.0000051734155],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022309253,0.00057529856,0.00017890177,0.00013021094,0.00083438074,0.000092307666,0.0023696746,0.001266608,0.07589925,0.547153,0.32102785,0.048241593],"study_design_scores_gemma":[0.005280604,0.0017419698,0.00016185416,0.00013531471,0.00019031895,0.0013059988,0.0007220225,0.033740193,0.28635108,0.33591327,0.33376417,0.0006932069],"about_ca_topic_score_codex":0.0000029912987,"about_ca_topic_score_gemma":0.0000017261261,"teacher_disagreement_score":0.21123973,"about_ca_system_score_codex":0.00009995425,"about_ca_system_score_gemma":0.00012213619,"threshold_uncertainty_score":0.4070907},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.03466716520478621,"score_gpt":0.27180729049283825,"score_spread":0.23714012528805203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2898247218","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9566483,0.0021452534,0.000049265098,0.003974696,0.0010691129,0.00020497099,0.000036778794,0.000034118395,0.03583752],"genre_scores_gemma":[0.99328566,0.0008373363,0.00014473096,0.00084171724,0.002482681,4.909822e-7,0.0000043275954,0.000009433707,0.0023936084],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99883926,0.000071475144,0.00047362395,0.00010223377,0.00030395642,0.00020945986],"domain_scores_gemma":[0.99828535,0.000061028997,0.00040193083,0.00011453404,0.0010177859,0.00011934662],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003107919,0.00018048345,0.00032814505,0.00007564925,0.00040537008,0.000439501,0.0002331043,0.000056693625,0.00069616444],"category_scores_gemma":[0.00008336929,0.000080066915,0.00027249902,0.00013286856,0.001498723,0.0004548945,0.000038178063,0.0003422766,0.000015863085],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024071373,0.0005466904,0.00094559405,0.00015041458,0.0037062955,0.00014724214,0.19760533,0.00003696238,0.010461575,0.39738384,0.3838261,0.0027828238],"study_design_scores_gemma":[0.004399883,0.00092894165,0.0009871245,0.00014670087,0.0013474668,0.00026601274,0.054005504,0.0031867102,0.0007252256,0.003125365,0.9303656,0.00051545183],"about_ca_topic_score_codex":0.000045722856,"about_ca_topic_score_gemma":0.00016143952,"teacher_disagreement_score":0.54653955,"about_ca_system_score_codex":0.000016093576,"about_ca_system_score_gemma":0.000025491514,"threshold_uncertainty_score":0.7622512},"labels":[],"label_agreement":null},{"id":"W2899525700","doi":"10.22148/16.027","title":"Measuring Modernist Novelty","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Contemporary Literature and Criticism","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Novelty; Contradiction; Character (mathematics); Literature; Aesthetics; Philosophy; Art; Epistemology; Psychology; Social psychology; Mathematics","score_opus":0.12620595233161896,"score_gpt":0.2681152115589517,"score_spread":0.14190925922733272,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2899525700","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.405646,0.0014144086,0.0005739941,0.0043010977,0.0024264483,0.00007460293,0.000023415128,0.000036090965,0.58550394],"genre_scores_gemma":[0.98127794,0.000008824765,0.00013485733,0.0005000383,0.004170364,1.4054821e-7,0.0000015156455,0.0000063178422,0.013899978],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99919283,0.000017079949,0.00035011614,0.00006184572,0.0002512314,0.000126875],"domain_scores_gemma":[0.99868214,0.000016765169,0.00018414835,0.00007802386,0.00096249103,0.00007641307],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00014654465,0.00009587327,0.00018805319,0.00005890926,0.000191109,0.00031566873,0.00015537483,0.000025872698,0.0005560174],"category_scores_gemma":[0.00003753502,0.000058102392,0.00016139918,0.000030038065,0.00016959316,0.0005385805,0.000018365983,0.00017031336,0.000037329428],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021719588,0.00024518397,0.00025018275,0.00006827377,0.0005407706,0.00017857603,0.113763206,0.000017296927,0.0018023567,0.76244766,0.115290694,0.0051786206],"study_design_scores_gemma":[0.00037853865,0.0002809291,0.00009226045,0.00010386629,0.00010964445,0.00009565157,0.004582976,0.00046145095,0.00057833974,0.0047021154,0.9884481,0.00016614498],"about_ca_topic_score_codex":0.0000089662,"about_ca_topic_score_gemma":0.00012098609,"teacher_disagreement_score":0.8731574,"about_ca_system_score_codex":0.000025347506,"about_ca_system_score_gemma":0.000029725248,"threshold_uncertainty_score":0.60880005},"labels":[],"label_agreement":null},{"id":"W2902056984","doi":"10.22148/16.029","title":"Detecting Footnotes in 32 million pages of ECCO","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Enlightenment; German; Philosophy; Literature; Art history; History; Classics; Art; Epistemology; Linguistics","score_opus":0.040543892597873006,"score_gpt":0.2681849783307519,"score_spread":0.2276410857328789,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2902056984","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97308356,0.00015658383,0.023106407,0.0010497151,0.00047126945,0.00003674043,4.0495814e-7,0.0000091332,0.0020861675],"genre_scores_gemma":[0.9948922,0.000036842575,0.004679129,0.00013211957,0.00016797526,8.884482e-8,1.0628754e-7,0.0000011236532,0.0000904125],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988552,0.000032237378,0.0006016908,0.000057773734,0.00033247453,0.000120655655],"domain_scores_gemma":[0.9988692,0.00004616049,0.00056916295,0.00011128522,0.00035140716,0.000052779294],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00049638643,0.000060562154,0.000170814,0.00018884543,0.00004693564,0.00012601221,0.00057672255,0.000032161628,0.0000069573343],"category_scores_gemma":[0.00011225039,0.00003617136,0.00008089798,0.000690416,0.000054312095,0.0037629083,0.00006612392,0.00009488322,0.000009582528],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016173377,0.00041259773,0.23488966,0.0003888521,0.00035641785,0.0001371364,0.1861382,0.008677197,0.098574385,0.020512348,0.035892714,0.41385874],"study_design_scores_gemma":[0.0022560554,0.0031697715,0.10323222,0.0010775198,0.00002622546,0.0008942247,0.021086127,0.44467324,0.38450924,0.00641171,0.03182895,0.00083472166],"about_ca_topic_score_codex":0.000006217721,"about_ca_topic_score_gemma":0.000005683429,"teacher_disagreement_score":0.43599606,"about_ca_system_score_codex":0.000020516045,"about_ca_system_score_gemma":0.000047199817,"threshold_uncertainty_score":0.27280176},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Reading (process); Computer science; Scale (ratio); Information retrieval; Natural language processing; Artificial intelligence; Linguistics; Geography; Philosophy; Cartography","score_opus":0.052274167746249496,"score_gpt":0.3675912135698836,"score_spread":0.3153170458236341,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2905726793","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06720544,0.0008205953,0.92152405,0.009285451,0.00032078015,0.00025233676,0.000009424491,0.0001255306,0.000456418],"genre_scores_gemma":[0.5978286,0.000008924825,0.40128928,0.0003540416,0.00034617318,0.0000012921944,8.90703e-7,0.0000060492985,0.0001647375],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984361,0.00004422658,0.0005446535,0.00019217805,0.00041792408,0.00036493537],"domain_scores_gemma":[0.99772394,0.00012524685,0.0002279249,0.0001698928,0.0016152033,0.00013778156],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000888284,0.00013695682,0.00029296157,0.00020430029,0.00011765993,0.00026164338,0.0008480266,0.0001045515,0.0000084366575],"category_scores_gemma":[0.0011267469,0.00009049823,0.00013282661,0.00057556486,0.000119082615,0.0010486778,0.0001452344,0.00035348634,0.0000037795471],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000729073,0.0016376149,0.011412455,0.0012350654,0.00032622903,0.0011837778,0.043784868,0.00021048586,0.091040894,0.5926408,0.2174502,0.038348492],"study_design_scores_gemma":[0.0041231243,0.0040338705,0.0018769794,0.002367647,0.0002079781,0.004589188,0.0058934735,0.58139926,0.14892378,0.23307052,0.011866811,0.0016473479],"about_ca_topic_score_codex":0.0000042739666,"about_ca_topic_score_gemma":0.000021087768,"teacher_disagreement_score":0.5811888,"about_ca_system_score_codex":0.000115225375,"about_ca_system_score_gemma":0.00015548355,"threshold_uncertainty_score":0.3690411},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Race (biology); Computation; Racial differences; Sociology; Computer science; Gender studies; Ethnic group; Anthropology; Programming language","score_opus":0.03399302375473869,"score_gpt":0.35616223997810736,"score_spread":0.3221692162233687,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2909609490","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9869877,0.00085207605,0.004219784,0.004925872,0.00017291281,0.0000746933,0.0000014061353,0.0000055126397,0.0027600299],"genre_scores_gemma":[0.9937555,0.00029453306,0.0031167367,0.0002521264,0.00043324023,2.4430648e-7,8.6776276e-7,0.0000026568523,0.002144051],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99880695,0.00019615491,0.0003472099,0.000089919085,0.00043829912,0.000121466044],"domain_scores_gemma":[0.9984861,0.000585452,0.00036476774,0.0000364602,0.00043495256,0.000092254406],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001140888,0.000082777995,0.0002834364,0.000054887816,0.00031601367,0.0002352485,0.00012801019,0.000042441483,0.000053652562],"category_scores_gemma":[0.00023498865,0.000045125867,0.00011195139,0.00025546463,0.00033767958,0.00020576379,0.00003771484,0.00018100443,0.0000032215187],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010056576,0.00038469603,0.20837478,0.00012351651,0.0017434595,0.000046955658,0.16032529,0.021782894,0.0012894536,0.36096242,0.0026011171,0.24135977],"study_design_scores_gemma":[0.011824044,0.00044921873,0.5540549,0.00031364599,0.001570266,0.00036762442,0.036931276,0.2098789,0.000047593025,0.07553462,0.1079978,0.0010301231],"about_ca_topic_score_codex":0.00010317171,"about_ca_topic_score_gemma":0.000108156586,"teacher_disagreement_score":0.34568012,"about_ca_system_score_codex":0.000035315516,"about_ca_system_score_gemma":0.000100503,"threshold_uncertainty_score":0.24305536},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Hollywood; Hierarchy; Variety (cybernetics); Decentralization; Sociology; History; Political science; Computer science","score_opus":0.08365479870504956,"score_gpt":0.35254696572743793,"score_spread":0.2688921670223884,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2913116882","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9822398,0.00040894042,0.000018013412,0.0010331149,0.0005531756,0.000082493476,0.0000017257387,0.0000046001,0.015658157],"genre_scores_gemma":[0.9976857,0.00013595732,0.000023948463,0.00015281158,0.0010960232,2.1126701e-7,0.0000018103276,0.0000038828766,0.0008996949],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9987032,0.00018852403,0.00040473926,0.000064455366,0.0004850006,0.00015411216],"domain_scores_gemma":[0.9991041,0.00008597008,0.00043098838,0.000055360382,0.00026671385,0.000056874538],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008613788,0.00007024124,0.00021808248,0.000077922654,0.00011166379,0.000045306493,0.00020368527,0.000071233524,0.00006937956],"category_scores_gemma":[0.00022731538,0.00004174218,0.0001585101,0.00025222867,0.000119767494,0.00041913046,0.000016666152,0.00027353878,0.000003007982],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008540724,0.00009023529,0.02708348,0.00004114864,0.0001228055,0.00002403517,0.9591008,0.000101880476,0.0035615817,0.0023884077,0.0046947086,0.002705526],"study_design_scores_gemma":[0.0017587355,0.00014711221,0.06108411,0.0001426707,0.00008467921,0.000012238106,0.9133847,0.0004301406,0.0007762582,0.0012790253,0.020607645,0.00029268136],"about_ca_topic_score_codex":0.00010465754,"about_ca_topic_score_gemma":0.000043743028,"teacher_disagreement_score":0.045716077,"about_ca_system_score_codex":0.000091261565,"about_ca_system_score_gemma":0.00017854576,"threshold_uncertainty_score":0.17021967},"labels":[],"label_agreement":null},{"id":"W2915547389","doi":"10.22148/16.033","title":"Could Fiction Have an Information History? Statistical Probability and the Rise of the Novel","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Meaning (existential); Phenomenon; History of literature; Literary science; Psychology; History; Epistemology; Literature; Linguistics; Sociology; Literary criticism; Art; Philosophy","score_opus":0.050218066700435195,"score_gpt":0.3391083479877508,"score_spread":0.2888902812873156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2915547389","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9620988,0.00020042369,0.024643842,0.005185244,0.0005824642,0.00027728023,0.000011730794,0.0000059874687,0.006994234],"genre_scores_gemma":[0.9960218,0.000030524676,0.0032899468,0.00008371404,0.00009136061,3.1101237e-7,0.000001146468,0.0000010365897,0.00048014228],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987969,0.00026323076,0.00036589612,0.00003953645,0.00046990343,0.00006455664],"domain_scores_gemma":[0.9984718,0.00026715756,0.00049225386,0.000066672,0.00065537496,0.00004676835],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016894791,0.000047574074,0.00016030812,0.000028257438,0.00012038043,0.00004022826,0.00014674719,0.000036238405,0.00010458419],"category_scores_gemma":[0.0006085772,0.000021086713,0.00010464772,0.00011253748,0.00037104357,0.00049684907,0.000020035413,0.00014986214,9.5042157e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007318025,0.0003714245,0.045406796,0.000104948536,0.0004931988,7.49732e-7,0.1087632,0.029508227,0.00038772944,0.76759964,0.006753482,0.039878767],"study_design_scores_gemma":[0.0037100664,0.0003327771,0.26368842,0.00008195913,0.0010408342,0.000033235014,0.034262795,0.28876242,0.000045016153,0.073027894,0.33468932,0.00032525248],"about_ca_topic_score_codex":0.00031017998,"about_ca_topic_score_gemma":0.00016130925,"teacher_disagreement_score":0.6945718,"about_ca_system_score_codex":0.0001341638,"about_ca_system_score_gemma":0.00013312751,"threshold_uncertainty_score":0.13671255},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Historiography; Biography; Scale (ratio); History; Classics; Literature; Media studies; Art history; Sociology; Art; Geography; Cartography; Archaeology","score_opus":0.04836581654379901,"score_gpt":0.3157815727627241,"score_spread":0.2674157562189251,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2931020970","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95893294,0.010516496,0.000027221888,0.0028901673,0.0016316276,0.00014730719,0.000086293134,0.000013739088,0.0257542],"genre_scores_gemma":[0.9967351,0.0008910772,0.00034910816,0.00038374527,0.0011426896,5.4783453e-7,0.00001609593,0.0000037709253,0.0004778419],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99846166,0.00014847939,0.00038308525,0.00013113492,0.00073698163,0.00013864457],"domain_scores_gemma":[0.9986678,0.000054066626,0.00047365372,0.00014844757,0.00057498325,0.00008106625],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007702224,0.000103371356,0.00019711985,0.00021548767,0.00031015422,0.000018660721,0.00062631,0.0000830076,0.00013317812],"category_scores_gemma":[0.000112914524,0.000062455496,0.00013472617,0.0005632973,0.0010097998,0.0006039396,0.00006305732,0.00024581852,8.072851e-7],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023163251,0.0002579672,0.10779009,0.000048853144,0.00028220384,0.00001567689,0.16730115,0.00005789287,0.002289804,0.009190369,0.7113062,0.001228177],"study_design_scores_gemma":[0.00029615522,0.00008531546,0.030319603,0.00003120725,0.00006712016,0.000035108947,0.006334158,0.000052028805,0.0000047235926,0.001683969,0.9609998,0.0000908099],"about_ca_topic_score_codex":0.00016852116,"about_ca_topic_score_gemma":0.0027174454,"teacher_disagreement_score":0.24969362,"about_ca_system_score_codex":0.0004422808,"about_ca_system_score_gemma":0.00017416065,"threshold_uncertainty_score":0.37206492},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.2958110940572425,"score_gpt":0.4945049306874361,"score_spread":0.19869383663019363,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2942659792","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2551705,0.008317033,0.00014712787,0.72816485,0.004839561,0.00048197765,0.00016270616,0.000024007255,0.0026922321],"genre_scores_gemma":[0.75316525,0.07811005,0.012792681,0.03250051,0.008068154,0.0000010692735,0.00071150216,0.000078626865,0.11457219],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.996368,0.0005570363,0.0006779697,0.00026182155,0.001738906,0.00039625668],"domain_scores_gemma":[0.9969367,0.00034580557,0.0006977452,0.0008510137,0.00089870143,0.00027006897],"candidate_categories":["research_integrity"],"consensus_categories":[],"category_scores_codex":[0.00694487,0.00017734058,0.0004021524,0.00007862064,0.0006283489,0.0008953108,0.0031283104,0.0003929548,0.00031287695],"category_scores_gemma":[0.00474091,0.00009664485,0.00015331313,0.0006598193,0.00023067517,0.0049512247,0.00061529025,0.0025678829,0.00005551444],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000056424164,0.00014180814,0.004452419,0.000045738332,0.00042622932,0.000017758262,0.3111159,0.00023854914,0.00031196736,0.013630213,0.6670938,0.00246919],"study_design_scores_gemma":[0.00027699998,0.000018592573,0.00041060374,0.00028015947,0.0001430312,0.0000072598486,0.081637435,0.00017875881,0.000004122609,0.008923724,0.9079359,0.00018338692],"about_ca_topic_score_codex":0.0015007164,"about_ca_topic_score_gemma":0.01955016,"teacher_disagreement_score":0.69566435,"about_ca_system_score_codex":0.000113293354,"about_ca_system_score_gemma":0.0012276205,"threshold_uncertainty_score":0.9997332},"labels":[],"label_agreement":null},{"id":"W2945313780","doi":"10.22148/16.037","title":"De-Agentializing Data Practices: The Shifting Power of Metaphor in 1990s Discourses on Data Mining","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Metaphor; Big data; Resource (disambiguation); Drill; Data science; Power (physics); Crude oil; Business; Computer science; Engineering; Petroleum engineering; Data mining","score_opus":0.17349251402455415,"score_gpt":0.3861726535368184,"score_spread":0.21268013951226425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2945313780","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98694676,0.0010023026,0.005137448,0.0055311997,0.00054962974,0.00007743202,0.000009007238,0.000008571905,0.0007376653],"genre_scores_gemma":[0.9790309,0.00009446977,0.020450873,0.00024780244,0.00008709406,9.269584e-8,0.0000040452974,0.0000037978728,0.00008092665],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99847835,0.00013347533,0.00052773213,0.0002085391,0.0004519041,0.00020000737],"domain_scores_gemma":[0.996766,0.0005784896,0.0013771133,0.0011191944,0.000118814365,0.00004042164],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018631595,0.00010659345,0.00029237982,0.000090149144,0.000062672676,0.000335557,0.0037746213,0.000037940466,0.000013573005],"category_scores_gemma":[0.0014933611,0.000053681568,0.00006395654,0.0003224303,0.000058720063,0.001971374,0.0008844652,0.00026083796,0.000003718432],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011068862,0.0030149682,0.44206777,0.0008608487,0.0067510055,0.0020859581,0.17534427,0.038731646,0.037098452,0.06423875,0.117724955,0.110974476],"study_design_scores_gemma":[0.0023282985,0.00094004243,0.07513421,0.00169529,0.0009337244,0.0008985742,0.10073121,0.7697379,0.002931385,0.0014077236,0.042387594,0.0008740733],"about_ca_topic_score_codex":0.000075042175,"about_ca_topic_score_gemma":0.00008392266,"teacher_disagreement_score":0.7310062,"about_ca_system_score_codex":0.000024397961,"about_ca_system_score_gemma":0.00013808481,"threshold_uncertainty_score":0.70142484},"labels":[],"label_agreement":null},{"id":"W2945654368","doi":"10.22148/16.038","title":"Towards A Queer Futurity of Data","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Focus Groups and Qualitative Methods","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Manifesto; Queer; Sociology; Public relations; Focus (optics); Gender studies; Political science; Law","score_opus":0.2116111665954851,"score_gpt":0.48081158512408123,"score_spread":0.2692004185285961,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2945654368","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94030565,0.0012612136,0.0032411164,0.0072788447,0.0010523925,0.00013221712,0.00004466343,0.000009419452,0.046674497],"genre_scores_gemma":[0.96911716,0.00034821304,0.026846237,0.00010947264,0.000569104,7.8472326e-8,0.0000026164548,0.0000043324726,0.0030027512],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99839836,0.00037350057,0.0004227604,0.000084912324,0.0005790156,0.00014143331],"domain_scores_gemma":[0.9984264,0.00014468325,0.00048393544,0.00018034366,0.00066730386,0.00009736091],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003117963,0.000068520676,0.00030166446,0.000048073434,0.000065752276,0.00003869478,0.00057729956,0.00005944255,0.00023035299],"category_scores_gemma":[0.000622872,0.00004162372,0.00013654589,0.0002669188,0.0001351012,0.0006505598,0.00006909769,0.00018960482,0.000010128789],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034868505,0.0005103324,0.030027805,0.00025359908,0.0011219824,0.000042118896,0.17110789,0.00021623734,0.006269716,0.62189996,0.051248245,0.116953395],"study_design_scores_gemma":[0.0010911067,0.0005507858,0.016169276,0.000169296,0.00029585065,0.00001839657,0.09741191,0.0008332473,0.0008484924,0.018746806,0.86349547,0.00036937252],"about_ca_topic_score_codex":0.00039167106,"about_ca_topic_score_gemma":0.000162289,"teacher_disagreement_score":0.8122472,"about_ca_system_score_codex":0.000049034814,"about_ca_system_score_gemma":0.00017132351,"threshold_uncertainty_score":0.25222036},"labels":[],"label_agreement":null},{"id":"W2951029771","doi":"10.22148/16.041","title":"Data Set Failures and Intersectional Data","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Open source software; Software; Computer science; Set (abstract data type); Open source; White (mutation); Data science; Sociology; World Wide Web; Internet privacy; Programming language","score_opus":0.22462252172857636,"score_gpt":0.45039096312592525,"score_spread":0.2257684413973489,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2951029771","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.85063875,0.000918973,0.0001383464,0.12422897,0.0019315733,0.00015668481,0.00042213683,0.000021095504,0.021543505],"genre_scores_gemma":[0.99548656,0.00082967157,0.0005909184,0.00046732082,0.0008576536,2.3849045e-8,0.000039724444,0.0000037283428,0.0017244099],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989295,0.00008321556,0.00024100368,0.00011662572,0.00048829167,0.00014138689],"domain_scores_gemma":[0.9987435,0.00015387562,0.00024549486,0.00025409166,0.0004615548,0.00014152568],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014935206,0.000063430794,0.00016680735,0.00004049711,0.0002001803,0.0003254639,0.00089853094,0.00008517875,0.00014721327],"category_scores_gemma":[0.0011246138,0.000043501204,0.000036859285,0.00012077144,0.00015698723,0.0020006911,0.0002998422,0.00033415132,0.000011068931],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019493715,0.00022950624,0.041903704,0.000104202,0.0015089149,0.00008933613,0.07526557,0.00019324903,0.0011420874,0.1420313,0.7274227,0.009914492],"study_design_scores_gemma":[0.00038467898,0.00011837887,0.0029436583,0.000071758426,0.00012599595,0.000033951208,0.050700013,0.001261649,0.0000061078035,0.006587523,0.9375955,0.00017077773],"about_ca_topic_score_codex":0.00068096025,"about_ca_topic_score_gemma":0.0025812297,"teacher_disagreement_score":0.2101728,"about_ca_system_score_codex":0.000037789847,"about_ca_system_score_gemma":0.00020533781,"threshold_uncertainty_score":0.31384557},"labels":[],"label_agreement":null},{"id":"W2962799820","doi":"10.22148/16.040","title":"Archival Circulation on the Web: The Vine-Tweets Dataset","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital and Traditional Archives Management","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Vine; World Wide Web; Circulation (fluid dynamics); Library science; Computer science; Engineering; Biology; Ecology","score_opus":0.05410923311520564,"score_gpt":0.23831636898506325,"score_spread":0.1842071358698576,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2962799820","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.62864035,0.000015909067,0.000010163709,0.020905009,0.0005971926,0.00015868682,0.00040440157,0.00000804032,0.3492602],"genre_scores_gemma":[0.993518,0.00001932804,0.0000061391174,0.0013891119,0.0006966598,6.153043e-7,0.00010316695,0.000004385596,0.004262602],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991979,0.000034889064,0.00024392734,0.00006249483,0.0003586032,0.00010219411],"domain_scores_gemma":[0.999405,0.00015469118,0.00020807066,0.00012497169,0.000073706615,0.00003353322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00015510271,0.000088165616,0.000107704196,0.000039597642,0.00017291703,0.00028859766,0.0002835903,0.0000042705924,0.00042502012],"category_scores_gemma":[0.000019766863,0.000033149172,0.00013707721,0.000029659295,0.000119512384,0.00029069732,0.000036935533,0.00017620451,0.00022221562],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003644902,0.000075376425,0.00006684967,0.000010517069,0.00018416066,0.000007783945,0.0014952723,0.0013951658,0.000035506546,0.86877215,0.1265632,0.0013575444],"study_design_scores_gemma":[0.00027618697,0.0002393913,0.016493805,0.0000667395,0.00010652174,0.000027015203,0.0019309678,0.0023455492,0.000009688884,0.039050054,0.93932784,0.00012626356],"about_ca_topic_score_codex":0.0000043425875,"about_ca_topic_score_gemma":0.000007711826,"teacher_disagreement_score":0.8297221,"about_ca_system_score_codex":0.00001545465,"about_ca_system_score_gemma":0.000014275736,"threshold_uncertainty_score":0.46536723},"labels":[],"label_agreement":null},{"id":"W2962827351","doi":"10.22148/16.042","title":"TED Talks as Data","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Educational Tools and Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Event (particle physics); Currency; Inclusion (mineral); Political science; Telecommunications; History; Computer science; Sociology; Gender studies; Linguistics; Philosophy; Physics","score_opus":0.17891380582912028,"score_gpt":0.4915616461961003,"score_spread":0.31264784036698,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2962827351","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8078258,0.0005926427,0.000121326775,0.0153984735,0.0020055438,0.00010223304,0.000009636185,0.000011116983,0.17393324],"genre_scores_gemma":[0.91189504,0.00066483684,0.009693711,0.00067114213,0.001811854,1.4921383e-7,0.0000113780625,0.0000058311557,0.07524605],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990132,0.00010278301,0.00024376727,0.00007421265,0.0004378356,0.00012818868],"domain_scores_gemma":[0.998891,0.00013480039,0.0002587432,0.00015023141,0.00043899554,0.00012624617],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011027508,0.000051276096,0.00014459826,0.000031823536,0.00010096292,0.000087077904,0.0005154965,0.000048843758,0.0007886895],"category_scores_gemma":[0.0008197467,0.000033242603,0.000072347306,0.00021214836,0.000050222105,0.0005322381,0.00004344733,0.00016255083,0.00009323486],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001900202,0.00076663494,0.0629693,0.00008181465,0.0010325242,0.00008190089,0.036993008,0.0009164061,0.009700141,0.2718472,0.5056348,0.10978628],"study_design_scores_gemma":[0.00015775097,0.000051626223,0.0035565528,0.000027448541,0.00005376196,0.000020673959,0.009133806,0.00010904196,0.00005979764,0.0032960929,0.9834533,0.00008019167],"about_ca_topic_score_codex":0.00013567878,"about_ca_topic_score_gemma":0.000053947195,"teacher_disagreement_score":0.47781846,"about_ca_system_score_codex":0.000055036133,"about_ca_system_score_gemma":0.0003510116,"threshold_uncertainty_score":0.8635597},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scholarship; Framing (construction); Style (visual arts); Computer science; Sociology; Political science; Geography; Visual arts; Art","score_opus":0.030486235782564937,"score_gpt":0.3704559229220717,"score_spread":0.33996968713950676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2962933257","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97947305,0.0006534133,0.000016626325,0.005059681,0.00044353044,0.000065778266,2.2390212e-7,0.0000043812247,0.01428329],"genre_scores_gemma":[0.99647033,0.0011738463,0.00033786782,0.00021162201,0.00059484673,3.0049569e-7,6.6610284e-7,0.0000024590422,0.0012080518],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990332,0.00012438986,0.0002984001,0.000046125348,0.00032212143,0.00017576477],"domain_scores_gemma":[0.99928814,0.00011070663,0.00025388715,0.000072091,0.00020845045,0.00006674958],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006888166,0.000054657543,0.00019056065,0.000030616553,0.00012711622,0.00004776641,0.0002417747,0.000023734734,0.00013138729],"category_scores_gemma":[0.000104090264,0.000037068166,0.000090222944,0.000294796,0.00005414003,0.00020584706,0.000042175852,0.00024299398,0.000028396482],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021785543,0.0004609124,0.751902,0.00003493784,0.00041914827,0.000059028094,0.07654774,0.019563274,0.00046116833,0.0765556,0.04357783,0.030200498],"study_design_scores_gemma":[0.003756531,0.0006193025,0.15041572,0.00043233423,0.00021049313,0.000026841399,0.16001715,0.011889453,0.000048378122,0.008358775,0.66352224,0.0007027549],"about_ca_topic_score_codex":0.0002590399,"about_ca_topic_score_gemma":0.0021686212,"teacher_disagreement_score":0.61994445,"about_ca_system_score_codex":0.00009804107,"about_ca_system_score_gemma":0.000062518426,"threshold_uncertainty_score":0.1511596},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scholarship; Narrative; Race (biology); Criminology; Sociology; Oral history; Law; Political science; History; Gender studies; Anthropology; Art; Literature","score_opus":0.007586517233021723,"score_gpt":0.30242183974918596,"score_spread":0.29483532251616423,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2963370929","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96516114,0.0007984515,0.00015193222,0.010864574,0.0014082179,0.00024850407,0.000037559927,0.000011677251,0.021317955],"genre_scores_gemma":[0.977868,0.012311068,0.002872365,0.00023456209,0.0009858531,0.0000015476619,0.000005035517,0.000005419103,0.005716137],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9986863,0.00029621247,0.00028522153,0.000091973394,0.0004657517,0.0001745148],"domain_scores_gemma":[0.99870306,0.00018443144,0.00046791488,0.000064551175,0.00046329116,0.00011676822],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000673921,0.0000914423,0.00017341526,0.000006659189,0.001064971,0.00020436397,0.0001832741,0.00006432087,0.000113962415],"category_scores_gemma":[0.00009545254,0.00004703856,0.00014827943,0.00013801492,0.0005984776,0.00050050457,0.000031702624,0.00047294947,0.000002345527],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029228325,0.00009909587,0.3242927,0.000023548673,0.001963024,0.000015948304,0.42926762,0.00093526463,0.0002514978,0.05209095,0.18144388,0.0093241865],"study_design_scores_gemma":[0.00069129217,0.0003341433,0.022293814,0.000053721677,0.00019861487,0.000052648047,0.6725146,0.00026060233,0.000019235966,0.030549485,0.2728663,0.00016552207],"about_ca_topic_score_codex":0.00017767012,"about_ca_topic_score_gemma":0.000383209,"teacher_disagreement_score":0.30199888,"about_ca_system_score_codex":0.00014988784,"about_ca_system_score_gemma":0.00015703063,"threshold_uncertainty_score":0.8191003},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Indigenous; Geography; History; Data science; Computer science; Biology","score_opus":0.1595948908674861,"score_gpt":0.26622637286834067,"score_spread":0.10663148200085457,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2968217872","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9782867,0.0010100224,0.00000573741,0.001301025,0.0002711361,0.00015467149,0.00003291737,0.000006078248,0.018931724],"genre_scores_gemma":[0.9885315,0.00019807785,0.0000766016,0.00013142083,0.0003739805,2.3605287e-7,0.000027542677,0.0000041065373,0.010656576],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99919176,0.00004374606,0.00036952816,0.00006534459,0.00023995714,0.00008966485],"domain_scores_gemma":[0.99898934,0.00008053652,0.0003841203,0.00017493682,0.00034750003,0.000023546812],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003113819,0.000085678854,0.00025462636,0.00003133334,0.00012566225,0.0001853724,0.0003833132,0.000017624177,0.00038963978],"category_scores_gemma":[0.000040018796,0.00003844487,0.000067689005,0.000042466945,0.000230537,0.0006244159,0.00013857623,0.00012059575,0.0000059622935],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007522794,0.00016197309,0.0052153394,0.0005380962,0.0012372069,0.000016076687,0.2736377,0.00023292993,0.00058681733,0.6704766,0.042801145,0.0043437975],"study_design_scores_gemma":[0.0049445834,0.0005448488,0.008885056,0.00026917373,0.00076716597,0.000056114488,0.24154331,0.006052597,0.0000754667,0.011926805,0.72448003,0.0004548344],"about_ca_topic_score_codex":0.00012437956,"about_ca_topic_score_gemma":0.0003654854,"teacher_disagreement_score":0.6816789,"about_ca_system_score_codex":0.000008801249,"about_ca_system_score_gemma":0.000008808009,"threshold_uncertainty_score":0.42662823},"labels":[],"label_agreement":null},{"id":"W2968503355","doi":"10.22148/16.045","title":"Performative Data: Cultures of Government Data Practice","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"E-Government and Public Services","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Performative utterance; Government (linguistics); Computer science; Art; Aesthetics; Linguistics; Philosophy","score_opus":0.07885178304177048,"score_gpt":0.39105278712180563,"score_spread":0.31220100408003515,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2968503355","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.66654086,0.0021615336,0.000092491064,0.052446395,0.0020481409,0.00041135668,0.000698687,0.000019695608,0.27558082],"genre_scores_gemma":[0.99018717,0.0018286481,0.0019246853,0.00068272144,0.0007333056,1.3147299e-7,0.000044319007,0.000005216469,0.0045938278],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971977,0.0001922168,0.00048450712,0.00013040187,0.0018220718,0.00017309762],"domain_scores_gemma":[0.9973993,0.00033806803,0.0012310887,0.0004548007,0.0004697442,0.00010702047],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023903844,0.00009765499,0.00025962098,0.000018703902,0.0001421377,0.00015111805,0.0018516206,0.00006610918,0.00037045416],"category_scores_gemma":[0.0011945341,0.00006016811,0.000063840154,0.00022984638,0.00012399137,0.0051248395,0.0004126991,0.0002775203,0.000023001023],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008117576,0.0011850107,0.06284987,0.00032896496,0.003131214,0.00004323645,0.16563904,0.0005095043,0.0012276629,0.051311914,0.7009012,0.012060625],"study_design_scores_gemma":[0.0003420817,0.000108795204,0.0013197992,0.000057334273,0.00022196342,0.000010840338,0.24888551,0.00091173325,0.00008408855,0.0000801194,0.7478714,0.00010631334],"about_ca_topic_score_codex":0.00042455958,"about_ca_topic_score_gemma":0.00045261031,"teacher_disagreement_score":0.32364628,"about_ca_system_score_codex":0.00013677504,"about_ca_system_score_gemma":0.0001413527,"threshold_uncertainty_score":0.40562132},"labels":[],"label_agreement":null},{"id":"W2977975511","doi":"10.22148/16.046","title":"The Canon of Dutch Literature According to Google","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Canon; Key (lock); Amazon rainforest; Literature; World Wide Web; History; History of literature; Art; Computer science","score_opus":0.03527795122692926,"score_gpt":0.25082351017514626,"score_spread":0.215545558948217,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2977975511","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9102051,0.0007463814,0.000001309351,0.00089444156,0.0007961166,0.00006317233,0.000015419622,0.0000041648746,0.08727385],"genre_scores_gemma":[0.95288837,0.00005175715,0.000021643193,0.00022851952,0.0006154741,2.4269e-7,0.0000017981477,0.0000066697435,0.046185546],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990695,0.000019275329,0.00039815714,0.00005723344,0.00030099193,0.00015482453],"domain_scores_gemma":[0.998719,0.000079297344,0.00028881093,0.00011079672,0.0007316599,0.00007047899],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023796965,0.00009845857,0.00021796436,0.00005932772,0.00016277363,0.00088262995,0.00030031454,0.000025803474,0.00020429124],"category_scores_gemma":[0.00006774244,0.000047596488,0.00020225997,0.00006260209,0.000057864985,0.00066828745,0.000041386884,0.00022604773,0.000027080088],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036386694,0.00013656708,0.0042711045,0.00014413892,0.0005394634,0.000049720275,0.09904726,0.00048058748,0.0011852018,0.82150584,0.062050536,0.010225688],"study_design_scores_gemma":[0.000344406,0.00059042167,0.0005213014,0.00038647288,0.00007493294,0.000046838413,0.058116294,0.000053191572,0.00033347355,0.0061149094,0.933218,0.00019976308],"about_ca_topic_score_codex":0.00001412344,"about_ca_topic_score_gemma":0.00015958925,"teacher_disagreement_score":0.8711675,"about_ca_system_score_codex":0.000035169676,"about_ca_system_score_gemma":0.000029072091,"threshold_uncertainty_score":0.8511221},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.022835250776226435,"score_gpt":0.23880331165707125,"score_spread":0.2159680608808448,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2978801148","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96420753,0.0030718811,0.00006113436,0.0030489592,0.0007787096,0.000071662485,0.000057298163,0.000015020601,0.028687825],"genre_scores_gemma":[0.9942238,0.0002667779,0.00007843381,0.00028849996,0.0017304376,2.042912e-7,0.000002186399,0.000003937807,0.0034057363],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992676,0.000041357715,0.0003498849,0.000051718904,0.00020836493,0.00008108049],"domain_scores_gemma":[0.99826777,0.00005374294,0.00031456942,0.000051850104,0.0012767541,0.000035301182],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002982369,0.00006193849,0.00017527318,0.00005515408,0.000222115,0.00021082282,0.000106090716,0.000032496453,0.00011934199],"category_scores_gemma":[0.000064446955,0.000036472848,0.00013813528,0.000064872795,0.0007400205,0.0002054233,0.000014616785,0.00018283605,0.000001053233],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046451582,0.00028737233,0.006802016,0.0002091713,0.001408069,0.000033141867,0.1323117,0.000063221494,0.0002109142,0.35708275,0.48657796,0.014549156],"study_design_scores_gemma":[0.0030673777,0.0005470925,0.027478686,0.0005253283,0.00053102616,0.0005856379,0.01839867,0.00054960453,0.00012443321,0.012885342,0.935012,0.00029480073],"about_ca_topic_score_codex":0.00016531633,"about_ca_topic_score_gemma":0.00024779205,"teacher_disagreement_score":0.44843403,"about_ca_system_score_codex":0.000005469132,"about_ca_system_score_gemma":0.000011862421,"threshold_uncertainty_score":0.27266362},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"English language; Computer science; Non-fiction; Resilience (materials science); Fragility; History; Linguistics; Literature; Art; Philosophy","score_opus":0.07261985779052788,"score_gpt":0.2707456352499021,"score_spread":0.1981257774593742,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2979474259","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7058294,0.00361364,0.00076593267,0.011165639,0.005655103,0.0007503128,0.0076933154,0.00021528108,0.26431137],"genre_scores_gemma":[0.9878774,0.000030715775,0.00016094922,0.001757169,0.0054817633,9.614287e-7,0.00016915197,0.000014309895,0.0045075547],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99901503,0.00001314211,0.0004520359,0.00009495837,0.00025189435,0.00017292165],"domain_scores_gemma":[0.9988045,0.000055013097,0.00030058544,0.000079506644,0.0006070635,0.00015330529],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00014079663,0.00013228766,0.00024799258,0.00004422431,0.00019034988,0.0007363522,0.00025660868,0.000031490556,0.0007397504],"category_scores_gemma":[0.00030134656,0.00008750244,0.00023746176,0.000038614522,0.00008070723,0.0011752866,0.000028775434,0.00020963258,0.0000133260655],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017262882,0.00010978993,0.000034750898,0.000086812644,0.00024681294,0.000042161148,0.113931224,0.0001209023,0.000116433956,0.062380668,0.8211472,0.0016106396],"study_design_scores_gemma":[0.00052616117,0.0003828309,0.000015702768,0.000029397874,0.00011802907,0.000011934884,0.04807862,0.00023905376,0.000091642985,0.00039347564,0.9499643,0.0001488402],"about_ca_topic_score_codex":0.0000060587226,"about_ca_topic_score_gemma":0.00003945186,"teacher_disagreement_score":0.28204805,"about_ca_system_score_codex":0.00002667174,"about_ca_system_score_gemma":0.000027381939,"threshold_uncertainty_score":0.8099748},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.0752353978742625,"score_gpt":0.2643746749599136,"score_spread":0.18913927708565112,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2985141934","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97192144,0.00043813538,0.00010060255,0.010166446,0.0013864756,0.00059041264,0.0005711118,0.00002459931,0.0148007795],"genre_scores_gemma":[0.9654097,0.0000062786794,0.000041163174,0.00052211556,0.0021997003,0.0000041225094,0.000024069781,0.000012512373,0.031780314],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.999051,0.000010319753,0.00039958154,0.00012697962,0.0002546539,0.00015750052],"domain_scores_gemma":[0.9984675,0.00010066624,0.00027169863,0.00009910279,0.0010012246,0.000059816022],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0001736801,0.00015223908,0.00023399977,0.00009006075,0.00038252745,0.0023781317,0.00018099071,0.000026335545,0.00078006176],"category_scores_gemma":[0.00012189372,0.000085660795,0.0001465485,0.00003030345,0.00013218638,0.0020907472,0.000037649545,0.0001555894,0.000020276202],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004260858,0.00010394737,0.00020522656,0.00008785954,0.0006535939,0.000006080457,0.3686572,0.00033513838,0.000247246,0.5494363,0.07542585,0.0044154595],"study_design_scores_gemma":[0.00063718355,0.0010140301,0.0010126705,0.00007422518,0.00011035271,0.000043431064,0.16382806,0.00023636402,0.000030723437,0.0066282,0.8261178,0.0002669625],"about_ca_topic_score_codex":0.0000057963002,"about_ca_topic_score_gemma":0.00008442579,"teacher_disagreement_score":0.75069195,"about_ca_system_score_codex":0.000038853814,"about_ca_system_score_gemma":0.000017222148,"threshold_uncertainty_score":0.9986575},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.3084406244997505,"score_gpt":0.31523065243641046,"score_spread":0.006790027936659948,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2985974955","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97947806,0.0008807393,0.000021616645,0.0008212765,0.0004616847,0.00027671497,0.0002334207,0.0000068624586,0.017819602],"genre_scores_gemma":[0.9863748,0.000048313563,0.000039211714,0.00013484184,0.0005477629,0.0000018612053,0.00003299834,0.000007784608,0.012812433],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99871516,0.000014240073,0.0006721123,0.000084393345,0.00040204535,0.000112018766],"domain_scores_gemma":[0.99693215,0.0001276143,0.00054375984,0.000092189795,0.00227569,0.000028574323],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00055918045,0.00011715586,0.00021804601,0.00007588842,0.00016299306,0.0009152458,0.00015270163,0.000028202603,0.00017454225],"category_scores_gemma":[0.00030728563,0.00006321221,0.00017504889,0.000020062704,0.00015294985,0.001405691,0.00002627215,0.000106453765,0.00000366272],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003894525,0.00035144165,0.0023363591,0.00042471715,0.0020256143,0.000010338414,0.11184716,0.001379724,0.0009557154,0.75572467,0.07210046,0.048949264],"study_design_scores_gemma":[0.007272746,0.0032237156,0.010355498,0.0007787804,0.001440554,0.00012936126,0.12728067,0.024344837,0.00011851834,0.03711849,0.7871725,0.0007643027],"about_ca_topic_score_codex":0.000015056475,"about_ca_topic_score_gemma":0.00011337829,"teacher_disagreement_score":0.7186062,"about_ca_system_score_codex":0.000026939624,"about_ca_system_score_gemma":0.0000257079,"threshold_uncertainty_score":0.8825736},"labels":[],"label_agreement":null},{"id":"W2986349701","doi":"10.22148/16.047","title":"Foreword to the Special Issue “A Shared Task for the Digital Humanities: Annotating Narrative Levels”","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Digital humanities; Narrative; Task (project management); Humanities; Computer science; World Wide Web; Art; Literature; Engineering","score_opus":0.08420987353616861,"score_gpt":0.28036276194827225,"score_spread":0.19615288841210365,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2986349701","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6704832,0.0005016544,0.00016096774,0.0155140245,0.007209793,0.0015773914,0.0017999372,0.000045475863,0.30270755],"genre_scores_gemma":[0.83803135,0.000004576295,0.000035368434,0.0015000345,0.015557947,0.000005419489,0.000013156204,0.000019162811,0.14483297],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99855673,0.000021881118,0.0005643888,0.00012237705,0.000442271,0.00029236672],"domain_scores_gemma":[0.9980772,0.000383756,0.00040731797,0.00016108342,0.00090288714,0.000067740824],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0003175413,0.00020329453,0.00030321363,0.00006284175,0.00081406033,0.002769169,0.000612892,0.000032232303,0.0023783972],"category_scores_gemma":[0.00022474478,0.00009145845,0.00037014615,0.000058122547,0.00016644863,0.0014955228,0.000074701464,0.0003173551,0.00010689561],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021036301,0.00006373705,0.00006496589,0.000042648822,0.0004588321,0.00000863694,0.41385543,0.00059570523,0.000011459167,0.15433541,0.42191824,0.0084345685],"study_design_scores_gemma":[0.00027008264,0.00040189933,0.00011659459,0.00006260437,0.00005797041,0.000014047542,0.2277548,0.0002243357,0.0000056256,0.0017882723,0.7691626,0.00014116228],"about_ca_topic_score_codex":0.000009508993,"about_ca_topic_score_gemma":0.0004036749,"teacher_disagreement_score":0.34724438,"about_ca_system_score_codex":0.000070999464,"about_ca_system_score_gemma":0.000046684825,"threshold_uncertainty_score":0.99853355},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.10184203879222058,"score_gpt":0.37150933491015053,"score_spread":0.26966729611792994,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2988314392","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09434935,0.008749652,0.8791858,0.0149947805,0.001134822,0.0006680347,0.00002481141,0.00019228521,0.0007004633],"genre_scores_gemma":[0.7944111,0.000020472287,0.20376682,0.00055716996,0.0003017345,0.0000018797699,0.0000026551954,0.0000064561004,0.00093171466],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99895656,0.000011017945,0.00043127683,0.00009892352,0.00038457505,0.00011765987],"domain_scores_gemma":[0.9974825,0.00015716886,0.0005021181,0.00014921396,0.0016847177,0.000024284158],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00044008173,0.00009930227,0.00015434517,0.00006475738,0.000107426844,0.00068752875,0.0006813385,0.00003307668,0.00000864346],"category_scores_gemma":[0.0005134908,0.000049375067,0.0001491836,0.000121365774,0.0000252599,0.0014114064,0.00008158749,0.00013513143,0.0000045786137],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023085717,0.00012122965,0.0011317579,0.0002820738,0.0007565171,0.00003825226,0.01477267,0.005629469,0.045667097,0.2856811,0.030491896,0.61519706],"study_design_scores_gemma":[0.0012332265,0.0012517091,0.00045635056,0.0005597718,0.00019143909,0.00042457206,0.002257594,0.8769921,0.005303404,0.09174637,0.019003384,0.0005801012],"about_ca_topic_score_codex":0.0000020713464,"about_ca_topic_score_gemma":0.0000012149158,"teacher_disagreement_score":0.8713626,"about_ca_system_score_codex":0.00003419332,"about_ca_system_score_gemma":0.00003553005,"threshold_uncertainty_score":0.66298556},"labels":[],"label_agreement":null},{"id":"W2989723222","doi":"10.22148/16.054","title":"Annotating Narrative Levels: Review of Guideline No. 2","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Soundness; Annotation; Guideline; Narrative; Computer science; Information retrieval; Psychology; Linguistics; Political science; Artificial intelligence; Philosophy; Programming language; Law","score_opus":0.02791739552624839,"score_gpt":0.34099030708607025,"score_spread":0.31307291155982186,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2989723222","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05302538,0.33495727,0.5718695,0.027957834,0.002864063,0.0009931499,0.00002168811,0.00033910302,0.0079720095],"genre_scores_gemma":[0.104169674,0.0016658274,0.89108366,0.0019803108,0.00019620084,3.5636145e-7,0.0000017222483,0.0000072219,0.00089502236],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9982474,0.000058614583,0.00095464813,0.00011363568,0.000492907,0.00013283253],"domain_scores_gemma":[0.99536544,0.000061969266,0.0012776093,0.00020677208,0.0030295039,0.00005870233],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00080974225,0.00012045062,0.0004220772,0.00007949778,0.00003176936,0.00005040527,0.0008203741,0.0000439635,0.000057639187],"category_scores_gemma":[0.0007866957,0.00007084236,0.00018192628,0.00044617057,0.000028570748,0.00085323723,0.00011869773,0.0002734349,0.000011298548],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012058928,0.00052065944,0.0012761991,0.017978277,0.0011741868,0.00042343143,0.024319727,0.00073465507,0.2115745,0.048603624,0.49938354,0.19389063],"study_design_scores_gemma":[0.0058358484,0.007796915,0.0007092077,0.17376222,0.0010827254,0.005079494,0.008219025,0.2128436,0.26926994,0.086681746,0.22409397,0.0046253265],"about_ca_topic_score_codex":0.0000030282831,"about_ca_topic_score_gemma":4.144672e-7,"teacher_disagreement_score":0.33329147,"about_ca_system_score_codex":0.00004960259,"about_ca_system_score_gemma":0.00010644502,"threshold_uncertainty_score":0.2888868},"labels":[],"label_agreement":null},{"id":"W2990468116","doi":"10.22148/16.051","title":"Annotation Guideline No. 1: Narrative Boundaries Annotation Guide","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Topic Modeling","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Natural (archaeology); Cover (algebra); Test (biology); Computer science; World Wide Web; History; Literature; Artificial intelligence; Engineering; Art; Biology","score_opus":0.020478133666557263,"score_gpt":0.2973478058838749,"score_spread":0.2768696722173176,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2990468116","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.29405695,0.00022031684,0.6938315,0.006995599,0.0020666108,0.00012216331,0.000001795504,0.00003691427,0.0026681435],"genre_scores_gemma":[0.7345929,0.000036563382,0.25838527,0.0010439846,0.00063696544,5.42093e-7,0.000006799429,0.0000078252615,0.005289115],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983061,0.00005399783,0.0008213304,0.0001485915,0.0005103493,0.00015961117],"domain_scores_gemma":[0.9965843,0.000041150706,0.0006633011,0.00020225118,0.0024329673,0.00007603487],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00049312075,0.000121447905,0.00023650592,0.00010539707,0.0001285152,0.00044878817,0.00046258658,0.000052684078,0.00004141829],"category_scores_gemma":[0.0002838054,0.000083907325,0.00013094342,0.00027004024,0.00004002643,0.0016062789,0.00006140605,0.00018911365,0.00006651461],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021808739,0.00030626287,0.0053327903,0.00015937707,0.0008270473,0.00017846304,0.10526529,0.328229,0.05458526,0.092310466,0.3449393,0.067648634],"study_design_scores_gemma":[0.000860953,0.00038266656,0.0008123994,0.000088179775,0.00004168843,0.00014502443,0.00290741,0.7859255,0.0012587539,0.004236328,0.20305724,0.0002838262],"about_ca_topic_score_codex":0.000016876496,"about_ca_topic_score_gemma":0.0000070203364,"teacher_disagreement_score":0.4576965,"about_ca_system_score_codex":0.00011180565,"about_ca_system_score_gemma":0.00020314314,"threshold_uncertainty_score":0.43276745},"labels":[],"label_agreement":null},{"id":"W2990647120","doi":"10.22148/16.052","title":"Annotation Guideline No. 2: For Annotating Anachronies and Narrative Levels in Fiction","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Perspective (graphical); Guideline; Encoding (memory); Computer science; World Wide Web; Literature; Art; Artificial intelligence; Medicine","score_opus":0.03843314680947173,"score_gpt":0.31443756346654733,"score_spread":0.2760044166570756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2990647120","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.69117486,0.00015977981,0.30711627,0.0010263914,0.0003083888,0.0000956033,0.0000013796114,0.000006934174,0.00011039418],"genre_scores_gemma":[0.8819983,0.00001833373,0.117514804,0.000115946,0.00015325595,5.751929e-7,0.0000011232327,0.0000026210068,0.000195011],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990822,0.000026713806,0.0004956294,0.00011208728,0.00017174725,0.00011161262],"domain_scores_gemma":[0.9987392,0.00006751814,0.00034945543,0.00007565969,0.00073334156,0.00003482839],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00040593816,0.000075807104,0.00017505565,0.00009494776,0.00005099915,0.000093511466,0.00015182076,0.000034892975,0.000003819899],"category_scores_gemma":[0.00018653416,0.000054597163,0.0000507721,0.00015456071,0.000011403573,0.00095224805,0.000029860348,0.0001116274,0.0000013617284],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018636546,0.00019318663,0.015167566,0.00035391698,0.00032213525,0.0000411461,0.1073374,0.63817704,0.06926482,0.017558556,0.0128964195,0.13850142],"study_design_scores_gemma":[0.0006692309,0.00021503294,0.0029079742,0.00007416404,0.0000088940305,0.00002981663,0.002608233,0.99067277,0.00030552543,0.0013233819,0.0010909757,0.00009397588],"about_ca_topic_score_codex":0.000008575352,"about_ca_topic_score_gemma":0.000010004802,"teacher_disagreement_score":0.35249573,"about_ca_system_score_codex":0.000069784255,"about_ca_system_score_gemma":0.00004402217,"threshold_uncertainty_score":0.22264077},"labels":[],"label_agreement":null},{"id":"W2991233348","doi":"10.22148/16.053","title":"Annotating Narrative Levels: Review of Guideline No. 1.","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Guideline; Computer science; Linguistics; Political science; Artificial intelligence; Philosophy","score_opus":0.052936451256500996,"score_gpt":0.33169971006662957,"score_spread":0.2787632588101286,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2991233348","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22321554,0.034986645,0.70316935,0.02215997,0.0035094928,0.0005557259,0.000009218417,0.0000616257,0.012332439],"genre_scores_gemma":[0.48064014,0.0030634808,0.50917137,0.003746946,0.00065816834,4.787445e-7,0.0000018701772,0.000011774797,0.0027057806],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981751,0.000055094595,0.0010672517,0.00011003787,0.00046465115,0.00012787667],"domain_scores_gemma":[0.9965148,0.00005565879,0.0009977539,0.00022234042,0.0021444997,0.00006493665],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007832979,0.00010007675,0.0003947405,0.000054203792,0.000027807835,0.000033439817,0.0005992368,0.000030789928,0.000074065254],"category_scores_gemma":[0.00045189256,0.00006233641,0.00018707618,0.00026469218,0.00001651989,0.00059210754,0.000092070375,0.00019224061,0.000021452097],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013445519,0.0008417478,0.0068144836,0.020311004,0.002728767,0.00042922422,0.066505514,0.110033914,0.09061426,0.073084936,0.3668572,0.26164448],"study_design_scores_gemma":[0.0017390044,0.0010309817,0.0005858107,0.019927923,0.00020872038,0.0006064602,0.0033695297,0.8910688,0.0033114275,0.0031969456,0.074185655,0.0007687512],"about_ca_topic_score_codex":0.0000033429344,"about_ca_topic_score_gemma":5.640714e-7,"teacher_disagreement_score":0.7810349,"about_ca_system_score_codex":0.00004163756,"about_ca_system_score_gemma":0.00010163244,"threshold_uncertainty_score":0.25420052},"labels":[],"label_agreement":null},{"id":"W2992420505","doi":"10.22148/16.056","title":"Annotation Guideline No. 5: Annotation Guidelines for Narrative Levels and Narrative Acts","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Plot (graphics); Conjunction (astronomy); Guideline; Task (project management); Computer science; Narrative criticism; Natural language processing; Information retrieval; Narrative inquiry; Linguistics; Artificial intelligence; Political science; Engineering; Mathematics; Philosophy","score_opus":0.06077488116399827,"score_gpt":0.38185205362106406,"score_spread":0.3210771724570658,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2992420505","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15479103,0.00217608,0.82196295,0.019490978,0.00083824305,0.00048974185,0.0000132649675,0.00010239315,0.0001353505],"genre_scores_gemma":[0.3875033,0.00002836028,0.6105307,0.00089554256,0.00028958134,0.0000017756753,0.0000070457645,0.0000070491465,0.0007366755],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99843085,0.00004867308,0.0007877416,0.00019308651,0.00037256314,0.00016710292],"domain_scores_gemma":[0.9907376,0.0001053044,0.0009141764,0.00013384319,0.008027492,0.00008157905],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00064002175,0.00017060748,0.00031368746,0.00014168855,0.00013027029,0.00023494013,0.0003897282,0.00008539641,0.000009071528],"category_scores_gemma":[0.0013411101,0.000108926004,0.000106533684,0.00030408142,0.000036312988,0.0019335013,0.000064577594,0.00018461431,0.0000033623805],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006653574,0.00027078914,0.0006827551,0.00050796074,0.0008525361,0.000079078076,0.16238211,0.0057893437,0.34088913,0.050050538,0.32246467,0.1153657],"study_design_scores_gemma":[0.005302763,0.0044530137,0.0008873949,0.0012796177,0.00026231696,0.00076322164,0.022479897,0.6526838,0.08845371,0.17212272,0.04954567,0.0017658515],"about_ca_topic_score_codex":0.000003846198,"about_ca_topic_score_gemma":0.000004322862,"teacher_disagreement_score":0.64689445,"about_ca_system_score_codex":0.00007205738,"about_ca_system_score_gemma":0.000108186185,"threshold_uncertainty_score":0.44418734},"labels":[],"label_agreement":null},{"id":"W2992775586","doi":"10.22148/16.055","title":"Annotation Guideline No. 4: Annotating Narrative Levels in Literature","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narratology; Narrative; Interpretation (philosophy); Task (project management); Field (mathematics); Annotation; Intersubjectivity; Focus (optics); Computer science; Narrative structure; Epistemology; Linguistics; Sociology; Literature; Art; Artificial intelligence; Philosophy","score_opus":0.016682114255803713,"score_gpt":0.3160439169189949,"score_spread":0.2993618026631912,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2992775586","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.74688834,0.008619765,0.231453,0.009227328,0.0016990199,0.00042390678,0.0000106098605,0.000203249,0.0014747875],"genre_scores_gemma":[0.5920626,0.000017712166,0.40683874,0.00041602142,0.0001485968,3.2490465e-7,0.0000028513032,0.000004464422,0.0005087015],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985351,0.000066168686,0.0006821618,0.00015174046,0.0003918099,0.00017304656],"domain_scores_gemma":[0.99725574,0.00005656032,0.00063054357,0.00015228309,0.001848515,0.000056333727],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005455765,0.00013804338,0.00026868866,0.00021176881,0.000040639734,0.00028056916,0.00059381296,0.00008587293,0.000014033668],"category_scores_gemma":[0.00040272632,0.00008794599,0.00009948475,0.00081803766,0.000017210106,0.0019451309,0.0000772051,0.00047817157,0.000011186217],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033924403,0.00065856124,0.011709575,0.0007167252,0.00048693427,0.0021977767,0.25230834,0.016967423,0.42852065,0.065624535,0.0773726,0.14309765],"study_design_scores_gemma":[0.0057233656,0.0027846766,0.0064242235,0.0062803756,0.00010405675,0.0024005508,0.009951738,0.75178355,0.06277062,0.13126236,0.01800721,0.002507245],"about_ca_topic_score_codex":0.0000039241945,"about_ca_topic_score_gemma":0.0000033362198,"teacher_disagreement_score":0.73481613,"about_ca_system_score_codex":0.0000989658,"about_ca_system_score_gemma":0.00007343621,"threshold_uncertainty_score":0.35863334},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.14958929182489766,"score_gpt":0.44556380261633605,"score_spread":0.2959745107914384,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2993640618","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2754123,0.16399625,0.0005782277,0.54435873,0.0044403775,0.00051844906,0.00002693544,0.00006618686,0.0106025385],"genre_scores_gemma":[0.8947737,0.078823976,0.0025102021,0.0063212006,0.004937559,0.0000035452317,0.000004192535,0.00004776517,0.012577866],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980402,0.00016604713,0.00064347516,0.0001623183,0.00060683204,0.0003811186],"domain_scores_gemma":[0.99823046,0.00009664144,0.0005433508,0.00015862599,0.00038423607,0.0005866952],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00031704266,0.00015328573,0.0003815482,0.00008273369,0.000092895476,0.00018592662,0.00040422566,0.00011261232,0.001993747],"category_scores_gemma":[0.000114792376,0.000100181736,0.00022117216,0.00039227994,0.00007181274,0.0005068785,0.00007168676,0.00077207095,0.00042394284],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00086798746,0.00045179756,0.0050799106,0.000626872,0.0007434299,0.0021006619,0.0288665,0.00023858894,0.0006941616,0.004383584,0.50115997,0.45478654],"study_design_scores_gemma":[0.0010006467,0.00067882857,0.0011024083,0.00056666514,0.00006939725,0.00027399266,0.03233403,0.00029381807,0.00016500714,0.0004750005,0.96287036,0.00016983233],"about_ca_topic_score_codex":0.0000048146744,"about_ca_topic_score_gemma":0.000008080524,"teacher_disagreement_score":0.6193614,"about_ca_system_score_codex":0.00012200544,"about_ca_system_score_gemma":0.00006197015,"threshold_uncertainty_score":0.9989186},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Research data; Data science; Computer science; Sociology; Data curation","score_opus":0.1263592046849277,"score_gpt":0.3235582065088711,"score_spread":0.19719900182394343,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2993740281","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25250888,0.0030968913,0.000011857243,0.0025645485,0.0062947655,0.00051780563,0.004866995,0.00006479594,0.73007345],"genre_scores_gemma":[0.7220341,0.00095142855,0.00044217354,0.000978183,0.024664644,4.972925e-7,0.002137095,0.000054755612,0.24873711],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99655515,0.0000759668,0.0013394122,0.00041936716,0.0011720699,0.00043804452],"domain_scores_gemma":[0.99553233,0.00007456904,0.0012517325,0.0013028448,0.0015969308,0.0002415953],"candidate_categories":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00060106383,0.0004208313,0.0009077869,0.000144907,0.00025053532,0.0013481803,0.0033541056,0.00015100146,0.007021094],"category_scores_gemma":[0.0003482439,0.00025097062,0.00033770842,0.00013183962,0.00041179993,0.0053681624,0.0008185334,0.0007932212,0.0003070674],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018753308,0.00027185472,0.00031241897,0.0001427322,0.00082890125,0.00008504861,0.020567508,0.00013745559,0.00013308373,0.048179846,0.92782104,0.0013325562],"study_design_scores_gemma":[0.00071986945,0.00042435422,0.000070505754,0.00016936679,0.0004812406,0.00013939974,0.047802452,0.00044480021,0.000058715133,0.0013685134,0.9479311,0.00038965492],"about_ca_topic_score_codex":0.00009359731,"about_ca_topic_score_gemma":0.0006624048,"teacher_disagreement_score":0.48133633,"about_ca_system_score_codex":0.00008156996,"about_ca_system_score_gemma":0.00013788874,"threshold_uncertainty_score":0.9999943},"labels":[],"label_agreement":null},{"id":"W2993877978","doi":"10.22148/16.057","title":"Annotating Narrative Levels: Review of Guideline No. 4","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Perspective (graphical); Guideline; Computer science; Narrative review; World Wide Web; Psychology; Linguistics; Political science; Artificial intelligence; Philosophy","score_opus":0.05391653072855011,"score_gpt":0.33259020478020473,"score_spread":0.2786736740516546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2993877978","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22374782,0.03441896,0.7041692,0.02129116,0.0034634208,0.0005463459,0.0000092446935,0.000059271435,0.01229459],"genre_scores_gemma":[0.48484772,0.0029676373,0.505138,0.0037050704,0.00064545614,4.681596e-7,0.0000018697849,0.000011390767,0.0026824235],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99818695,0.000054558357,0.0010607512,0.00010916238,0.00046169982,0.00012688576],"domain_scores_gemma":[0.9965529,0.000055371194,0.000991861,0.0002203874,0.0021151071,0.000064409025],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007785962,0.00009923708,0.00039195217,0.000053731135,0.000027616245,0.000033217813,0.0005955379,0.000030525716,0.00007387405],"category_scores_gemma":[0.00045129523,0.00006177182,0.00018564824,0.00026425466,0.000016372866,0.0005892965,0.00009139079,0.0001909713,0.000020799218],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013642198,0.00084105146,0.006831062,0.020354075,0.0027362513,0.00042960883,0.06693781,0.10957013,0.09053745,0.073363245,0.367832,0.26043087],"study_design_scores_gemma":[0.001719534,0.001033931,0.00057623745,0.019857561,0.00020616093,0.0005991629,0.0033411153,0.8926197,0.0032378447,0.0031876152,0.072862156,0.0007589454],"about_ca_topic_score_codex":0.0000033013346,"about_ca_topic_score_gemma":5.5941763e-7,"teacher_disagreement_score":0.7830496,"about_ca_system_score_codex":0.00004134421,"about_ca_system_score_gemma":0.0001014849,"threshold_uncertainty_score":0.25189817},"labels":[],"label_agreement":null},{"id":"W2994106476","doi":"10.22148/16.058","title":"Annotating Narrative Levels: Review of Guideline No. 5","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Guideline; Narrative criticism; Narrative history; Narrative network; Narrative inquiry; Linguistics; Psychology; Computer science; Literature; History; Philosophy; Art; Political science","score_opus":0.030970730186825144,"score_gpt":0.34129124731195526,"score_spread":0.3103205171251301,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2994106476","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93680274,0.05366286,0.00132937,0.0034750414,0.0007672214,0.00015477228,0.000024987194,0.0000069532807,0.003776053],"genre_scores_gemma":[0.9392551,0.015321112,0.03595583,0.003103566,0.00099398,7.5184664e-7,0.00004487782,0.000016915683,0.0053078616],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989014,0.000044824756,0.0006519442,0.00008556745,0.00021025547,0.000105989806],"domain_scores_gemma":[0.99811906,0.000019530067,0.0005876852,0.00010988802,0.0011089774,0.00005487248],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00044308574,0.00009181615,0.00030830395,0.000022604592,0.000019221043,0.0000063508473,0.0001760264,0.000077861805,0.00009730485],"category_scores_gemma":[0.0010362456,0.000054457374,0.00018432806,0.00009622238,0.000068021545,0.000005058519,0.00004219512,0.0001217533,0.000008161996],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019509181,0.0001656796,0.0024331708,0.0025017469,0.00083908485,0.00002453634,0.0010295482,0.00022515493,0.5824577,0.00007459654,0.37598175,0.034071926],"study_design_scores_gemma":[0.0018883086,0.0037412858,0.0014582922,0.007512826,0.0003065009,0.00036306411,0.006565584,0.00074354274,0.053345405,0.00014572745,0.9233968,0.0005326505],"about_ca_topic_score_codex":0.0000016245981,"about_ca_topic_score_gemma":8.739743e-7,"teacher_disagreement_score":0.5474151,"about_ca_system_score_codex":0.000009176489,"about_ca_system_score_gemma":0.00007489184,"threshold_uncertainty_score":0.22207072},"labels":[],"label_agreement":null},{"id":"W2998938469","doi":"10.22148/001c.11772","title":"Annotation Guideline No. 7: Guidelines for annotation of narrative structure","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Plot (graphics); Character (mathematics); Perspective (graphical); Narrative structure; Guideline; Sequence (biology); Linguistics; Literature; History; Computer science; Artificial intelligence; Philosophy; Art; Mathematics; Political science","score_opus":0.06051591390825047,"score_gpt":0.37230084988058393,"score_spread":0.3117849359723335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2998938469","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014740775,0.0011393927,0.9615185,0.022128971,0.00024256851,0.00014316596,0.000019925328,0.000048264785,0.000018405706],"genre_scores_gemma":[0.2484837,0.000020853147,0.7495829,0.0013990102,0.00045830364,4.5063445e-7,0.000011192878,0.0000050337935,0.000038577888],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99841225,0.000031619013,0.000939187,0.00012980483,0.0003774267,0.00010968877],"domain_scores_gemma":[0.9894649,0.000050030332,0.001151753,0.0000970992,0.009158357,0.000077899305],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00024543065,0.00012659666,0.00029261562,0.000077858596,0.000057718615,0.00008119334,0.0005433166,0.00006907644,0.000006810148],"category_scores_gemma":[0.0021702782,0.000079882724,0.00014507791,0.00041164973,0.000030914372,0.0009275566,0.000050975425,0.00015806388,5.636021e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003902896,0.000079152655,0.0001434627,0.00038780723,0.0003526917,0.000034354794,0.028030021,0.012731765,0.45052242,0.014409465,0.43316132,0.05975723],"study_design_scores_gemma":[0.0015589945,0.0017982718,0.00007009529,0.00027529,0.0001703882,0.00011517376,0.0025500166,0.7369973,0.19082074,0.046754148,0.018386414,0.0005031576],"about_ca_topic_score_codex":0.0000030860533,"about_ca_topic_score_gemma":0.0000020766906,"teacher_disagreement_score":0.7242656,"about_ca_system_score_codex":0.000034137734,"about_ca_system_score_gemma":0.000096061805,"threshold_uncertainty_score":0.3257523},"labels":[],"label_agreement":null},{"id":"W2999500319","doi":"10.22148/001c.11747","title":"Annotation Guideline No. 6: SANTA 6 Collaborative Annotation as a Teaching Tool Between Theory and Practice","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Media Influence and Health","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Annotation; Guideline; Test (biology); Computer science; Literature; Library science; Mathematics education; Psychology; Art; Artificial intelligence; Medicine","score_opus":0.06641907348666143,"score_gpt":0.3674051233373484,"score_spread":0.300986049850687,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2999500319","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9307894,0.0011886943,0.0013793695,0.042177897,0.00051244686,0.0003354751,0.0000742155,0.00003414441,0.023508329],"genre_scores_gemma":[0.9776455,0.0006309093,0.003814609,0.012139551,0.0049911635,0.0000011877866,0.00003544108,0.000012452979,0.0007291618],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984984,0.00028581326,0.0006707806,0.000095058625,0.00032725537,0.0001226942],"domain_scores_gemma":[0.9966179,0.00054954865,0.0007580136,0.000044999284,0.0018964079,0.00013314534],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012672356,0.00011210582,0.0002569701,0.000053166576,0.00026881119,0.00023400063,0.00008703036,0.000037275142,0.00013504412],"category_scores_gemma":[0.0049965065,0.00007332322,0.00005308079,0.00006218732,0.000088505905,0.0013850842,0.000016675636,0.00034180415,0.00004171392],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015902037,0.00012806137,0.0010171663,0.00016935341,0.0005975817,0.00009817821,0.57230675,0.00028643507,0.0013449537,0.23496409,0.165106,0.022391206],"study_design_scores_gemma":[0.00085252296,0.0010903071,0.00034326795,0.00009515792,0.00038023238,0.000035966783,0.117203884,0.00064385356,0.00005537082,0.0031888757,0.87591827,0.00019227999],"about_ca_topic_score_codex":0.000027946999,"about_ca_topic_score_gemma":0.000013074625,"teacher_disagreement_score":0.7108123,"about_ca_system_score_codex":0.000036381116,"about_ca_system_score_gemma":0.00013741087,"threshold_uncertainty_score":0.59816456},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Binary opposition; Psychology; Characterization (materials science); Linguistics; Sociology; Gender studies; Philosophy","score_opus":0.10683158573027929,"score_gpt":0.34832036482402784,"score_spread":0.24148877909374855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3000500424","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9042105,0.00395465,0.002589642,0.0035782869,0.00096172123,0.00019533167,0.00002164389,0.000035782643,0.084452435],"genre_scores_gemma":[0.9963316,0.00017756285,0.0018155838,0.00037507052,0.00090263283,3.484619e-7,8.110698e-7,0.000008480134,0.00038793753],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9981249,0.000116709925,0.00063364924,0.00012192439,0.0007175206,0.0002853276],"domain_scores_gemma":[0.99867445,0.00009268059,0.00047106002,0.00009079715,0.00049637037,0.00017465596],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006420391,0.0001269861,0.0004089496,0.00007909884,0.00009441883,0.000053363576,0.0004508129,0.00009031422,0.00009419511],"category_scores_gemma":[0.0011265151,0.000094479285,0.00020105559,0.00040587786,0.00015067309,0.00052708323,0.00008598823,0.00036479213,0.0000039755973],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026435639,0.000050381954,0.00090685254,0.000036433605,0.00011370066,0.00007766186,0.98437166,0.008604225,0.00076885556,0.0022456471,0.0018738742,0.000924255],"study_design_scores_gemma":[0.0011485908,0.00013203184,0.00062886125,0.000079319536,0.00018341017,0.000007908106,0.9848407,0.0043353713,0.0007191774,0.0017750491,0.0058001233,0.0003494832],"about_ca_topic_score_codex":0.00013486044,"about_ca_topic_score_gemma":0.0004292149,"teacher_disagreement_score":0.092121065,"about_ca_system_score_codex":0.00017592756,"about_ca_system_score_gemma":0.00017748089,"threshold_uncertainty_score":0.38527536},"labels":[],"label_agreement":null},{"id":"W3001536763","doi":"10.22148/001c.11773","title":"Annotation Guideline No. 8: Annotation Guidelines for Narrative Levels","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Phenomenon; Set (abstract data type); Class (philosophy); Value (mathematics); Perspective (graphical); Element (criminal law); Linguistics; Mathematics education; Computer science; Psychology; Epistemology; Artificial intelligence; Philosophy","score_opus":0.1214185467744917,"score_gpt":0.3977763826092004,"score_spread":0.2763578358347087,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3001536763","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022753074,0.0008446167,0.9527644,0.043633334,0.00024012095,0.000121356126,0.00000753058,0.000077479956,0.000035874466],"genre_scores_gemma":[0.12018773,0.000024621471,0.87383884,0.004864298,0.0009043166,0.0000016215663,0.000008731329,0.000007638376,0.00016217661],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983051,0.00003792361,0.00093442365,0.00016549669,0.0004051417,0.00015191693],"domain_scores_gemma":[0.9897891,0.00006407198,0.000875493,0.00010362812,0.009051025,0.00011668252],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00043540244,0.00014831616,0.00027564232,0.00007998366,0.00010460602,0.00020562441,0.00061647274,0.00006796995,0.000007760695],"category_scores_gemma":[0.0030033905,0.000096563825,0.00016833105,0.0004337459,0.000026637104,0.0014235799,0.000059927585,0.0001809218,0.000004748842],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019194857,0.000075764896,0.000049193713,0.0001490274,0.0002332576,0.000059239363,0.019959835,0.005392575,0.099157,0.012270841,0.77744806,0.08501324],"study_design_scores_gemma":[0.0016432801,0.0016283423,0.00006560163,0.00023994579,0.00014211838,0.00016057631,0.002030762,0.82119966,0.0545569,0.0488481,0.06881979,0.00066494104],"about_ca_topic_score_codex":0.0000024232745,"about_ca_topic_score_gemma":0.0000013452525,"teacher_disagreement_score":0.81580704,"about_ca_system_score_codex":0.00005501456,"about_ca_system_score_gemma":0.00011062346,"threshold_uncertainty_score":0.39377585},"labels":[],"label_agreement":null},{"id":"W3002364789","doi":"10.22148/001c.11776","title":"Annotating Narrative Levels: Review of Guideline No. 8","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Narrative Theory and Analysis","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Narratology; Computer science; Phrase; Linguistics; Scripting language; Focalization; Natural language processing; Philosophy","score_opus":0.10665602000777705,"score_gpt":0.32159007484615904,"score_spread":0.214934054838382,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3002364789","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.34535477,0.13079382,0.0045454316,0.19609457,0.0029203028,0.0010322935,0.0005858957,0.00014327664,0.31852964],"genre_scores_gemma":[0.9729463,0.0036528155,0.0023668464,0.0107620545,0.0038436472,6.225785e-7,0.000018443296,0.000017998347,0.006391256],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983072,0.00011049207,0.0011003371,0.00007671127,0.0003033136,0.00010192212],"domain_scores_gemma":[0.9959623,0.000060666876,0.0011064758,0.000060049955,0.0027154626,0.00009504931],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00048497997,0.00012654535,0.0005563665,0.00004202231,0.00012063614,0.000040308907,0.00019096537,0.000018208044,0.003802449],"category_scores_gemma":[0.0008634045,0.00007327299,0.00037390887,0.00012571794,0.0001433874,0.00036083488,0.000023346367,0.00020115032,0.00002398769],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002608829,0.00022401438,0.00007701528,0.004218473,0.0025929085,0.000104836,0.34898147,0.00075089623,0.00721419,0.095902145,0.5348141,0.0048590796],"study_design_scores_gemma":[0.00093785586,0.0012748772,0.000011698879,0.006121175,0.0013412285,0.000030934156,0.1633366,0.0031420286,0.0015355832,0.0029022351,0.818877,0.0004887799],"about_ca_topic_score_codex":0.0000034155207,"about_ca_topic_score_gemma":0.0000058013024,"teacher_disagreement_score":0.62759155,"about_ca_system_score_codex":0.000016421609,"about_ca_system_score_gemma":0.000045685316,"threshold_uncertainty_score":0.9971082},"labels":[],"label_agreement":null},{"id":"W3002533498","doi":"10.22148/001c.11775","title":"Annotating Narrative Levels: Review of Guideline No. 7","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narratology; Narrative; Guideline; Grammar; Interpretation (philosophy); Hierarchy; Focus (optics); Linguistics; Plot (graphics); Computer science; Scope (computer science); Sociology; Philosophy; Political science; Mathematics; Programming language","score_opus":0.10099942756746239,"score_gpt":0.3415630571593329,"score_spread":0.2405636295918705,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3002533498","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01096453,0.014685289,0.9175479,0.05434497,0.00051538897,0.00013728565,0.0000046926084,0.000029561445,0.0017703875],"genre_scores_gemma":[0.30469096,0.0033451759,0.67571026,0.0147474175,0.0012010911,4.4090646e-7,0.0000017542685,0.000010492431,0.00029237894],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981909,0.000056668254,0.0010931181,0.000109777764,0.0004361597,0.00011339027],"domain_scores_gemma":[0.9968871,0.000043536984,0.00091004954,0.00013073106,0.0019115114,0.000117059644],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00047489486,0.000097994656,0.00037991378,0.000028979082,0.00003670439,0.00003401998,0.00060661533,0.00002594632,0.00003641106],"category_scores_gemma":[0.0011203588,0.00006249371,0.0001822757,0.0003329396,0.000020823672,0.0005004197,0.00010131808,0.00020447333,0.0000071157842],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010139593,0.00034368734,0.00084762485,0.012105233,0.0016113509,0.00064029515,0.09852656,0.06609725,0.045509238,0.025643637,0.5620918,0.1864819],"study_design_scores_gemma":[0.00080180995,0.0006498972,0.00009039982,0.006050444,0.00015015234,0.00020312144,0.0022741614,0.93210924,0.0022552216,0.0009788692,0.054045122,0.00039158727],"about_ca_topic_score_codex":0.0000019316733,"about_ca_topic_score_gemma":3.3432033e-7,"teacher_disagreement_score":0.866012,"about_ca_system_score_codex":0.000026340624,"about_ca_system_score_gemma":0.000099802506,"threshold_uncertainty_score":0.25484195},"labels":[],"label_agreement":null},{"id":"W3003077735","doi":"10.22148/001c.11774","title":"Annotating Narrative Levels: Review of Guideline No. 6","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Annotation; Ambiguity; Computer science; Narrative; Guideline; Class (philosophy); Framing (construction); Interpretation (philosophy); Task (project management); Information retrieval; Natural language processing; Linguistics; Artificial intelligence; Political science; History","score_opus":0.16877383920545086,"score_gpt":0.31912394516923387,"score_spread":0.150350105963783,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3003077735","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18574199,0.05659851,0.00016496069,0.040427398,0.0021334225,0.00050462276,0.00035242608,0.0000698032,0.7140069],"genre_scores_gemma":[0.9664273,0.0023751897,0.0010783733,0.014746506,0.0037869783,5.9954016e-7,0.00002034584,0.000021840673,0.01154287],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983743,0.000035200246,0.001047126,0.00006787841,0.00035968187,0.000115828116],"domain_scores_gemma":[0.9964568,0.000043912874,0.0008096605,0.000057029625,0.0025314218,0.00010117117],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00030161304,0.00012000163,0.00042149375,0.000029086948,0.0001018174,0.00015122138,0.00021676841,0.000018323484,0.0022532155],"category_scores_gemma":[0.00071723067,0.00007299798,0.0002707006,0.00005281409,0.00011455319,0.0007656585,0.00003234529,0.00022339543,0.000022244376],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000108364,0.00013095335,0.00005120368,0.003786258,0.0005838747,0.0000906749,0.15617527,0.00010457202,0.00033312253,0.07583291,0.7574433,0.0053595095],"study_design_scores_gemma":[0.00031961428,0.00062807335,0.000013168997,0.0029258567,0.00015670378,0.000020619835,0.043895967,0.00026609396,0.000075677286,0.0007543375,0.95076114,0.00018274663],"about_ca_topic_score_codex":0.000004399223,"about_ca_topic_score_gemma":0.000009722983,"teacher_disagreement_score":0.7806853,"about_ca_system_score_codex":0.00001954475,"about_ca_system_score_gemma":0.00004795746,"threshold_uncertainty_score":0.99865884},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.19506992183085015,"score_gpt":0.292525711556765,"score_spread":0.09745578972591484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3003312070","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7872122,0.0006551065,0.0000025390223,0.008581355,0.000532428,0.000069596346,0.00011279271,0.000016815136,0.20281716],"genre_scores_gemma":[0.98679614,0.00007943761,0.000016462083,0.0039203544,0.0023293996,1.9897429e-7,0.000011656456,0.000010882695,0.006835485],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998819,0.000022978436,0.00048881397,0.00012146501,0.00037039837,0.00017735927],"domain_scores_gemma":[0.9992285,0.000049440143,0.0002890542,0.00013318179,0.00020317706,0.00009667697],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00022795725,0.00013203794,0.00026439028,0.000053387364,0.00011501205,0.0008091169,0.00046492022,0.000029987006,0.0015386053],"category_scores_gemma":[0.000108619206,0.00008300057,0.000116892035,0.00005997915,0.000086577944,0.0012280336,0.00007386616,0.00040036067,0.00007494389],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044613404,0.00040868518,0.004339511,0.00019084595,0.0005112353,0.00053782103,0.17789969,0.0003458996,0.00003222658,0.1404863,0.6666762,0.008125448],"study_design_scores_gemma":[0.00038064749,0.00025529173,0.00023319831,0.00017857857,0.00004650369,0.000010464066,0.02903624,0.0005623952,0.000012326773,0.00079204515,0.96833766,0.0001546495],"about_ca_topic_score_codex":0.000018629957,"about_ca_topic_score_gemma":0.000232279,"teacher_disagreement_score":0.30166143,"about_ca_system_score_codex":0.000034756624,"about_ca_system_score_gemma":0.000033684104,"threshold_uncertainty_score":0.99937415},"labels":[],"label_agreement":null},{"id":"W3003443395","doi":"10.22148/001c.11828","title":"Are We Breaking the Social Contract?","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Political Science Research and Education","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scholarship; Rhetorical question; Argument (complex analysis); Epistemology; Humanism; Verifiable secret sharing; Virtue; Social contract; Sociology; Law and economics; Philosophy; Political science; Law; Computer science; Politics; Linguistics","score_opus":0.196493887523447,"score_gpt":0.45588442046883215,"score_spread":0.25939053294538517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3003443395","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24580443,0.00020924823,0.00020085734,0.74336165,0.00022898789,0.000060001745,0.000002567267,0.000009028599,0.010123212],"genre_scores_gemma":[0.99519354,0.00016476752,0.0000321785,0.0018634279,0.0025149477,2.1013146e-7,1.2772175e-7,0.000001481693,0.00022929344],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.998606,0.0001294027,0.00019191673,0.00004822,0.00063634367,0.00038806346],"domain_scores_gemma":[0.998777,0.00012897074,0.0002732482,0.00002337669,0.0003730251,0.0004244161],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00090649264,0.000043366057,0.000112797316,0.000014907472,0.0006117182,0.00020759745,0.0003321616,0.000036227142,0.00018755194],"category_scores_gemma":[0.002154913,0.000021829093,0.00011882011,0.00032422494,0.0003009265,0.00034328757,0.000017503326,0.00024367569,0.000016408008],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017731734,0.00029047098,0.02905706,0.000055269367,0.00031198826,0.00013471964,0.36702576,0.00017938906,0.0026478781,0.21067563,0.31923252,0.07021199],"study_design_scores_gemma":[0.00031696263,0.00010012293,0.05447185,0.000047553836,0.00008479103,0.000015688312,0.201258,0.00077262265,0.0001317184,0.008044856,0.73458326,0.00017257455],"about_ca_topic_score_codex":0.00016013363,"about_ca_topic_score_gemma":0.00013846502,"teacher_disagreement_score":0.7493891,"about_ca_system_score_codex":0.00013515951,"about_ca_system_score_gemma":0.0002775219,"threshold_uncertainty_score":0.47049037},"labels":[],"label_agreement":null},{"id":"W3003556322","doi":"10.22148/001c.11821","title":"From Reproducible to Productive","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social and Cultural Dynamics","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Sociology; Quarter (Canadian coin); Discipline; Epistemology; Set (abstract data type); Scale (ratio); Social science; Computer science; History; Geography; Philosophy","score_opus":0.06684697976209006,"score_gpt":0.38633516757554204,"score_spread":0.319488187813452,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3003556322","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9471579,0.000101171885,0.00004351117,0.03525479,0.0014499599,0.00010656738,0.000008430022,0.000013743433,0.015863933],"genre_scores_gemma":[0.97949094,0.00013203119,0.0014701794,0.00020814363,0.005429059,6.8170766e-7,0.0000013867001,0.0000054062416,0.013262188],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.99876916,0.00006251308,0.00029115018,0.00016764375,0.0005020778,0.00020743867],"domain_scores_gemma":[0.99792403,0.00002784847,0.0005544711,0.00027303075,0.001006519,0.00021409031],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00058555027,0.000092793576,0.00024149938,0.000029809911,0.0010923403,0.00040302827,0.00067013723,0.00006957441,0.00008908376],"category_scores_gemma":[0.002693116,0.000060453483,0.00017192613,0.00012868377,0.00020737029,0.00074588653,0.00006871819,0.00019091446,0.000045881276],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005347792,0.0005407159,0.042622805,0.000024358937,0.0013595213,0.00025754984,0.5260131,0.0009894224,0.014590139,0.04721503,0.26299363,0.102858946],"study_design_scores_gemma":[0.00087229925,0.00035993897,0.10697741,0.00020680932,0.0004529027,0.000018616583,0.15611199,0.00007896609,0.0020844096,0.062186576,0.6698776,0.0007724647],"about_ca_topic_score_codex":0.0021066766,"about_ca_topic_score_gemma":0.0011311698,"teacher_disagreement_score":0.40688396,"about_ca_system_score_codex":0.00013447533,"about_ca_system_score_gemma":0.00011744596,"threshold_uncertainty_score":0.84015095},"labels":[],"label_agreement":null},{"id":"W3003714380","doi":"10.22148/001c.11831","title":"Gender Dynamics and Critical Reception: A Study of Early 20th-century Book Reviews from The New York Times","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Audience measurement; Publishing; Dynamics (music); Reading (process); Order (exchange); Lemma (botany); Taste; Copying; Scale (ratio); History; Sociology; Psychology; Literature; Political science; Law; Geography; Art; Cartography","score_opus":0.12925868815169114,"score_gpt":0.39672365252175396,"score_spread":0.26746496437006284,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3003714380","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.78772956,0.07268919,0.018994844,0.11478877,0.0008582577,0.0006875457,0.000023142908,0.000030520732,0.00419819],"genre_scores_gemma":[0.98748493,0.0035584979,0.0054486315,0.001270982,0.0015334566,4.528329e-7,0.0000025669963,0.0000054587344,0.00069501234],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.99821556,0.0004998747,0.0005564004,0.00010568247,0.0005155371,0.00010695792],"domain_scores_gemma":[0.9985892,0.0004608414,0.0003586539,0.00006214954,0.0003332025,0.00019590766],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006630736,0.0000922273,0.0003619816,0.000024082588,0.00018991968,0.000107069085,0.00023589865,0.000043349362,0.0003836811],"category_scores_gemma":[0.00083529146,0.00005069475,0.00019368737,0.00036319168,0.00013534204,0.0002234391,0.000040757317,0.00022198264,0.0000059175636],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000462118,0.00089900335,0.06377897,0.00007376402,0.0021995488,0.000057322086,0.4688124,0.0013196899,0.00019897702,0.04688999,0.27106705,0.14424117],"study_design_scores_gemma":[0.002726721,0.002103804,0.092292726,0.00031192842,0.0053420393,0.000025154015,0.274756,0.020656025,0.000011863069,0.030104816,0.5708136,0.00085528276],"about_ca_topic_score_codex":0.000295999,"about_ca_topic_score_gemma":0.00023802517,"teacher_disagreement_score":0.29974657,"about_ca_system_score_codex":0.00003789488,"about_ca_system_score_gemma":0.000103350765,"threshold_uncertainty_score":0.4201039},"labels":[],"label_agreement":null},{"id":"W3004150426","doi":"10.22148/001c.11827","title":"Response by the Special Interest Group on Digital Literary Stylistics to Nan Z. Da's Study","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Stylistics; Alliance; Statement (logic); Special Interest Group; Interest group; Epistemology; Sociology; Political science; Linguistics; Philosophy; Politics; Law","score_opus":0.11116118933099015,"score_gpt":0.2785396355135806,"score_spread":0.16737844618259043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3004150426","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9669269,0.000057438716,0.000005746534,0.008540072,0.00055004394,0.00014885647,0.00017638349,0.000014819152,0.023579791],"genre_scores_gemma":[0.98308283,0.00000278826,0.000004951509,0.003912604,0.004083488,7.7377297e-7,0.000009640605,0.000015589985,0.008887324],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99856246,0.00008974649,0.00055499206,0.00013300411,0.0004617006,0.00019812062],"domain_scores_gemma":[0.9989006,0.00019378433,0.00024197921,0.00012759361,0.00030827697,0.00022778992],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.00027813128,0.00019564074,0.00028308967,0.000059261194,0.00027773983,0.0035733092,0.00051739893,0.000024973022,0.0004411351],"category_scores_gemma":[0.00038970463,0.000103134385,0.00020507666,0.0000733985,0.00013530538,0.00088894786,0.00010289206,0.00043174822,0.000083132465],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0047842083,0.00092024205,0.0003986893,0.0000152379625,0.00039595072,0.0003920379,0.19391538,0.00003781529,0.000053798038,0.05753938,0.73399854,0.007548749],"study_design_scores_gemma":[0.0004662142,0.0043375427,0.00030579275,0.00004724455,0.00006113332,0.000018674647,0.052109834,0.000020302417,0.000007277137,0.00080896995,0.9416302,0.00018684902],"about_ca_topic_score_codex":0.0000029778303,"about_ca_topic_score_gemma":0.00005734708,"teacher_disagreement_score":0.20763166,"about_ca_system_score_codex":0.000057570964,"about_ca_system_score_gemma":0.000021289392,"threshold_uncertainty_score":0.9974611},"labels":[],"label_agreement":null},{"id":"W3004189244","doi":"10.22148/001c.11824","title":"Send us your null results","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Replication (statistics); Null (SQL); Null hypothesis; Code (set theory); Statistical power; Econometrics; Computer science; Power (physics); Statistics; Data science; Mathematics; Data mining; Physics; Programming language","score_opus":0.7901292379649458,"score_gpt":0.6004316014343586,"score_spread":0.18969763653058724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3004189244","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94842243,0.0008971519,0.002469895,0.035644073,0.00074162544,0.000085070635,0.000058851285,0.000011704706,0.011669188],"genre_scores_gemma":[0.99363977,0.00035532954,0.0018938449,0.0008328196,0.0006952384,9.709656e-8,0.0000013190562,0.0000054904813,0.0025760718],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98964584,0.00013970646,0.0014241226,0.00027940536,0.0081526255,0.00035827886],"domain_scores_gemma":[0.9899043,0.0010652099,0.0010036564,0.0002484028,0.0068996893,0.00087876135],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0077225333,0.00013076706,0.00043468107,0.008206807,0.0001291892,0.0018304865,0.0019923747,0.00008524578,0.00026286612],"category_scores_gemma":[0.055487018,0.00006681694,0.00039391074,0.07469781,0.00009722919,0.00082917477,0.0002901918,0.00049611495,0.00030742562],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005385134,0.00021053001,0.03997182,0.000010980084,0.0001807652,0.0006418949,0.0020778358,0.004386955,0.0052616885,0.00047476194,0.8962834,0.04996087],"study_design_scores_gemma":[0.002612688,0.0015404208,0.06370868,0.000024982905,0.000078084726,0.0002941995,0.005124151,0.043108065,0.0028441185,0.0032363443,0.8770221,0.00040618714],"about_ca_topic_score_codex":0.000009514587,"about_ca_topic_score_gemma":0.0000027226981,"teacher_disagreement_score":0.066491,"about_ca_system_score_codex":0.0000749528,"about_ca_system_score_gemma":0.00015864697,"threshold_uncertainty_score":0.9992057},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Reductionism; Variance (accounting); Literary criticism; Epistemology; Sociology; Linguistics; Philosophy; Economics","score_opus":0.07676143944045041,"score_gpt":0.3253034171408474,"score_spread":0.24854197770039696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3004295495","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7229555,0.000109378045,0.19393508,0.082485124,0.00006731479,0.0001357048,0.0000066144603,0.000046712707,0.00025854126],"genre_scores_gemma":[0.9509771,0.000018658693,0.046113327,0.0027299044,0.000072460185,4.6964612e-7,4.1765995e-7,0.000004318076,0.00008335767],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984972,0.00020435988,0.0004917736,0.00014567153,0.00051919447,0.00014179798],"domain_scores_gemma":[0.99814385,0.00023968589,0.00044376586,0.00029596357,0.0007064937,0.0001702384],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00047971946,0.00013369214,0.0003386385,0.000109642024,0.00007782883,0.00015603603,0.001213543,0.00003896379,0.000014951277],"category_scores_gemma":[0.00096926524,0.000068103815,0.00028027841,0.0010709929,0.00009372162,0.00036658545,0.00015435586,0.0003350454,0.0000075297935],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031309356,0.00046268618,0.00035778488,0.000070748356,0.0008633867,0.00046853776,0.05947386,0.011537629,0.5478448,0.29753843,0.07103467,0.0072165234],"study_design_scores_gemma":[0.0043067853,0.01934083,0.059850223,0.0028983864,0.0009137309,0.0009448749,0.006916361,0.3437669,0.19717212,0.2848944,0.07596045,0.0030348988],"about_ca_topic_score_codex":0.0000014421067,"about_ca_topic_score_gemma":0.000001824049,"teacher_disagreement_score":0.35067266,"about_ca_system_score_codex":0.000054738768,"about_ca_system_score_gemma":0.000043033368,"threshold_uncertainty_score":0.2777193},"labels":[],"label_agreement":null},{"id":"W3004383240","doi":"10.22148/001c.11822","title":"Other people's data: humanities edition","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Digital humanities; Process (computing); Raw data; Resource (disambiguation); Computer science; Digital curation; World Wide Web; Data science","score_opus":0.1393494022932569,"score_gpt":0.2780875505455687,"score_spread":0.13873814825231182,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3004383240","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4402121,0.0010114362,0.00032238735,0.0060960995,0.00541975,0.00017415071,0.00083536597,0.000099192395,0.54582953],"genre_scores_gemma":[0.92847574,0.00007989542,0.00003280375,0.000494661,0.0064508403,4.3856488e-7,0.00001307892,0.000015360267,0.064437166],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99883157,0.000028252018,0.00046207977,0.00010534522,0.00038312047,0.0001896483],"domain_scores_gemma":[0.99868137,0.00008496119,0.00036788906,0.0002157813,0.0005844868,0.000065511114],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00027169663,0.00013826102,0.0002470786,0.000093155955,0.0002137908,0.0007103415,0.0004990957,0.00003168016,0.003462616],"category_scores_gemma":[0.00010095621,0.0000672535,0.00014210492,0.000025237383,0.0002143055,0.0034175927,0.00007699413,0.00013625405,0.00009014018],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000076227494,0.00014517944,0.0004229634,0.000036843274,0.0003010053,0.00003487358,0.015591228,0.0000038703674,0.00012657716,0.68359226,0.2924218,0.007247171],"study_design_scores_gemma":[0.0003846739,0.00018989663,0.00024518368,0.00013325029,0.00009814078,0.00004654371,0.011103513,0.000009604294,0.00003666776,0.011213458,0.9763655,0.00017353863],"about_ca_topic_score_codex":0.000024975177,"about_ca_topic_score_gemma":0.0011113535,"teacher_disagreement_score":0.68394375,"about_ca_system_score_codex":0.00005043137,"about_ca_system_score_gemma":0.000031314205,"threshold_uncertainty_score":0.9974483},"labels":[],"label_agreement":null},{"id":"W3004478131","doi":"10.22148/001c.11882","title":"Topic Modeling the Hàn diăn Ancient Classics (汉典古籍)","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"College of Humanities and Social Sciences, United Arab Emirates University; Xi’an Jiaotong University; National Endowment for the Humanities","keywords":"Pinyin; Pronunciation; Topic model; Linguistics; Computer science; History; Natural language processing; Artificial intelligence; Chinese characters; Philosophy","score_opus":0.1623321213294615,"score_gpt":0.43606651150148973,"score_spread":0.2737343901720282,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3004478131","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89958274,0.00068625476,0.0378389,0.037160885,0.0012996134,0.00008654252,0.0000018416631,0.000012891143,0.023330325],"genre_scores_gemma":[0.99296963,0.00021160142,0.0022573867,0.00024676012,0.0012252078,2.6930192e-7,4.1251158e-7,0.0000028180748,0.0030859401],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985998,0.000141235,0.00035552937,0.000077974466,0.00065178634,0.00017362993],"domain_scores_gemma":[0.9983306,0.00012147266,0.00053866423,0.00016298401,0.0007348202,0.00011148306],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.001455591,0.000077224395,0.00019826768,0.00003809879,0.0015036417,0.00046043194,0.00065297796,0.000046354322,0.000055831886],"category_scores_gemma":[0.0008231798,0.000040071765,0.00029913257,0.00012523388,0.00020454012,0.00038871585,0.000061251725,0.00022636398,0.000006631648],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010523513,0.00039112396,0.02890623,0.000035501645,0.001442871,0.00014468921,0.055408698,0.32167673,0.00069082173,0.31147042,0.021024862,0.2587028],"study_design_scores_gemma":[0.00074057706,0.00011663322,0.019915132,0.00010584991,0.0007741136,0.000037244106,0.01902127,0.57378125,0.00006901726,0.09536629,0.2896356,0.00043699626],"about_ca_topic_score_codex":0.00018898086,"about_ca_topic_score_gemma":0.00041361796,"teacher_disagreement_score":0.26861075,"about_ca_system_score_codex":0.000089014946,"about_ca_system_score_gemma":0.00014724366,"threshold_uncertainty_score":0.9997963},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Computer science; Reuse; Natural language processing; Phrase; Sequence (biology); Artificial intelligence; Engineering","score_opus":0.011587360958005563,"score_gpt":0.3090124364441694,"score_spread":0.29742507548616387,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3005998127","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18641785,0.0009082883,0.81045216,0.0016058189,0.00010236211,0.00040887142,0.000031076335,0.000061100036,0.000012440692],"genre_scores_gemma":[0.41331095,0.000022066073,0.5862447,0.00027779874,0.00004669969,0.000004354605,0.0000084254725,0.000007299993,0.00007770504],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99873817,0.000044175213,0.00038075846,0.00020044074,0.00040087377,0.00023560534],"domain_scores_gemma":[0.99840504,0.00015059755,0.00061089505,0.00026431907,0.00046015694,0.00010899968],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00044210764,0.00018318018,0.00030534773,0.00012256445,0.00007245774,0.00023231245,0.00058083236,0.000049453945,0.000015243807],"category_scores_gemma":[0.00012221388,0.00010453903,0.000090594585,0.000302438,0.000027993441,0.0007186912,0.00011988619,0.00016953367,0.0000015648787],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0051156795,0.002779701,0.046590682,0.0032657424,0.003814781,0.00489786,0.053699892,0.016047541,0.1889545,0.029031876,0.035694264,0.6101075],"study_design_scores_gemma":[0.017693609,0.005526351,0.0011630923,0.0010937107,0.0005699162,0.0027214668,0.004679005,0.9208737,0.03315221,0.009682055,0.0012948881,0.001549991],"about_ca_topic_score_codex":0.000016202112,"about_ca_topic_score_gemma":0.000010288826,"teacher_disagreement_score":0.90482616,"about_ca_system_score_codex":0.0001012225,"about_ca_system_score_gemma":0.00011052711,"threshold_uncertainty_score":0.42629778},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.034029507757561826,"score_gpt":0.32027837498356765,"score_spread":0.28624886722600584,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3008187175","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.29345214,0.002076058,0.000035515033,0.06970374,0.0023637433,0.00022806948,0.00003183209,0.000037067493,0.63207185],"genre_scores_gemma":[0.9890972,0.00070680137,0.000056614776,0.00041779006,0.002667897,1.4907343e-7,9.442134e-7,0.0000044461585,0.007048186],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.99894935,0.000060597435,0.00029876953,0.00005577702,0.00048007365,0.00015541982],"domain_scores_gemma":[0.9989662,0.00011490422,0.0003635869,0.00003536735,0.0003669001,0.0001530204],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00027813858,0.000074386844,0.00018839624,0.000020865293,0.00016924804,0.00022150164,0.00026422646,0.000052429183,0.000009965146],"category_scores_gemma":[0.0005870427,0.00003864878,0.00021492042,0.00025359378,0.00016349913,0.00025875942,0.000018433218,0.00022857926,0.0000033470985],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033299968,0.00012229604,0.00065755146,0.0000536162,0.0004193027,0.00009435,0.7447567,0.00005682576,0.0017286888,0.14610189,0.042767454,0.06290832],"study_design_scores_gemma":[0.00019549139,0.00029164995,0.0002984812,0.00011349635,0.00005970577,0.0000036927288,0.041208837,0.00001831364,0.0003485201,0.0005759493,0.9568025,0.00008336167],"about_ca_topic_score_codex":0.0000030154529,"about_ca_topic_score_gemma":0.000023912811,"teacher_disagreement_score":0.914035,"about_ca_system_score_codex":0.00004114015,"about_ca_system_score_gemma":0.00008645724,"threshold_uncertainty_score":0.21359454},"labels":[],"label_agreement":null},{"id":"W3011762003","doi":"10.22148/001c.12266","title":"What’s in a Face? Gender representation of faces in Time, 1940s-1990s","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Media, Gender, and Advertising","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"State University of New York Polytechnic Institute; State University of New York","keywords":"Representation (politics); Face (sociological concept); Context (archaeology); Reading (process); Photography; History; Visual arts; Psychology; Art; Sociology; Politics; Linguistics; Social science; Political science; Philosophy","score_opus":0.12280218420559069,"score_gpt":0.37625489868433115,"score_spread":0.25345271447874046,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3011762003","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99238974,0.0006356006,0.00008454848,0.0048947316,0.00022747264,0.00007814899,0.0000011326682,0.0000041032754,0.001684544],"genre_scores_gemma":[0.997376,0.0013443703,0.00025494775,0.00026114605,0.00022830487,2.1713838e-7,0.0000013156123,0.0000034069494,0.0005302736],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99847937,0.00015357716,0.0005355785,0.00009185107,0.0005775725,0.00016207356],"domain_scores_gemma":[0.9992411,0.00007560667,0.00035814542,0.000042699703,0.00015478785,0.00012763485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00042358245,0.00007155056,0.00027634608,0.00009386848,0.000044366217,0.000087117376,0.00020957312,0.000062777755,0.000094754294],"category_scores_gemma":[0.0005157118,0.00005380505,0.00010506393,0.00059983897,0.00008483004,0.00084375846,0.000023138677,0.00019283386,0.0000068086274],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020027316,0.00024054934,0.17881374,0.00011082776,0.00016974407,0.00021320893,0.76674306,0.019194238,0.01707086,0.0007273455,0.0078816265,0.008634505],"study_design_scores_gemma":[0.0031340776,0.0002866171,0.10241818,0.00059881376,0.00017538354,0.000023442937,0.8617318,0.011419262,0.006774413,0.004485976,0.008389873,0.0005621391],"about_ca_topic_score_codex":0.00019222588,"about_ca_topic_score_gemma":0.00029991963,"teacher_disagreement_score":0.09498873,"about_ca_system_score_codex":0.0000780075,"about_ca_system_score_gemma":0.00011595936,"threshold_uncertainty_score":0.21941063},"labels":[],"label_agreement":null},{"id":"W3011815798","doi":"10.22148/001c.12265","title":"Faces extracted from Time Magazine 1923-2014","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Face recognition and analysis","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"State University of New York Polytechnic Institute; State University of New York","keywords":"Metadata; Computer science; Sample (material); Raw data; Publishing; Artificial intelligence; Information retrieval; World Wide Web; Art","score_opus":0.026504282015613098,"score_gpt":0.24511915210157198,"score_spread":0.21861487008595887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3011815798","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.44783413,0.0021194431,0.42656657,0.11667198,0.00070163957,0.00016184343,0.00007246463,0.00025381392,0.005618121],"genre_scores_gemma":[0.9609024,0.00046884292,0.03244036,0.003536056,0.0008060018,2.9498372e-7,0.000019079222,0.000010660405,0.0018163117],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99879277,0.00005319794,0.0004572764,0.00014388816,0.00040903225,0.00014381252],"domain_scores_gemma":[0.9987333,0.00005730855,0.00041054207,0.00011198386,0.00042908368,0.00025780874],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00011195865,0.00012506395,0.0003267366,0.00006763563,0.00005369187,0.00024274949,0.00057854713,0.000049954953,0.00055855606],"category_scores_gemma":[0.00012402619,0.00008062912,0.0002799321,0.00044098083,0.00002766651,0.0007395815,0.00006983372,0.00023605286,0.0008860375],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013032726,0.00049301854,0.0023618836,0.0000547605,0.002597738,0.00089609454,0.0065634544,0.009739615,0.35047656,0.0012157738,0.53607994,0.08939082],"study_design_scores_gemma":[0.0015839987,0.0004415028,0.004748277,0.00010357036,0.0005387436,0.00013411658,0.0005390767,0.84919506,0.008945358,0.00196125,0.13115132,0.00065775064],"about_ca_topic_score_codex":0.000004598843,"about_ca_topic_score_gemma":0.0000028940465,"teacher_disagreement_score":0.8394554,"about_ca_system_score_codex":0.000023073408,"about_ca_system_score_gemma":0.000039388935,"threshold_uncertainty_score":0.9998919},"labels":[],"label_agreement":null},{"id":"W3016176758","doi":"10.22148/001c.12556","title":"Dramatic Structure and Social Status in Shakespeare’s Plays","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language, Metaphor, and Cognition","field":"Psychology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Tragedy (event); Narrative structure; Pyramid (geometry); Literature; Social structure; Scale (ratio); History; Sociology; Art; Politics; Geography; Political science; Mathematics; Law; Cartography","score_opus":0.032988609329113426,"score_gpt":0.30749944021815256,"score_spread":0.27451083088903916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3016176758","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99598694,0.00051238074,0.00006854148,0.0019815709,0.00015268446,0.000046232337,0.00002131333,0.000005908493,0.0012244151],"genre_scores_gemma":[0.99814165,0.00002989327,0.0001766102,0.0009853625,0.0005042771,2.536413e-7,0.000008354878,0.0000068648,0.00014674985],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9991097,0.00007526067,0.00035574296,0.000090942915,0.00019160523,0.0001767605],"domain_scores_gemma":[0.99947476,0.000029501454,0.00024222626,0.00003515681,0.00011667575,0.00010166676],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00008361359,0.00010035628,0.00026150414,0.00005585524,0.00004131276,0.000055349996,0.00008710704,0.00008585254,0.0006055029],"category_scores_gemma":[0.000062442974,0.00006588706,0.0000846765,0.00017585568,0.000041801086,0.00014192174,0.000015308542,0.00033607936,0.0000119947745],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022584195,0.00051403116,0.059629224,0.0005417393,0.0021121379,0.002765901,0.7764086,0.0003252089,0.019683586,0.018591646,0.062067267,0.055102207],"study_design_scores_gemma":[0.017660785,0.0025469328,0.5692193,0.0002713819,0.002022718,0.0016333613,0.35201472,0.0032847836,0.0013530297,0.020360729,0.02794914,0.0016831546],"about_ca_topic_score_codex":0.00001463375,"about_ca_topic_score_gemma":0.00003084817,"teacher_disagreement_score":0.50959,"about_ca_system_score_codex":0.000028300181,"about_ca_system_score_gemma":0.000021791868,"threshold_uncertainty_score":0.66298324},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.08545780393490782,"score_gpt":0.3457816242500205,"score_spread":0.2603238203151127,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3040858410","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8556786,0.00049884105,0.05519286,0.087692134,0.0004649278,0.0001835502,0.000014107807,0.000030567597,0.00024441275],"genre_scores_gemma":[0.9909091,0.00001759021,0.003909915,0.0048503396,0.00015368577,5.439432e-7,0.0000015948841,0.0000032572937,0.00015399038],"study_design_codex":"qualitative","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9982576,0.00022333197,0.0005660471,0.00015641122,0.0004980472,0.0002985725],"domain_scores_gemma":[0.99904084,0.000109771274,0.00033136088,0.00014539946,0.00021966673,0.0001529526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015166375,0.00012316104,0.00026864448,0.000054327804,0.000094293915,0.00022635862,0.0009412127,0.000060142513,0.000010601514],"category_scores_gemma":[0.00029168383,0.00007159697,0.00014894083,0.0010883671,0.000048039252,0.0004876753,0.0001363494,0.00065485825,0.000024438199],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010388946,0.0019963053,0.20220315,0.00030154348,0.0004377284,0.012153178,0.45711452,0.108959734,0.01938217,0.031842586,0.104720406,0.059849784],"study_design_scores_gemma":[0.002575503,0.00061523553,0.05547602,0.00014485054,0.000044543336,0.00084389717,0.00439297,0.89485735,0.0007576405,0.00288914,0.03686895,0.0005338833],"about_ca_topic_score_codex":0.000015320857,"about_ca_topic_score_gemma":0.000055312183,"teacher_disagreement_score":0.7858976,"about_ca_system_score_codex":0.00009059838,"about_ca_system_score_gemma":0.00008813373,"threshold_uncertainty_score":0.291964},"labels":[],"label_agreement":null},{"id":"W3044340157","doi":"10.22148/001c.14112","title":"Familial Places in Jim Crow Spaces: Kinship, Demography, and the Color Line in William Faulkner’s Yoknapatawpha County","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"American Environmental and Regional History","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"National Endowment for the Humanities","keywords":"Kinship; Context (archaeology); Scholarship; White (mutation); Metaphor; Anthropology; Genealogy; Agency (philosophy); Sociology; History; Ethnology; Media studies; Social science; Law; Archaeology; Political science","score_opus":0.0162316207879963,"score_gpt":0.22140808599488374,"score_spread":0.20517646520688743,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3044340157","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9920551,0.000739389,0.00005092514,0.006321817,0.00006992384,0.000107071675,0.000007856127,0.000003664363,0.0006442161],"genre_scores_gemma":[0.996439,0.0013700132,0.00031623588,0.0015446111,0.00010266793,0.0000014167726,0.0000018021065,0.0000067031006,0.00021758763],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9987019,0.000097447526,0.0004212897,0.00015985352,0.00043297224,0.00018659081],"domain_scores_gemma":[0.9993971,0.00008493351,0.00029828568,0.00006362062,0.000009175311,0.00014688616],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00029561907,0.00015404922,0.00032801466,0.000039253995,0.00007503945,0.000020524141,0.00025568707,0.000056578356,0.00008293561],"category_scores_gemma":[0.000055033564,0.00008119084,0.00012353598,0.00034900603,0.0021785405,0.00027442427,0.00010494344,0.00036913547,0.000011853858],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0045249825,0.0006260166,0.75523114,0.00008239326,0.00022577836,0.0006709974,0.040082026,0.1549926,0.009272707,0.00032947317,0.028419593,0.0055423],"study_design_scores_gemma":[0.0145199485,0.0019407626,0.67918605,0.00024171468,0.00036105214,0.00044955485,0.06880568,0.04897991,0.00032030712,0.0012989844,0.18264627,0.0012497639],"about_ca_topic_score_codex":0.00023217635,"about_ca_topic_score_gemma":0.0003675304,"teacher_disagreement_score":0.15422668,"about_ca_system_score_codex":0.00017500337,"about_ca_system_score_gemma":0.000013826392,"threshold_uncertainty_score":0.80269235},"labels":[],"label_agreement":null},{"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.04380876233099833,"score_gpt":0.26081456624981286,"score_spread":0.21700580391881452,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3085332162","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.44942558,0.0010138783,0.371239,0.17415003,0.001887353,0.0003324081,0.000017177805,0.00030098986,0.0016335737],"genre_scores_gemma":[0.951152,0.00003614496,0.046315383,0.0015248646,0.000869702,3.9798036e-7,5.9685624e-7,0.0000075762946,0.00009329316],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981041,0.00006293324,0.0006687034,0.00024571095,0.00061499875,0.00030357853],"domain_scores_gemma":[0.99815303,0.00017415127,0.000439406,0.00025864362,0.0005841444,0.00039063033],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00028361834,0.00019647538,0.00040226092,0.000076080665,0.00009416886,0.00038505706,0.0013581582,0.00006208047,0.000016812968],"category_scores_gemma":[0.00049863954,0.00013284321,0.0002944052,0.00064755575,0.000053315925,0.0005963298,0.00035431734,0.0004568165,0.000013149454],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022742995,0.0028644807,0.09566623,0.00093912863,0.0022192218,0.00859012,0.092291415,0.12510456,0.046734788,0.024309684,0.12296578,0.47808716],"study_design_scores_gemma":[0.0017234907,0.0018553428,0.022707159,0.00015190008,0.00018014234,0.0014919662,0.0014150325,0.9344832,0.0035354171,0.0106726205,0.02080879,0.00097494555],"about_ca_topic_score_codex":0.000010345333,"about_ca_topic_score_gemma":0.0000084713165,"teacher_disagreement_score":0.8093786,"about_ca_system_score_codex":0.00012591909,"about_ca_system_score_gemma":0.0001116695,"threshold_uncertainty_score":0.5417189},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.06567579411856664,"score_gpt":0.3085290246590696,"score_spread":0.24285323054050298,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3091866587","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9798921,0.002361003,0.00004048786,0.016828319,0.00003919198,0.000063182895,0.0000027239214,0.000002690697,0.0007703457],"genre_scores_gemma":[0.9973534,0.0018772187,0.00013687971,0.00024921715,0.0001188617,1.4578309e-7,3.4938975e-7,9.111364e-7,0.0002630179],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9992983,0.00010695317,0.00022393854,0.000053859716,0.00023239969,0.0000845539],"domain_scores_gemma":[0.99947774,0.000034736404,0.00020923931,0.00002493979,0.00017432478,0.00007901432],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022458252,0.00005419886,0.00018815049,0.000010627113,0.00011451368,0.000034992616,0.00013325944,0.000037345137,0.000026442203],"category_scores_gemma":[0.00033995943,0.000024184437,0.000051349532,0.00019357582,0.00080418953,0.0001733278,0.000036730366,0.00015294949,4.7764496e-7],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005552588,0.00008494874,0.067922994,0.000108642314,0.00015843178,0.000039852654,0.77556413,0.00010760256,0.0071424753,0.13355733,0.0056492486,0.00910907],"study_design_scores_gemma":[0.006828574,0.0006781253,0.4780393,0.00040503006,0.00052768417,0.000095259245,0.4243494,0.006300847,0.0008228703,0.063211806,0.01808647,0.00065466034],"about_ca_topic_score_codex":0.00040916508,"about_ca_topic_score_gemma":0.00066385296,"teacher_disagreement_score":0.41011629,"about_ca_system_score_codex":0.0000164406,"about_ca_system_score_gemma":0.000021704123,"threshold_uncertainty_score":0.296307},"labels":[],"label_agreement":null},{"id":"W3106923209","doi":"10.22148/001c.18120","title":"A Computational Approach to Urban Space in Science Fiction","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Human Mobility and Location-Based Analysis","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Space (punctuation); Techno-thriller; Fiction theory; Point (geometry); Urban space; Urban planning; Literary fiction; Geography; Computer science; Regional science; Mathematics; Literature; Art; Literary criticism; Engineering; Civil engineering","score_opus":0.0469397853434405,"score_gpt":0.3225254073441537,"score_spread":0.2755856220007132,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3106923209","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90254235,0.00008996346,0.045981,0.03314259,0.0001335103,0.00018402972,0.0000032232751,0.000020942976,0.017902367],"genre_scores_gemma":[0.99656415,0.000009202474,0.0024020683,0.00055471965,0.00031535758,4.9000874e-7,0.0000013826775,0.0000018071663,0.00015082335],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99866486,0.000077028155,0.00030668732,0.00011710134,0.00068607787,0.00014826648],"domain_scores_gemma":[0.99888664,0.000048416132,0.00015889395,0.000041727548,0.00061276683,0.00025154138],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00091656164,0.000056523942,0.0001542243,0.00015583803,0.00026530496,0.0001350016,0.0002722915,0.000030027142,0.0000381475],"category_scores_gemma":[0.0007214145,0.000044214583,0.00009058724,0.0017646049,0.00021813475,0.00036615523,0.00001706072,0.00016127525,0.000010298475],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005162221,0.00023836357,0.016464919,0.000019477571,0.00004762201,0.00000645581,0.13659713,0.81438386,0.00036517213,0.02035274,0.009914874,0.0015577816],"study_design_scores_gemma":[0.0012867859,0.000484724,0.0590757,0.000109410386,0.00023773256,0.000010973497,0.16705461,0.70473236,0.00019350977,0.003068915,0.063103355,0.0006419169],"about_ca_topic_score_codex":0.0002734556,"about_ca_topic_score_gemma":0.00016604403,"teacher_disagreement_score":0.109651476,"about_ca_system_score_codex":0.00022292607,"about_ca_system_score_gemma":0.00038625352,"threshold_uncertainty_score":0.2040538},"labels":[],"label_agreement":null},{"id":"W3113041327","doi":"10.22148/001c.18238","title":"Sermons as data: Introducing a corpus of 11,955 Danish sermons","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Media, Religion, Digital Communication","field":"Arts and Humanities","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Danish; Sermon; Corpus linguistics; Metadata; Agency (philosophy); Field (mathematics); Sociology; Resource (disambiguation); Linguistics; Social science; Computer science; Philosophy; World Wide Web; Theology","score_opus":0.12887272772949065,"score_gpt":0.29618114165925435,"score_spread":0.1673084139297637,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3113041327","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.85346735,0.0028139339,0.00026381342,0.031657465,0.0010686729,0.000281224,0.00025517138,0.000068881774,0.110123515],"genre_scores_gemma":[0.9964963,0.00053234765,0.00032321812,0.00055124925,0.0011890164,3.317926e-7,0.000070644,0.000013467998,0.00082341605],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99849737,0.000060654562,0.0007440561,0.00013381451,0.0004134569,0.00015064502],"domain_scores_gemma":[0.99777585,0.000101220336,0.00078472245,0.0004224899,0.0007431845,0.00017253705],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023877781,0.00014036128,0.00037726754,0.00006435336,0.0001543646,0.00027978385,0.0008613995,0.00003942889,0.00013522766],"category_scores_gemma":[0.0006284518,0.00009779787,0.00014647744,0.00011542428,0.00021060112,0.0009558954,0.00023444326,0.00030484147,0.000022178014],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037862227,0.0005880649,0.0032312332,0.00048939686,0.0012497108,0.00015778137,0.14459959,0.0021284036,0.008504579,0.3175353,0.51569605,0.0054412815],"study_design_scores_gemma":[0.002744216,0.0019675496,0.0013400916,0.0007712412,0.0019581332,0.0005343936,0.021277657,0.018660555,0.0035619442,0.030759586,0.9152586,0.0011660544],"about_ca_topic_score_codex":0.00011232327,"about_ca_topic_score_gemma":0.00036036584,"teacher_disagreement_score":0.39956254,"about_ca_system_score_codex":0.00003691543,"about_ca_system_score_gemma":0.00012184863,"threshold_uncertainty_score":0.39880815},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.04070182394424551,"score_gpt":0.27458966544404423,"score_spread":0.23388784149979872,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3117881903","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97642136,0.00033582633,0.000008225958,0.022588216,0.000047567482,0.000041242332,0.0000028463787,0.000010217663,0.0005445046],"genre_scores_gemma":[0.9977159,0.00043143367,0.00019597789,0.0006826632,0.0007408099,2.5851102e-7,0.0000033303493,6.287959e-7,0.00022899131],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9990359,0.00006272483,0.00039074974,0.00014026907,0.00021417972,0.00015615889],"domain_scores_gemma":[0.99922776,0.00007380922,0.00035129036,0.000022391274,0.00016255204,0.00016219699],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00018425182,0.00010625111,0.00029178782,0.0000108785025,0.000082899496,0.00006565958,0.00014747474,0.000046274617,0.000055141463],"category_scores_gemma":[0.00020288015,0.000036348985,0.000119759185,0.0005151688,0.000070246344,0.0002549664,0.00005792639,0.0003046737,0.000003291136],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004958146,0.00038227814,0.4639484,0.00005295463,0.00027054374,0.00070876296,0.009468979,0.0014389524,0.31523034,0.0002823206,0.10306907,0.10465159],"study_design_scores_gemma":[0.0006813226,0.0012843302,0.80740523,0.000077942816,0.00012215796,0.00033733324,0.012096754,0.002694948,0.002176039,0.00034773213,0.17235596,0.00042020876],"about_ca_topic_score_codex":0.00014100209,"about_ca_topic_score_gemma":0.00015860409,"teacher_disagreement_score":0.34345686,"about_ca_system_score_codex":0.000017604074,"about_ca_system_score_gemma":0.0000057217367,"threshold_uncertainty_score":0.14822686},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.041766873716279955,"score_gpt":0.339984818055101,"score_spread":0.29821794433882104,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3123995752","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98387456,0.004087089,0.00017889419,0.0050755464,0.0008205444,0.000092159484,0.000005984932,0.000020854091,0.0058443784],"genre_scores_gemma":[0.9907226,0.0058230776,0.00025515477,0.0004767884,0.0018613792,0.0000013395129,0.000014043845,0.0000072792855,0.0008383678],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9986112,0.00013970303,0.0003842045,0.00011358114,0.00043668694,0.000314615],"domain_scores_gemma":[0.9990614,0.00003963373,0.00023230333,0.000065009575,0.00043886548,0.0001627961],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00028837527,0.00011907538,0.00027701407,0.0000444758,0.00029020218,0.00019102423,0.00018792109,0.00011551503,0.000106838816],"category_scores_gemma":[0.00019297596,0.000089824,0.00022379139,0.0005764548,0.00009498228,0.0005632933,0.00003042661,0.00034528627,0.000003979451],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038528847,0.0015009318,0.0406865,0.00015440231,0.0010382547,0.0030555949,0.567637,0.17145255,0.0032242634,0.09589391,0.005596128,0.10937517],"study_design_scores_gemma":[0.0024721285,0.0001509828,0.014588547,0.00064289535,0.0004277427,0.00014511988,0.7403783,0.19555613,0.000056930247,0.0020855053,0.042324144,0.0011715693],"about_ca_topic_score_codex":0.0011656778,"about_ca_topic_score_gemma":0.0059493836,"teacher_disagreement_score":0.1727413,"about_ca_system_score_codex":0.0002571798,"about_ca_system_score_gemma":0.00012248984,"threshold_uncertainty_score":0.36629167},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Minor (academic); Hierarchy; German; History; Media studies; Literature; Sociology; Political science; Art; Law","score_opus":0.09238203772860684,"score_gpt":0.2961933228713019,"score_spread":0.20381128514269506,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3129389071","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.44367427,0.08113819,0.0005548189,0.016471462,0.0031408768,0.00009785902,0.00008866387,0.000058539143,0.45477533],"genre_scores_gemma":[0.98587584,0.00285229,0.00056218036,0.00042854654,0.0023426572,1.4702889e-7,0.000014951296,0.000010232162,0.007913144],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989668,0.000093337054,0.0004493821,0.00009074353,0.00026078997,0.00013895551],"domain_scores_gemma":[0.9982497,0.000043332933,0.0002820485,0.00007465212,0.0012766947,0.000073544485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019069217,0.00012836263,0.00023022325,0.00004023284,0.00026475833,0.0007251269,0.00011755802,0.000024214543,0.000425168],"category_scores_gemma":[0.00004700436,0.00007479249,0.00025750915,0.00006895447,0.00005265328,0.0010094381,0.000023240294,0.00028969144,0.000030188949],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003094525,0.00054113177,0.00047862556,0.00038092112,0.0034285686,0.0026780458,0.65380895,0.023660535,0.0018483624,0.13129471,0.14167501,0.039895695],"study_design_scores_gemma":[0.00040542075,0.00006294937,0.000022699263,0.00015947438,0.00024513432,0.00024124004,0.04373733,0.002867893,0.000037999507,0.0004393774,0.95159924,0.00018127004],"about_ca_topic_score_codex":0.000008327123,"about_ca_topic_score_gemma":0.00010272284,"teacher_disagreement_score":0.8099242,"about_ca_system_score_codex":0.000016604916,"about_ca_system_score_gemma":0.00002162302,"threshold_uncertainty_score":0.6992415},"labels":[],"label_agreement":null},{"id":"W3133718336","doi":"10.22148/001c.21374","title":"Images of the arXiv: Reconfiguring large scientific image datasets","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Data Visualization and Analytics","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Computer science; Metadata; Data science; Set (abstract data type); Representation (politics); Point (geometry); Range (aeronautics); Information retrieval; Scientific literature; World Wide Web","score_opus":0.03356266331921172,"score_gpt":0.30702755598187076,"score_spread":0.27346489266265905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3133718336","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24447094,0.0035221,0.7189158,0.019631064,0.0067996653,0.00024477096,0.0015121697,0.00008245497,0.004821013],"genre_scores_gemma":[0.9797845,0.00021610371,0.015552901,0.000560828,0.00020761722,1.8472569e-7,0.00007253112,0.0000092075825,0.003596146],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99856144,0.00008671716,0.0005453725,0.00015605937,0.0004765443,0.00017385701],"domain_scores_gemma":[0.9978196,0.000041800093,0.0005880053,0.0005028255,0.0009596376,0.000088122804],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006087896,0.00009609632,0.00021768227,0.000075884665,0.00017550802,0.0005388761,0.0010725018,0.000029704326,0.00006437677],"category_scores_gemma":[0.00043738078,0.000055890654,0.00020438104,0.0008827537,0.00010570852,0.0010730452,0.00033124053,0.00017623327,0.000010177845],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017606653,0.0011085906,0.005141062,0.00031482117,0.0006325783,0.00052524055,0.0026436707,0.0015080909,0.16795164,0.08605418,0.7267988,0.007303729],"study_design_scores_gemma":[0.0018006876,0.00010845954,0.006403073,0.0006405114,0.00035944488,0.0010273075,0.002267494,0.1157267,0.5734975,0.0045009814,0.29304987,0.0006179816],"about_ca_topic_score_codex":0.0000011810553,"about_ca_topic_score_gemma":0.0000063909265,"teacher_disagreement_score":0.73531353,"about_ca_system_score_codex":0.00002706541,"about_ca_system_score_gemma":0.00018024468,"threshold_uncertainty_score":0.51963943},"labels":[],"label_agreement":null},{"id":"W3137867769","doi":"10.22148/001c.21599","title":"Measuring Canonicity: Graduate Reading Lists in Departments of Hispanic Studies","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Gender Diversity and Inequality","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Diversity (politics); Dominance (genetics); Computer science; Reading (process); Graduate students; Measure (data warehouse); Psychology; Mathematics education; Natural language processing; Sociology; Linguistics; Pedagogy; Data mining; Biology; Anthropology","score_opus":0.4901887461836222,"score_gpt":0.39634176886769973,"score_spread":0.09384697731592245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3137867769","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99608713,0.00078124163,0.000012941841,0.0012329286,0.00025272823,0.00002441263,0.0000025170634,0.0000027859894,0.0016033115],"genre_scores_gemma":[0.99695003,0.0019730194,0.00023004669,0.000056759833,0.00009322422,8.288052e-8,5.5961715e-7,0.0000016992703,0.00069458596],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.9986193,0.00020899142,0.0003897171,0.00007303819,0.0005399595,0.0001689553],"domain_scores_gemma":[0.99854547,0.00006789636,0.00034713908,0.000055443907,0.0009084825,0.00007557758],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010107186,0.00006351066,0.0002938085,0.00006119257,0.00013597842,0.000029835017,0.00014727876,0.000038635404,0.000018713048],"category_scores_gemma":[0.00081323663,0.00005030596,0.00013368574,0.0003488824,0.000087871114,0.00038773456,0.000053290147,0.00016311885,0.0000016410373],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020058008,0.0009705503,0.4791074,0.0006004105,0.0030512,0.0014220164,0.4669864,0.0041725463,0.010578606,0.020505607,0.008462977,0.0039417217],"study_design_scores_gemma":[0.0028617857,0.00035048323,0.16966136,0.0010529032,0.0008938898,0.000088204375,0.77118737,0.00023870301,0.02107176,0.014830263,0.017020306,0.0007429878],"about_ca_topic_score_codex":0.00045220583,"about_ca_topic_score_gemma":0.0039556893,"teacher_disagreement_score":0.30944604,"about_ca_system_score_codex":0.00030443273,"about_ca_system_score_gemma":0.00023580025,"threshold_uncertainty_score":0.22073673},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Politics; Popularity; Centrality; Literature; History; Classics; Law; Art; Political science","score_opus":0.03109520831080825,"score_gpt":0.29509826779773196,"score_spread":0.2640030594869237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3151065226","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030765293,0.0013529379,0.9618062,0.005014315,0.00036155176,0.000031855285,0.0000016753288,0.000028052962,0.00063814555],"genre_scores_gemma":[0.98889655,0.00015837654,0.0091577545,0.00036542598,0.00022856756,7.5378495e-7,0.0000027013493,0.0000040872383,0.0011857874],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998877,0.00011204757,0.00042284117,0.00013401276,0.00027069464,0.00018342047],"domain_scores_gemma":[0.9979333,0.000116257856,0.00038563722,0.00015597706,0.001247704,0.00016109421],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00034109486,0.00009816188,0.00019005458,0.000074420146,0.0002461563,0.00023989088,0.00032456257,0.00005774838,0.000031889605],"category_scores_gemma":[0.00023362509,0.00007547496,0.00016781694,0.00057527353,0.000079415106,0.0013828258,0.000063157815,0.00042035343,0.0000063140865],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004355334,0.00071062427,0.03394311,0.00006563248,0.0017273257,0.0010799627,0.037972465,0.39233968,0.0039020781,0.45949242,0.031160334,0.03756283],"study_design_scores_gemma":[0.00043566,0.0000810282,0.01430773,0.00011951889,0.00013145161,0.00045614544,0.0008613285,0.9650978,0.0027928257,0.013271546,0.0021886374,0.00025633286],"about_ca_topic_score_codex":0.0000041666135,"about_ca_topic_score_gemma":0.000008099201,"teacher_disagreement_score":0.95813125,"about_ca_system_score_codex":0.00008223462,"about_ca_system_score_gemma":0.00012985614,"threshold_uncertainty_score":0.30777797},"labels":[],"label_agreement":null},{"id":"W3152040157","doi":"10.22148/001c.22086","title":"An Institutional Perspective on Genres: Generic Subtitles in German Literature from 1500-2020","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Perspective (graphical); German; Relation (database); Computer science; Poetry; Field (mathematics); Linguistics; Artificial intelligence; Philosophy","score_opus":0.018246796622688802,"score_gpt":0.32749123006626346,"score_spread":0.30924443344357466,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3152040157","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8837528,0.00316813,0.10911946,0.0027193285,0.00023434026,0.00005341212,0.000015797601,0.00005405135,0.0008826338],"genre_scores_gemma":[0.92929965,0.00053177966,0.06912657,0.00051971804,0.00038989825,6.704493e-7,0.0000144096375,0.000005540188,0.000111741305],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99861115,0.00009566064,0.00040920902,0.0002586843,0.0004628324,0.0001624724],"domain_scores_gemma":[0.9983572,0.0000464455,0.0002806462,0.00029429563,0.00090157334,0.00011989235],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00013215347,0.00015759747,0.00030527933,0.00017785162,0.000076916716,0.00026636827,0.00062006665,0.00008482002,0.000026934424],"category_scores_gemma":[0.00011474938,0.000111922,0.00020264361,0.0010138426,0.000049883565,0.0012429599,0.00007577583,0.0004321056,0.000007435689],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022400888,0.0024893072,0.008001218,0.00004481774,0.0012948588,0.017693363,0.033069644,0.06904971,0.14651306,0.6802204,0.011661452,0.029738132],"study_design_scores_gemma":[0.0047954023,0.0021392847,0.09484317,0.0020574017,0.00060566474,0.0033275231,0.010351163,0.34114784,0.1283394,0.36696473,0.042111255,0.0033171668],"about_ca_topic_score_codex":0.000016086933,"about_ca_topic_score_gemma":0.00008818674,"teacher_disagreement_score":0.3132557,"about_ca_system_score_codex":0.00035267492,"about_ca_system_score_gemma":0.00016105757,"threshold_uncertainty_score":0.4564047},"labels":[],"label_agreement":null},{"id":"W3153487356","doi":"10.22148/001c.22330","title":"The Asian American Literature We’ve Constructed","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Asian Culture and Media Studies","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scholarship; Asian americans; Gender studies; Conflation; History; Chinese americans; American literature; Hegemony; Ethnic group; Literary criticism; Asian studies; Rhetoric; Sociology; Political science; Literature; Anthropology; Law; China; Art; Epistemology; Linguistics","score_opus":0.01652242526843748,"score_gpt":0.3165288529940538,"score_spread":0.3000064277256163,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3153487356","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13615443,0.044910256,0.0001655803,0.5375311,0.0052888193,0.00027309352,0.000033337754,0.000071098126,0.2755723],"genre_scores_gemma":[0.9589086,0.021754595,0.0013856756,0.0006405711,0.0024412554,6.799148e-7,0.0000031112982,0.0000068053955,0.014858676],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99876225,0.00015907265,0.00029656023,0.000075600496,0.00047331664,0.00023320688],"domain_scores_gemma":[0.99798286,0.00011580935,0.00037377322,0.00008027558,0.0012883327,0.00015892508],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002609394,0.000094438496,0.00024145347,0.000023137554,0.00067956786,0.00025901006,0.00021086528,0.000050436443,0.000030242589],"category_scores_gemma":[0.0008835277,0.000046218967,0.00021120765,0.00068721105,0.000643806,0.00019378339,0.000027733626,0.00034139067,0.00000588255],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008513004,0.00010328669,0.00421299,0.000024017296,0.00130188,0.0010501402,0.36035004,0.000011821327,0.0008920264,0.053109348,0.41515175,0.16370758],"study_design_scores_gemma":[0.00013481599,0.000037674883,0.00078362366,0.00006259886,0.00009138452,0.00009521397,0.3189519,0.0000022412285,0.00010944861,0.0010819185,0.6785723,0.00007686642],"about_ca_topic_score_codex":0.00001004762,"about_ca_topic_score_gemma":0.0007244648,"teacher_disagreement_score":0.8227542,"about_ca_system_score_codex":0.0000684279,"about_ca_system_score_gemma":0.00025734142,"threshold_uncertainty_score":0.5226755},"labels":[],"label_agreement":null},{"id":"W3153509735","doi":"10.22148/001c.22220","title":"Content-Era Ethics","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Anecdote; The arts; Content (measure theory); Sociology; Media studies; Digital media; Digital content; Criticism; Perspective (graphical); Argument (complex analysis); Aesthetics; Literature; Law; Art; Political science; Visual arts","score_opus":0.24935973582751333,"score_gpt":0.4165319943597316,"score_spread":0.16717225853221826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3153509735","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.72091144,0.0008299728,0.0011583662,0.10016897,0.0016499041,0.00006812997,0.000005812617,0.000025733578,0.17518164],"genre_scores_gemma":[0.9785374,0.0013305134,0.0009707173,0.0030791769,0.00051620934,2.0759279e-8,0.0000016405824,0.0000025289048,0.015561814],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.99878,0.000115477385,0.0003318988,0.000030937455,0.00060337933,0.00013828182],"domain_scores_gemma":[0.997732,0.00010342,0.0002815557,0.000049648,0.0016476831,0.00018573625],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008319771,0.000049252354,0.00013840382,0.000035461853,0.00023319612,0.00018180956,0.00012504471,0.00008679549,0.0005451232],"category_scores_gemma":[0.0029411681,0.000032937525,0.00013363222,0.0002431838,0.000082232415,0.0006036731,0.00001294965,0.00040807226,0.000030451209],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006604521,0.00022123594,0.0017389882,0.000058993806,0.00042718882,0.00026972152,0.51540095,0.00065955217,0.0036169505,0.2470098,0.2106381,0.019892443],"study_design_scores_gemma":[0.0006524948,0.00006993457,0.003884774,0.00009582046,0.00009106174,0.00012914606,0.25579935,0.00017574294,0.0015375008,0.0015222824,0.73587084,0.00017102926],"about_ca_topic_score_codex":0.00003163667,"about_ca_topic_score_gemma":0.0001696388,"teacher_disagreement_score":0.52523273,"about_ca_system_score_codex":0.000073690346,"about_ca_system_score_gemma":0.00055490184,"threshold_uncertainty_score":0.5968717},"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.12326729342130129,"score_gpt":0.28908285481989715,"score_spread":0.16581556139859588,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3153523203","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.36410615,0.010369006,0.000021793354,0.016081132,0.0065224543,0.00010474074,0.00023828555,0.000045254248,0.60251117],"genre_scores_gemma":[0.94257504,0.000093661,0.00010985359,0.00055037375,0.0061255684,2.2402727e-7,0.00002964912,0.0000059703557,0.050509676],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99873996,0.00002177713,0.00040293662,0.00017059548,0.00050459435,0.00016010964],"domain_scores_gemma":[0.99760455,0.0000131873385,0.00022904684,0.00031445854,0.0017257236,0.00011301437],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00036592074,0.00011117326,0.00021056106,0.00006497731,0.00026881712,0.0016028523,0.0005398634,0.000023699939,0.0019585674],"category_scores_gemma":[0.00020547234,0.00006856412,0.00010742529,0.000109648085,0.000517439,0.0050586304,0.00014092363,0.00028403796,0.000032284617],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012101081,0.0002432858,0.00033991883,0.000056312714,0.00024902687,0.0007263847,0.026236603,0.000019297699,0.0004389888,0.21795407,0.75204635,0.0015687601],"study_design_scores_gemma":[0.00016732795,0.00007784643,0.000116061565,0.000036750807,0.00004971862,0.00022850525,0.019145869,0.000024601106,0.00018399178,0.0011839173,0.9786664,0.000118970594],"about_ca_topic_score_codex":0.0000035476085,"about_ca_topic_score_gemma":0.000054232958,"teacher_disagreement_score":0.57846886,"about_ca_system_score_codex":0.00004536046,"about_ca_system_score_gemma":0.00023533488,"threshold_uncertainty_score":0.9994336},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Amateur; Publishing; Criticism; Literary criticism; Reading (process); Point (geometry); Literature; Computer science; History; Sociology; Art; Linguistics; Philosophy","score_opus":0.03654861629295647,"score_gpt":0.3283822056129931,"score_spread":0.2918335893200366,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3153865529","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9874661,0.0004687541,0.00006165027,0.0042121615,0.0001983781,0.00006309587,0.0000037225411,0.0000047971025,0.007521317],"genre_scores_gemma":[0.9910288,0.00023042598,0.00022640456,0.00012104053,0.00020085645,2.9197176e-7,0.0000010325915,0.000003892486,0.008187277],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9986924,0.00011657008,0.00040035407,0.000070851296,0.0005691059,0.00015074018],"domain_scores_gemma":[0.9983179,0.00027525518,0.00028070086,0.00005811903,0.00093634287,0.00013169358],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004424581,0.000071116905,0.00020760324,0.000027065056,0.00015837325,0.00021638545,0.00014510757,0.000036355574,0.0000109578605],"category_scores_gemma":[0.0005549587,0.000041910687,0.00009058225,0.00022294601,0.00026816392,0.00021033839,0.000034885135,0.0001422533,9.760194e-7],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029702956,0.003645896,0.035961933,0.00014663949,0.0021748282,0.0008235689,0.70624787,0.0053513935,0.00079948717,0.08301898,0.039571878,0.12196051],"study_design_scores_gemma":[0.0016931844,0.0005959484,0.021700757,0.00010932063,0.00029608267,0.00011068234,0.80382115,0.00062266533,0.00010117425,0.004962782,0.16578309,0.00020318848],"about_ca_topic_score_codex":0.000023440069,"about_ca_topic_score_gemma":0.00040972533,"teacher_disagreement_score":0.12621121,"about_ca_system_score_codex":0.000030930594,"about_ca_system_score_gemma":0.00015481049,"threshold_uncertainty_score":0.20866099},"labels":[],"label_agreement":null},{"id":"W3156333834","doi":"10.22148/001c.22223","title":"A Network Analysis of Postwar American Poetry in the Age of Digital Audio Archives","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Music Technology and Sound Studies","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Poetry; Literature; American poetry; Reading (process); Formalism (music); Spoken word; Perspective (graphical); Art; Computer science; History; Visual arts; Linguistics; Philosophy","score_opus":0.019787903427388696,"score_gpt":0.26148755891428205,"score_spread":0.24169965548689334,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3156333834","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97184676,0.0009122914,0.023111947,0.003016228,0.00007139312,0.00002114846,0.000004446073,0.000005443315,0.0010103309],"genre_scores_gemma":[0.99582684,0.0001471563,0.0037262884,0.000179672,0.000039235692,1.5981833e-7,0.0000015783514,0.0000012029316,0.00007784415],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989853,0.00006864879,0.00047218564,0.00008877091,0.00025101515,0.00013409587],"domain_scores_gemma":[0.99889344,0.00022311218,0.0004881881,0.00021972683,0.00015526694,0.000020278674],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00018627604,0.000078548655,0.00047892454,0.00017062062,0.000042800668,0.00004224235,0.0006092766,0.000025568079,0.0000021348976],"category_scores_gemma":[0.00015492325,0.00004312697,0.00031314886,0.0022672978,0.00027204677,0.00016227606,0.00013220613,0.0002130065,1.8969783e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012887546,0.0009655418,0.758126,0.000072625946,0.011251608,0.0019064333,0.062867895,0.038210772,0.0015408406,0.08531431,0.006582694,0.033032387],"study_design_scores_gemma":[0.0005834766,0.0006098083,0.93630517,0.00012463944,0.0014801321,0.00026451322,0.01360251,0.01138887,0.00023811894,0.031512447,0.0035876362,0.00030270117],"about_ca_topic_score_codex":0.00000809925,"about_ca_topic_score_gemma":0.00006645318,"teacher_disagreement_score":0.17817913,"about_ca_system_score_codex":0.000010051332,"about_ca_system_score_gemma":0.00004057762,"threshold_uncertainty_score":0.17586668},"labels":[],"label_agreement":null},{"id":"W3156898303","doi":"10.22148/001c.22333","title":"Feminist Bestsellers: A Digital History of 1970s Feminism","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Cultural History and Identity Formation","field":"Arts and Humanities","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Feminism; Mainstream; Historiography; Gender studies; Sociology; Salient; Media studies; Political science; Law","score_opus":0.05633336489340642,"score_gpt":0.23175837072097272,"score_spread":0.1754250058275663,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3156898303","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5210824,0.0042016283,0.00002244604,0.00064855244,0.003753455,0.00006180039,0.00005755246,0.000020886324,0.47015125],"genre_scores_gemma":[0.9023601,0.00010609702,0.00006647881,0.0001266659,0.0005956307,1.832008e-7,0.00001828384,0.000005266033,0.096721336],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99889666,0.000017819319,0.0005979495,0.00006549302,0.0003164425,0.000105656414],"domain_scores_gemma":[0.99845666,0.00002445404,0.00059847086,0.00008708622,0.00075838837,0.00007497017],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.000096824755,0.00010567422,0.00027490174,0.00006826805,0.00008843742,0.0001248129,0.00013878309,0.000038595375,0.0010886056],"category_scores_gemma":[0.00007855311,0.00007456526,0.00033722425,0.000041818003,0.00023194066,0.0009072123,0.000026170444,0.0001718443,0.00003734914],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010738188,0.00052520493,0.00026204027,0.0002515986,0.00053003884,0.00017597864,0.252268,0.00007415302,0.002834588,0.09222942,0.6471736,0.0035680563],"study_design_scores_gemma":[0.00027622675,0.00012635339,0.0000305246,0.00008189413,0.000118407464,0.00015431081,0.025556706,0.00005233017,0.00027895666,0.00044145808,0.9727674,0.000115429815],"about_ca_topic_score_codex":0.000012293192,"about_ca_topic_score_gemma":0.0000729734,"teacher_disagreement_score":0.38127762,"about_ca_system_score_codex":0.00027400313,"about_ca_system_score_gemma":0.0001497342,"threshold_uncertainty_score":0.9998245},"labels":[],"label_agreement":null},{"id":"W3156955396","doi":"10.22148/001c.22332","title":"Squatter Regionalism: Postwar Fiction, Geography, and the Program Era","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Cultural History and Identity Formation","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Regionalism (politics); Population; State (computer science); History; Sociology; Demography; Political science; Law; Computer science","score_opus":0.025514725722821138,"score_gpt":0.23801064463954352,"score_spread":0.21249591891672237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3156955396","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94973606,0.00460782,0.000025850031,0.028754953,0.0015209293,0.00018017579,0.000011796774,0.00003548019,0.015126914],"genre_scores_gemma":[0.9814605,0.0008226983,0.00016071707,0.0014992558,0.0013464041,0.0000019688418,0.000021625881,0.0000053790427,0.0146814175],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99923384,0.0000538428,0.0003346141,0.00005606426,0.00023486694,0.000086742926],"domain_scores_gemma":[0.9987123,0.000027488744,0.00026226655,0.000064704895,0.00088813604,0.000045112138],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001779024,0.00008127295,0.00016280382,0.00004285829,0.00039651114,0.00036997176,0.00008064725,0.000024761204,0.00033588056],"category_scores_gemma":[0.000045066427,0.00004006212,0.0002305255,0.0000560629,0.00031422387,0.00073764246,0.000019439127,0.00020023811,0.000008945962],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027309844,0.00026465318,0.0007490277,0.000104616505,0.0007053253,0.00006513983,0.08166204,0.000087548324,0.00009144847,0.3213294,0.58750886,0.007158855],"study_design_scores_gemma":[0.00081557,0.000085329855,0.001479912,0.00004406533,0.00022650982,0.00036332014,0.009410994,0.00018175525,0.000025056876,0.0029623518,0.984303,0.00010212592],"about_ca_topic_score_codex":0.00005228032,"about_ca_topic_score_gemma":0.00016256805,"teacher_disagreement_score":0.39679417,"about_ca_system_score_codex":0.000022271692,"about_ca_system_score_gemma":0.00001764713,"threshold_uncertainty_score":0.36776567},"labels":[],"label_agreement":null},{"id":"W3160788868","doi":"10.22148/001c.24077","title":"Putting the Sorting Hat on J.K. Rowling’s Reader: A digital inquiry into the age of the implied readership of the Harry Potter series","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Themes in Literature Analysis","field":"Arts and Humanities","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Harry potter; Audience measurement; Readability; Style (visual arts); Information Age; Literature; Psychology; History; Media studies; Sociology; Art; Linguistics; Law; Philosophy; Political science","score_opus":0.05403514568411098,"score_gpt":0.26353588962540336,"score_spread":0.20950074394129237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3160788868","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9479764,0.00117839,0.000013244553,0.041730516,0.001160321,0.00012732451,0.000015622412,0.000010089561,0.0077881087],"genre_scores_gemma":[0.9910122,0.000035405526,0.000060203532,0.0014522318,0.0013606414,7.9943965e-7,0.0000036336678,0.000014479706,0.0060604187],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9979707,0.00018568459,0.0008894157,0.00012442168,0.00062079093,0.0002089696],"domain_scores_gemma":[0.9972496,0.00029052424,0.0012783986,0.00048600053,0.00066597457,0.000029500843],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005323268,0.00019277498,0.00037567393,0.000043948112,0.00058053137,0.0005423396,0.00079727184,0.000059321574,0.00007800223],"category_scores_gemma":[0.0005738268,0.00006436625,0.0007541516,0.00022794289,0.0008473603,0.00047263716,0.00020173147,0.0006647993,0.0000016238774],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001232006,0.00015628144,0.0039288737,0.00016657502,0.002180882,0.000041656866,0.9025131,0.0012278935,0.007487001,0.06352466,0.014963934,0.003685932],"study_design_scores_gemma":[0.000750023,0.00025424443,0.0051353755,0.0018054751,0.0023759904,0.00034498263,0.7951436,0.00019669169,0.020879138,0.033338763,0.13929614,0.00047958037],"about_ca_topic_score_codex":0.000016132488,"about_ca_topic_score_gemma":0.00021730755,"teacher_disagreement_score":0.12433221,"about_ca_system_score_codex":0.000047639885,"about_ca_system_score_gemma":0.00006935305,"threshold_uncertainty_score":0.52297926},"labels":[],"label_agreement":null},{"id":"W3166272135","doi":"10.22148/001c.24722","title":"Modeling Conflict: Representations of Social Groups in Present-Day Dutch Literature","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narratology; Narrative; Representation (politics); Sociology; Epistemology; Psychology; Literature; Politics; Political science; Art; Philosophy; Law","score_opus":0.07713974378771252,"score_gpt":0.41915106394212137,"score_spread":0.3420113201544088,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3166272135","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96161145,0.002159989,0.013774625,0.012987397,0.00039493124,0.00008234547,0.000012210589,0.000009712259,0.008967355],"genre_scores_gemma":[0.9936772,0.00031383903,0.0031861153,0.000058557285,0.00069584604,4.1592358e-7,0.000010511721,0.0000036960853,0.0020538485],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99805355,0.00045658599,0.00062678143,0.00010319537,0.0006133075,0.00014660685],"domain_scores_gemma":[0.99765813,0.00021858698,0.0003115409,0.000057444104,0.0016775256,0.00007677322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010012857,0.000077718854,0.00031810874,0.00014348475,0.00017035876,0.00010990132,0.0001739303,0.000082182,0.00011173475],"category_scores_gemma":[0.0006808053,0.000060250175,0.0003241404,0.0011518396,0.00008119843,0.00036836718,0.0000353948,0.00026013894,0.0000010711648],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000105753614,0.0006962386,0.013807872,0.00009751755,0.0010144825,0.00045634925,0.3410993,0.4520219,0.004058088,0.17098445,0.0047299406,0.010928114],"study_design_scores_gemma":[0.0033929825,0.00013818181,0.034504134,0.0006050675,0.001240175,0.0001414247,0.217307,0.5831858,0.0013515535,0.10700072,0.050173327,0.000959637],"about_ca_topic_score_codex":0.00014719779,"about_ca_topic_score_gemma":0.00051453867,"teacher_disagreement_score":0.1311639,"about_ca_system_score_codex":0.000064160835,"about_ca_system_score_gemma":0.00020278062,"threshold_uncertainty_score":0.24569307},"labels":[],"label_agreement":null},{"id":"W3172137656","doi":"10.22148/001c.24911","title":"Can We Map Culture?","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language and cultural evolution","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Analogy; Imperfect; Appeal; Epistemology; Space (punctuation); Embodied cognition; Representation (politics); Domain (mathematical analysis); Ground truth; Computer science; Sociology; Artificial intelligence; Mathematics; Psychology; Philosophy","score_opus":0.02600725044092982,"score_gpt":0.31644482770217985,"score_spread":0.29043757726125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3172137656","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.60017484,0.03723183,0.0004139617,0.2519759,0.0054499367,0.00028046154,0.00003551678,0.00013028333,0.104307264],"genre_scores_gemma":[0.9426155,0.0029132138,0.00095784914,0.00060414383,0.0018667906,3.0458904e-7,0.000006827068,0.000005025896,0.051030315],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99876237,0.00011842265,0.00030383875,0.00008193862,0.0005205183,0.00021293339],"domain_scores_gemma":[0.9984275,0.000020206075,0.00025197252,0.00007039038,0.0010518767,0.00017805793],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023870239,0.00008778679,0.00019573625,0.000025897081,0.00027760534,0.00013849592,0.00018635763,0.00009210371,0.0004440805],"category_scores_gemma":[0.00028385886,0.000052932737,0.00023516362,0.00032854828,0.00009387207,0.00035535905,0.00002243151,0.00022440414,0.0000284537],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007156629,0.00036684962,0.0026982934,0.00008902449,0.0007353576,0.0018037669,0.27524942,0.00045221706,0.02335372,0.08292771,0.58932835,0.022923738],"study_design_scores_gemma":[0.00039553258,0.00006750784,0.00038325504,0.00010120556,0.00018368625,0.00015148742,0.12474031,0.000030405008,0.0017083631,0.0072235144,0.8648054,0.00020933244],"about_ca_topic_score_codex":0.00013262352,"about_ca_topic_score_gemma":0.0016398146,"teacher_disagreement_score":0.34244066,"about_ca_system_score_codex":0.00016756845,"about_ca_system_score_gemma":0.00023192506,"threshold_uncertainty_score":0.486237},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"American literature; History; Identity (music); Politics; National identity; Principal (computer security); Political science; Literature; Law; Aesthetics; Art","score_opus":0.020497545934156022,"score_gpt":0.22723858050912868,"score_spread":0.20674103457497267,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3174730804","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.70572,0.12614149,0.0015561144,0.074287,0.0018946179,0.00016686565,0.00009276655,0.000022608561,0.090118535],"genre_scores_gemma":[0.97131443,0.006557294,0.00038877907,0.0018584931,0.00042124602,6.909615e-7,0.0000020662844,0.0000073253595,0.019449662],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99908894,0.000024537669,0.00059524644,0.000115626375,0.00003712703,0.00013854429],"domain_scores_gemma":[0.9989878,0.00008208118,0.0005942482,0.00008821632,0.00017217612,0.0000755039],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00028991338,0.00010196536,0.0005567074,0.000060531103,0.0001733472,0.00015234169,0.00012459485,0.000037497663,0.000045835055],"category_scores_gemma":[0.00024172706,0.000065952765,0.00023698532,0.00042868685,0.00021860577,0.00018781239,0.000048897702,0.00029557582,0.000018035],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027444083,0.0001569135,0.018966673,0.000042472002,0.0015519381,0.00018700091,0.021313855,0.00022827272,0.000017700884,0.85610366,0.094636045,0.0065210075],"study_design_scores_gemma":[0.0014134948,0.00008456139,0.005718843,0.000028207245,0.000072151954,0.0000922178,0.001853353,0.0009457595,0.0000067732,0.032455232,0.9570887,0.00024071643],"about_ca_topic_score_codex":0.000042647163,"about_ca_topic_score_gemma":0.000059344424,"teacher_disagreement_score":0.8624526,"about_ca_system_score_codex":0.00012403264,"about_ca_system_score_gemma":0.000018298957,"threshold_uncertainty_score":0.26894757},"labels":[],"label_agreement":null},{"id":"W3180413404","doi":"10.22148/001c.25525","title":"Chance Encounters: World Literature Between the Unexpected and the Probable","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Philippine History and Culture","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Probabilistic logic; History; Epistemology; History of literature; Sociology; Literature; Philosophy; Art; Archaeology","score_opus":0.024376555183349694,"score_gpt":0.28799913643385716,"score_spread":0.26362258125050747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3180413404","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.56378835,0.09154625,0.0001342331,0.26967126,0.002079856,0.0005269539,0.000027383918,0.000054681834,0.07217104],"genre_scores_gemma":[0.96934175,0.0017376039,0.00010309496,0.00095298426,0.0019125728,0.0000012786901,0.000002784808,0.000003838717,0.025944065],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.998863,0.00027134793,0.00024993997,0.00008121058,0.00038076052,0.0001537803],"domain_scores_gemma":[0.99877155,0.000116945266,0.00027254163,0.000096869764,0.0006665529,0.000075549586],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006315503,0.00008796532,0.00021191672,0.000023387487,0.00066001044,0.00020782807,0.0002775741,0.000054094275,0.00005071708],"category_scores_gemma":[0.00033015537,0.00003600785,0.00014834061,0.00072227255,0.00042011146,0.0003668724,0.000030176454,0.00043216412,0.000002108625],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028312556,0.00014109499,0.003923466,0.00008542987,0.0010958558,0.0002744037,0.6405214,0.00019890249,0.0016574162,0.112025924,0.23592687,0.0038660963],"study_design_scores_gemma":[0.0006869245,0.000025037907,0.0011880029,0.00019594865,0.00022007315,0.000064262975,0.023571685,0.000023422781,0.00014630513,0.0077571985,0.965998,0.00012316281],"about_ca_topic_score_codex":0.000012242079,"about_ca_topic_score_gemma":0.00075903675,"teacher_disagreement_score":0.7300711,"about_ca_system_score_codex":0.00006475869,"about_ca_system_score_gemma":0.00011296252,"threshold_uncertainty_score":0.5076333},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.02610744377729758,"score_gpt":0.29019006063620456,"score_spread":0.264082616858907,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3183176260","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99000263,0.0036486192,0.00068927574,0.003952825,0.00051629986,0.00007744514,0.000003696418,0.0000020781176,0.0011071216],"genre_scores_gemma":[0.99761474,0.00088163774,0.000054828593,0.000049613092,0.00047837754,5.227847e-7,3.3321564e-7,0.0000030685546,0.0009168881],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.9975995,0.000710977,0.00052061083,0.00007541871,0.00087118987,0.00022229897],"domain_scores_gemma":[0.99755543,0.00070678,0.00069808104,0.00016732607,0.00081763504,0.000054746633],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016501952,0.000083087005,0.00031507562,0.000030303223,0.0004418626,0.00007328807,0.00070373184,0.000070881084,0.00001774405],"category_scores_gemma":[0.00062290044,0.00003439725,0.00052600226,0.0021427604,0.00056985527,0.0001282346,0.00007830772,0.00040271878,2.7572497e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030525067,0.00013363842,0.11687027,0.000013386699,0.0016329072,0.00002272825,0.096135415,0.757977,0.00022526356,0.013529923,0.0055712047,0.007583017],"study_design_scores_gemma":[0.001073261,0.00020238831,0.7190793,0.000604103,0.0039281393,0.00002827354,0.15007383,0.07424117,0.00029944032,0.041714974,0.008234491,0.0005205991],"about_ca_topic_score_codex":0.0002520569,"about_ca_topic_score_gemma":0.010592123,"teacher_disagreement_score":0.68373585,"about_ca_system_score_codex":0.000049759765,"about_ca_system_score_gemma":0.00027353174,"threshold_uncertainty_score":0.5910653},"labels":[],"label_agreement":null},{"id":"W3200612959","doi":"10.22148/001c.28215","title":"Celebrating 5 Years of Cultural Analytics","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Analytics; Data science; Computer science","score_opus":0.06948626028111493,"score_gpt":0.39423671341715394,"score_spread":0.324750453136039,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3200612959","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98340046,0.0009826869,0.0037067789,0.0027060767,0.00047399767,0.000043138116,0.00000646552,0.00001320515,0.008667212],"genre_scores_gemma":[0.97655034,0.00038337716,0.019093078,0.00012470818,0.0007055458,1.491589e-7,0.000005815295,0.000005800896,0.0031312096],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.99771583,0.0003117179,0.0007596272,0.00011816243,0.00088229065,0.00021238314],"domain_scores_gemma":[0.9964017,0.00033398814,0.0007626584,0.000094139876,0.0022409668,0.00016652263],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010011771,0.000108020045,0.0004422134,0.00011150622,0.00018438262,0.00012932869,0.00026864823,0.000081419756,0.00026774575],"category_scores_gemma":[0.0015027388,0.00008324025,0.0005192202,0.0010820888,0.00018420232,0.00038929793,0.00004971051,0.00025310542,0.0000061414526],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002600282,0.0018063602,0.06985416,0.00029700625,0.007360601,0.0017367925,0.16875978,0.17729092,0.035979033,0.20532674,0.045910135,0.28541845],"study_design_scores_gemma":[0.004496092,0.0008154072,0.08833983,0.0009061298,0.005649888,0.0005712887,0.39629,0.052174985,0.013586502,0.06586014,0.36893904,0.002370685],"about_ca_topic_score_codex":0.00009918852,"about_ca_topic_score_gemma":0.00028302037,"teacher_disagreement_score":0.32302892,"about_ca_system_score_codex":0.00010101023,"about_ca_system_score_gemma":0.00035498835,"threshold_uncertainty_score":0.3394439},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.06872344227832917,"score_gpt":0.3059254956830484,"score_spread":0.23720205340471923,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3210937173","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47322097,0.0019694827,0.51607394,0.007300537,0.00036328737,0.00007595741,7.675535e-7,0.000014027648,0.0009810557],"genre_scores_gemma":[0.98435473,0.0002380139,0.014807896,0.00034619216,0.000110317014,2.842258e-7,6.7650893e-7,0.0000018906471,0.00014002128],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989161,0.00012428088,0.00039743946,0.00010388782,0.00030924362,0.00014907574],"domain_scores_gemma":[0.9991978,0.00010861316,0.0002250031,0.00012949963,0.00027062706,0.00006846355],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008486422,0.00008281991,0.0001949349,0.0000460301,0.00009719329,0.0002516318,0.00033785842,0.000033049273,0.000004783069],"category_scores_gemma":[0.00012147637,0.000048807182,0.00007251258,0.00026124576,0.000029848916,0.0002474713,0.00010101215,0.0003422268,0.0000029458022],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000085136,0.00058405206,0.13614891,0.00033517732,0.0005927049,0.0034256126,0.084983565,0.1216437,0.0076639755,0.61021763,0.0037918536,0.03052768],"study_design_scores_gemma":[0.0027806216,0.00041563195,0.07550018,0.0009517676,0.00028345562,0.0048469827,0.019602468,0.81961554,0.0152280275,0.025923785,0.033868626,0.0009829415],"about_ca_topic_score_codex":0.0000023498076,"about_ca_topic_score_gemma":0.000011267326,"teacher_disagreement_score":0.6979718,"about_ca_system_score_codex":0.000059750575,"about_ca_system_score_gemma":0.00006896296,"threshold_uncertainty_score":0.24264912},"labels":[],"label_agreement":null},{"id":"W3215541196","doi":"10.22148/001c.30009","title":"The Generative Dissensus of Reading the Feminist Novel, 1995-2020: A Computational Analysis of Interpretive Communities","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Feminism; Ambivalence; Sociology; Reading (process); Publishing; Criticism; TRACE (psycholinguistics); Generative grammar; Literary criticism; Epistemology; Gender studies; Psychology; Literature; Political science; Linguistics; Art; Social psychology; Law; Philosophy","score_opus":0.0553978990341914,"score_gpt":0.3932323339961122,"score_spread":0.3378344349619208,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3215541196","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.928013,0.0015490239,0.046670884,0.012718008,0.0003586429,0.000114833274,0.000111834495,0.00000828678,0.010455459],"genre_scores_gemma":[0.9938357,0.00019462963,0.0052580857,0.000084516956,0.00013448145,4.850635e-7,0.000012609376,0.0000030194085,0.00047647936],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9976717,0.0006732675,0.0007505783,0.0000665476,0.0007132686,0.00012458659],"domain_scores_gemma":[0.99268836,0.003731498,0.0010523744,0.00011232633,0.0023578552,0.000057568268],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015119618,0.00009667487,0.00044243972,0.00010641689,0.0006335842,0.00011861511,0.00038867543,0.00003835232,0.000047563226],"category_scores_gemma":[0.000949368,0.00004949042,0.00060540787,0.0017445564,0.0007555431,0.00010197013,0.00007627553,0.00022618484,3.1845818e-7],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019598752,0.0007972877,0.013267731,0.0000244345,0.019930227,0.000028220746,0.23602629,0.33739564,0.0011351731,0.38305897,0.0015499127,0.0065901256],"study_design_scores_gemma":[0.0005821233,0.00011942774,0.020359205,0.00016173968,0.005684271,0.00003895734,0.77052534,0.18356042,0.0004462572,0.011899111,0.006387809,0.00023532147],"about_ca_topic_score_codex":0.00031926634,"about_ca_topic_score_gemma":0.00185014,"teacher_disagreement_score":0.53449905,"about_ca_system_score_codex":0.00008592239,"about_ca_system_score_gemma":0.00027155993,"threshold_uncertainty_score":0.48730814},"labels":[],"label_agreement":null},{"id":"W4200099029","doi":"10.22148/001c.30704","title":"Annotation Guidelines for Narrative Levels","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Narrative; CLARITY; Rule of thumb; Focus (optics); Set (abstract data type); Computer science; Sentence; Narrative criticism; Linguistics; Narrative history; Natural language processing; Philosophy; Programming language","score_opus":0.21613679332068877,"score_gpt":0.39836663804300365,"score_spread":0.18222984472231488,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200099029","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04132923,0.0003045081,0.94222957,0.015556013,0.000441802,0.000030443987,0.000001600793,0.00000810692,0.00009871832],"genre_scores_gemma":[0.49927276,0.0000338674,0.4976286,0.0010233474,0.000601393,6.1241076e-7,0.0000013486693,0.0000034518873,0.0014346174],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99907297,0.000025521576,0.00046389757,0.000096506905,0.000238928,0.00010219225],"domain_scores_gemma":[0.9953994,0.000053949312,0.00029947763,0.00011267504,0.00407943,0.00005506011],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00026042285,0.00006768445,0.00016292608,0.00003853604,0.000070261034,0.00012634229,0.00026047844,0.0000314386,0.0000072867783],"category_scores_gemma":[0.00068530405,0.00004552114,0.00014072032,0.00018693383,0.000010004865,0.00058865827,0.000038409216,0.00008259466,0.0000012448525],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006423337,0.00027740235,0.00064001104,0.00016555875,0.0009714855,0.00039781735,0.057178218,0.28015116,0.07120283,0.17250551,0.1621566,0.25428918],"study_design_scores_gemma":[0.0012030528,0.00021656725,0.0006362162,0.0001491537,0.00008550191,0.0005835831,0.005636989,0.89611983,0.019214192,0.044750553,0.031091822,0.00031256262],"about_ca_topic_score_codex":9.606218e-7,"about_ca_topic_score_gemma":0.0000043088758,"teacher_disagreement_score":0.61596864,"about_ca_system_score_codex":0.000038714847,"about_ca_system_score_gemma":0.00013406608,"threshold_uncertainty_score":0.18562983},"labels":[],"label_agreement":null},{"id":"W4200162648","doi":"10.22148/001c.30697","title":"On Organizing a Shared Task for the Digital Humanities – Conclusions and Future Paths","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Task (project management); Digital humanities; Narratology; Computer science; World Wide Web; Linguistics; Artificial intelligence; Engineering","score_opus":0.020825326980333212,"score_gpt":0.2705781422478529,"score_spread":0.24975281526751972,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200162648","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1377607,0.08291922,0.6482713,0.12450525,0.0032385471,0.0007804226,0.00018871862,0.0005169196,0.0018188959],"genre_scores_gemma":[0.9206602,0.00036789427,0.07565276,0.0016687731,0.0008700935,0.0000013648853,0.0000051725874,0.0000106708485,0.00076308125],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993049,0.000015511047,0.00022154352,0.000103244885,0.00023522013,0.000119562734],"domain_scores_gemma":[0.9987439,0.00022250578,0.00021387808,0.0001332368,0.00064287713,0.00004359283],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00012091932,0.00009814782,0.00015972415,0.000038756552,0.00028123104,0.0008341357,0.00042868775,0.000044137567,0.0000066199873],"category_scores_gemma":[0.00028369215,0.00004925344,0.00009779422,0.00019691612,0.000041322968,0.0006204535,0.00018133536,0.00022357816,5.6327957e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000087062035,0.00023631183,0.00018329329,0.00017786097,0.00061929424,0.0007726042,0.025082879,0.00008618268,0.018893477,0.7137698,0.08097918,0.15911207],"study_design_scores_gemma":[0.0037668226,0.0024323247,0.001056428,0.0014691252,0.0007654807,0.008309279,0.03393495,0.042281333,0.042479284,0.39856854,0.46298176,0.001954692],"about_ca_topic_score_codex":6.0992613e-7,"about_ca_topic_score_gemma":0.0000024574747,"teacher_disagreement_score":0.7828995,"about_ca_system_score_codex":0.000036704187,"about_ca_system_score_gemma":0.000069775066,"threshold_uncertainty_score":0.804359},"labels":[],"label_agreement":null},{"id":"W4200267238","doi":"10.22148/001c.30700","title":"On the Theory of Narrative Levels and Their Annotation in the Digital Context","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Narrative Theory and Analysis","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Context (archaeology); Categorization; Identification (biology); Computer science; Task (project management); Narrative criticism; Narrative network; Narrative inquiry; Narrative structure; Phenomenon; Epistemology; Linguistics; Artificial intelligence; History; Philosophy; Engineering; Archaeology","score_opus":0.06465114698293625,"score_gpt":0.25903134738114403,"score_spread":0.19438020039820778,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200267238","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98846126,0.00031703166,0.000043474516,0.003440767,0.000046719786,0.000032093947,0.000022917484,0.000001072154,0.007634653],"genre_scores_gemma":[0.9977485,0.000018806413,0.000002105836,0.0004963521,0.00012617446,3.6556273e-7,0.0000021248054,0.0000023731905,0.0016032088],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.999191,0.00026245968,0.00029030285,0.000049991344,0.00014194999,0.00006430764],"domain_scores_gemma":[0.9985638,0.00069277245,0.00027665007,0.000066262655,0.0003846318,0.000015828007],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00054103934,0.00008276269,0.0001948051,0.000040993382,0.00015516665,0.00016682301,0.00011451876,0.000015312846,0.000314234],"category_scores_gemma":[0.0002826375,0.000029700434,0.00013524528,0.00008431073,0.00027560454,0.00025008028,0.000011504122,0.00018095227,0.0000013219666],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000050638595,0.0000531532,0.000041528885,0.0000039588745,0.00015111487,0.000008532116,0.29223734,0.000045750905,0.00021287118,0.70503604,0.00048495654,0.0016741037],"study_design_scores_gemma":[0.00022906112,0.00014956213,0.00032060314,0.00006851704,0.00005011521,0.000021197508,0.83784264,0.00009351818,0.0008403801,0.1575369,0.0027875602,0.00005995142],"about_ca_topic_score_codex":0.000001573906,"about_ca_topic_score_gemma":0.00007102859,"teacher_disagreement_score":0.5474992,"about_ca_system_score_codex":0.000010588068,"about_ca_system_score_gemma":0.000023265262,"threshold_uncertainty_score":0.34406418},"labels":[],"label_agreement":null},{"id":"W4200327391","doi":"10.22148/001c.30698","title":"Narrative Boundaries Annotation Guide","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Scripting language; Annotation; Computer science; Schema (genetic algorithms); Narrative criticism; Narrative structure; World Wide Web; Information retrieval; Narrative inquiry; Artificial intelligence; Linguistics; Programming language; Philosophy","score_opus":0.03377907282816144,"score_gpt":0.3051666231684335,"score_spread":0.27138755034027207,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200327391","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2083913,0.00061066705,0.776616,0.011588328,0.00090094714,0.00002230585,5.276104e-7,0.000017716626,0.0018522141],"genre_scores_gemma":[0.82511246,0.00005142882,0.17110834,0.0005641423,0.00032022272,1.732265e-7,0.0000010574203,0.0000028988406,0.0028392961],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990752,0.000044441906,0.00037307318,0.000090443515,0.00031357337,0.00010330722],"domain_scores_gemma":[0.9983599,0.000029145802,0.00026167076,0.0001320189,0.0011562129,0.000061004106],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00018937261,0.000066186265,0.00014868854,0.000039348306,0.00014250481,0.0004817196,0.00027673683,0.00002914823,0.000016837785],"category_scores_gemma":[0.00023593262,0.000045872217,0.00010550236,0.00024192972,0.00003775898,0.0008257834,0.000060797596,0.00014070832,0.0000040913596],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004492132,0.00031419384,0.0022162895,0.000093907576,0.00096620043,0.0022439489,0.3058387,0.06910724,0.039214905,0.36949995,0.115925536,0.09453421],"study_design_scores_gemma":[0.0017368464,0.0004033288,0.0027811236,0.00027940795,0.00017064727,0.003210028,0.03628395,0.40658227,0.038125515,0.04647903,0.463194,0.0007538387],"about_ca_topic_score_codex":0.00000485633,"about_ca_topic_score_gemma":0.00001565499,"teacher_disagreement_score":0.61672115,"about_ca_system_score_codex":0.000062786436,"about_ca_system_score_gemma":0.0002805773,"threshold_uncertainty_score":0.46452332},"labels":[],"label_agreement":null},{"id":"W4200340511","doi":"10.22148/001c.30702","title":"Collaborative Annotation as a Teaching Tool","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Educational Tools and Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Annotation; Reading (process); Context (archaeology); Ambiguity; Interpretation (philosophy); Task (project management); Computer science; Close reading; Pedagogy; Mathematics education; Library science; Psychology; Linguistics; Literature; Art; Artificial intelligence; Engineering; History; Archaeology; Philosophy","score_opus":0.048586444208874334,"score_gpt":0.45047872418365054,"score_spread":0.4018922799747762,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200340511","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8423222,0.0010847654,0.0012042703,0.025987284,0.0011854077,0.00007341228,0.0000048588026,0.000010737252,0.1281271],"genre_scores_gemma":[0.9041894,0.0009476273,0.054832924,0.0010012354,0.0020683422,8.6036766e-7,0.000005231318,0.000005774261,0.036948573],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988345,0.00034673716,0.00025928894,0.00005765329,0.00039589067,0.00010595213],"domain_scores_gemma":[0.9977526,0.00020243389,0.00026835687,0.000039601215,0.0016527419,0.00008426042],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010234579,0.000050305687,0.00013276057,0.00002869984,0.0002946828,0.00014911615,0.00009014154,0.000042839456,0.00015033843],"category_scores_gemma":[0.003277119,0.000036816007,0.00009232258,0.00031718682,0.000053704272,0.00043638557,0.000009363733,0.00019723213,0.000008475996],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006295798,0.00047366906,0.0031095769,0.000020075666,0.00036301874,0.0001581604,0.13766177,0.0008790102,0.008936744,0.71362734,0.060117032,0.07459066],"study_design_scores_gemma":[0.0002597949,0.00006112317,0.0036402764,0.00005142244,0.00008297021,0.000043277447,0.09437555,0.000046635192,0.0007148877,0.014379554,0.8862318,0.00011269998],"about_ca_topic_score_codex":0.00003755429,"about_ca_topic_score_gemma":0.00007252892,"teacher_disagreement_score":0.8261148,"about_ca_system_score_codex":0.00011152317,"about_ca_system_score_gemma":0.000925199,"threshold_uncertainty_score":0.39232543},"labels":[],"label_agreement":null},{"id":"W4200492986","doi":"10.22148/001c.30701","title":"Annotation Guidelines for Narrative Levels and Narrative Acts v2","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Annotation; Narrative inquiry; Political science; Literature; Computer science; Art; Artificial intelligence","score_opus":0.10205961958090254,"score_gpt":0.3980552257095357,"score_spread":0.2959956061286332,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200492986","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07602962,0.005819841,0.8837485,0.033749178,0.00037593488,0.00013473231,0.0000072188313,0.000063616826,0.00007139582],"genre_scores_gemma":[0.425045,0.000051265546,0.5735566,0.0006838733,0.00019186382,0.0000010211915,0.0000019814124,0.00000403657,0.00046434073],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9989838,0.000046111592,0.00043353235,0.00014447692,0.00026269417,0.00012936545],"domain_scores_gemma":[0.9943601,0.00009478785,0.00046531216,0.000099193225,0.0049043186,0.000076317156],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003117243,0.00011768086,0.00023920623,0.00006893931,0.0001573299,0.0002623754,0.00027348954,0.000060766513,0.000004219421],"category_scores_gemma":[0.0014110334,0.00007514591,0.000096315336,0.00030787775,0.000038882474,0.0011715221,0.000072812494,0.00016157153,2.6422288e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001983797,0.00028997366,0.00029237953,0.00038086495,0.0010250896,0.0006914764,0.2765535,0.0011352404,0.29966655,0.11787827,0.13760689,0.16428135],"study_design_scores_gemma":[0.0025716021,0.0012795035,0.00068767974,0.0010260284,0.0002489994,0.002779136,0.039468378,0.07010942,0.47327176,0.38907412,0.018361038,0.001122339],"about_ca_topic_score_codex":0.0000012103886,"about_ca_topic_score_gemma":0.0000077098575,"teacher_disagreement_score":0.3490154,"about_ca_system_score_codex":0.00004847004,"about_ca_system_score_gemma":0.00014734047,"threshold_uncertainty_score":0.30643612},"labels":[],"label_agreement":null},{"id":"W4200572262","doi":"10.22148/001c.30703","title":"Annotation Guideline No. 7 (revised): Guidelines for annotation of narrative structure","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Annotation; Narrative; Focalization; Framing (construction); Computer science; Linguistics; Key (lock); Perspective (graphical); Literature; Artificial intelligence; History; Philosophy; Art","score_opus":0.04336450762914995,"score_gpt":0.37492497431710475,"score_spread":0.3315604666879548,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200572262","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018425245,0.005314887,0.96401316,0.01147299,0.00053113187,0.00014143153,0.000025774118,0.000044102937,0.00003130709],"genre_scores_gemma":[0.12966618,0.00009311544,0.8689475,0.0006398778,0.0003965568,6.6474854e-7,0.000025139154,0.0000061183405,0.00022483326],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9982097,0.00005136631,0.0010470478,0.00014946926,0.00041468555,0.00012772149],"domain_scores_gemma":[0.98117304,0.000070636066,0.0011574446,0.00016586717,0.017376352,0.00005668471],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00035992748,0.00013691442,0.00033356802,0.000104515595,0.00007052466,0.00010413796,0.00043876647,0.000081372404,0.000009302916],"category_scores_gemma":[0.0024183127,0.00008968391,0.00018788506,0.0004967409,0.0000317566,0.0009320452,0.00005835361,0.00016202693,4.6633565e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018819109,0.00016732224,0.00017838435,0.00048324192,0.00047383553,0.00011077116,0.009209484,0.007894415,0.5695393,0.028378602,0.27594656,0.1074299],"study_design_scores_gemma":[0.002137732,0.0009979998,0.0001298891,0.000881713,0.00030457124,0.0006636739,0.0035309815,0.27983394,0.5526498,0.13200442,0.02613296,0.00073230866],"about_ca_topic_score_codex":0.0000026357059,"about_ca_topic_score_gemma":0.0000049688038,"teacher_disagreement_score":0.27193952,"about_ca_system_score_codex":0.00005415878,"about_ca_system_score_gemma":0.00019777204,"threshold_uncertainty_score":0.36572036},"labels":[],"label_agreement":null},{"id":"W4200594119","doi":"10.22148/001c.30699","title":"Annotation Guidelines For narrative levels, time features, and subjective narration styles in fiction (SANTA 2)","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Narrative Theory and Analysis","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Computer science; Markup language; Literature; XML; History; Linguistics; World Wide Web; Art; Philosophy","score_opus":0.08781428286646534,"score_gpt":0.3294392872258638,"score_spread":0.24162500435939846,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200594119","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9907203,0.0016897792,0.0006730994,0.00431618,0.00023612595,0.000134642,0.00006573508,0.0000105724375,0.0021536169],"genre_scores_gemma":[0.98935795,0.00014870228,0.00052917394,0.00019892555,0.000778311,0.0000018171838,0.00004409574,0.000006720153,0.008934335],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99903435,0.00011277338,0.00047354374,0.00011328714,0.00016475511,0.00010131295],"domain_scores_gemma":[0.9968833,0.00009830598,0.00037213028,0.00004552617,0.0025621806,0.000038590646],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00037254984,0.00011995144,0.00028502353,0.00012585481,0.0002437192,0.00017127363,0.000050485785,0.000043117612,0.00018963433],"category_scores_gemma":[0.00044898808,0.000081622624,0.00013452803,0.00011080427,0.00008198792,0.0006587338,0.000009979484,0.0001381573,0.0000012226017],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012323384,0.00046823724,0.00065147836,0.00016100163,0.0017318022,0.00010369311,0.6509075,0.006255348,0.06373414,0.13745062,0.12764359,0.009660251],"study_design_scores_gemma":[0.0034345181,0.0010023941,0.004573291,0.0005270977,0.0007790413,0.0002686889,0.8788946,0.010240188,0.012443773,0.04030841,0.046772547,0.00075543404],"about_ca_topic_score_codex":0.000013597647,"about_ca_topic_score_gemma":0.0006746847,"teacher_disagreement_score":0.22798711,"about_ca_system_score_codex":0.0000577215,"about_ca_system_score_gemma":0.0000568645,"threshold_uncertainty_score":0.3328474},"labels":[],"label_agreement":null},{"id":"W4205760491","doi":"10.22148/001c.32036","title":"The Humanities in Public: A Computational Analysis of US National and Campus Newspapers","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Digital humanities; Newspaper; Humanities; Sociology; Meaning (existential); Publics; Media studies; Political science; Epistemology; Politics; Art; Law; Philosophy","score_opus":0.076590508968831,"score_gpt":0.3698167006261597,"score_spread":0.29322619165732866,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4205760491","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98353666,0.0008725983,0.0006189196,0.0076431786,0.00017181183,0.00006559468,0.000020194711,0.000005113182,0.0070659197],"genre_scores_gemma":[0.9976081,0.00008623871,0.001001898,0.00020339005,0.00008225305,0.0000012286459,0.000008363603,0.000002051916,0.0010064213],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.997716,0.00044139434,0.00051151565,0.00007427107,0.0011320573,0.00012473062],"domain_scores_gemma":[0.99759907,0.0008970327,0.00054254266,0.000031755964,0.00088473456,0.000044846445],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0022228463,0.00006362228,0.00025337536,0.0004445245,0.000877816,0.0001514486,0.00023192239,0.000019378573,0.00019206284],"category_scores_gemma":[0.00048699154,0.000042529784,0.00023855879,0.0017130657,0.00032188353,0.00020144689,0.000058180987,0.00018622132,2.4691684e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000050531944,0.00013437953,0.05318661,0.0000045034085,0.0022850032,0.000010901159,0.016882004,0.6549707,0.0000180617,0.26423743,0.0017469866,0.0064728847],"study_design_scores_gemma":[0.00091904664,0.00018142146,0.3594899,0.000011280236,0.0014990353,0.00003287185,0.08872892,0.16725525,0.0000019834538,0.06496342,0.31661162,0.00030526318],"about_ca_topic_score_codex":0.00039752087,"about_ca_topic_score_gemma":0.0033776592,"teacher_disagreement_score":0.48771542,"about_ca_system_score_codex":0.00019413605,"about_ca_system_score_gemma":0.00037171543,"threshold_uncertainty_score":0.675154},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.1678217392894345,"score_gpt":0.31922858610839955,"score_spread":0.15140684681896505,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4211124067","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9487361,0.00512875,0.00003222679,0.004614071,0.00176127,0.00025676875,0.0010527101,0.00004721202,0.03837095],"genre_scores_gemma":[0.9907667,0.00007886449,0.00024014064,0.00053554296,0.00077338144,8.1621056e-7,0.00011051785,0.000010312866,0.007483694],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988867,0.00008050478,0.00038915974,0.00011988147,0.00038341922,0.00014030698],"domain_scores_gemma":[0.9992137,0.000057070225,0.000229462,0.00016643261,0.0002710333,0.00006230567],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00030373843,0.00010653976,0.00021842407,0.000105506704,0.0004732278,0.00029320197,0.00035977023,0.00001272027,0.0011987154],"category_scores_gemma":[0.000047532936,0.00007436768,0.00006819075,0.000061615705,0.00011119483,0.00074777514,0.00013765525,0.000267753,0.0000035787348],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002511633,0.00039366304,0.0014550939,0.00014816895,0.00062003604,0.00040469939,0.34106418,0.00040020092,0.000086634594,0.13102047,0.51630425,0.0078514395],"study_design_scores_gemma":[0.00040475183,0.00020862844,0.0004978753,0.000022190898,0.00012350082,0.0002902256,0.06613311,0.0032125232,0.0000020311695,0.00066104345,0.92831,0.00013412148],"about_ca_topic_score_codex":0.00005749779,"about_ca_topic_score_gemma":0.00010334985,"teacher_disagreement_score":0.41200575,"about_ca_system_score_codex":0.000024404897,"about_ca_system_score_gemma":0.00006721276,"threshold_uncertainty_score":0.9997143},"labels":[],"label_agreement":null},{"id":"W4214538649","doi":"10.22148/001c.33086","title":"Cultures of E/valuation on the Social Web. A very short introduction to the special issue","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Valuation (finance); Discipline; Sociology; Epistemology; Social science; Business; Accounting; Philosophy","score_opus":0.07782660500584064,"score_gpt":0.29273903958087555,"score_spread":0.2149124345750349,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4214538649","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8563543,0.000078847625,0.0000021118792,0.0706842,0.003196303,0.00020661755,0.00006699973,0.000008718049,0.0694019],"genre_scores_gemma":[0.94554436,0.000009545018,0.0000032309047,0.0011341367,0.03821203,0.0000029798366,0.0000074970553,0.000006223356,0.01508002],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986426,0.000109737484,0.00036863837,0.000071236725,0.0006949206,0.00011283343],"domain_scores_gemma":[0.9991835,0.000049216884,0.00023008283,0.000085719985,0.00042457064,0.000026952805],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00064507546,0.00009506329,0.0001676818,0.00006197277,0.0009539681,0.00033706042,0.0003145376,0.000014776608,0.00332384],"category_scores_gemma":[0.00010340887,0.000043907003,0.00021379763,0.00007967546,0.000113974434,0.00028629915,0.000060927956,0.00035939284,0.000013887656],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013368578,0.00007316385,0.000009347714,0.000004143985,0.00009835808,0.0000024404171,0.047257222,0.0010536,0.000057790636,0.16428903,0.7850101,0.0020111077],"study_design_scores_gemma":[0.000086885004,0.00036313248,0.00014728992,0.0000063459806,0.000078861136,0.000012498265,0.07582159,0.0000494501,0.000054304965,0.001896249,0.92141664,0.00006675135],"about_ca_topic_score_codex":0.000007703474,"about_ca_topic_score_gemma":0.00014125813,"teacher_disagreement_score":0.16239278,"about_ca_system_score_codex":0.000095797,"about_ca_system_score_gemma":0.000035134704,"threshold_uncertainty_score":0.99758726},"labels":[],"label_agreement":null},{"id":"W4220668970","doi":"10.22148/001c.33570","title":"Between consumers and fans: Writing fan reports as a multifunctional evaluation practice","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language, Discourse, Communication Strategies","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Ticket; Feeling; Agency (philosophy); Value (mathematics); Resource (disambiguation); Linguistics; Advertising; Sociology; Psychology; Social psychology; Computer science; Business; Social science; Computer security","score_opus":0.10218838274259834,"score_gpt":0.36712676445686454,"score_spread":0.2649383817142662,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220668970","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9434158,0.0025282456,0.000040296993,0.006378234,0.0004814567,0.00018932644,0.000027625754,0.0000230996,0.046915866],"genre_scores_gemma":[0.9978738,0.0000653609,0.00036567848,0.00027877552,0.00039140877,0.000004819415,0.000035971738,0.000007773558,0.00097645127],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99814224,0.00026556544,0.0006003799,0.00009836671,0.0007814797,0.00011196998],"domain_scores_gemma":[0.9973731,0.00030163035,0.0010232703,0.00013305248,0.0011046723,0.00006428439],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0015120605,0.0001002248,0.00019232437,0.00009619601,0.00065617455,0.00031895633,0.00012440821,0.000019167637,0.0013411018],"category_scores_gemma":[0.000606958,0.00007465684,0.000094618976,0.00006036587,0.00015878389,0.0008896705,0.00007719619,0.00038513617,0.000003457293],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006107961,0.0011828141,0.03289317,0.00019458904,0.006520277,0.0011054948,0.5489893,0.009186986,0.0015093179,0.26897648,0.059092302,0.069738485],"study_design_scores_gemma":[0.00082139316,0.00029508033,0.0026766763,0.00004947264,0.0012295758,0.0018427966,0.70025826,0.00079732435,0.000033706958,0.0048886454,0.2868365,0.00027060456],"about_ca_topic_score_codex":0.00013927839,"about_ca_topic_score_gemma":0.00005832172,"teacher_disagreement_score":0.26408786,"about_ca_system_score_codex":0.00011607132,"about_ca_system_score_gemma":0.00016721044,"threshold_uncertainty_score":0.9995718},"labels":[],"label_agreement":null},{"id":"W4220823893","doi":"10.22148/001c.33994","title":"Post-War British Women Writers and their Cultural Impact: A Quantitative Approach","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Gender Politics and Representation","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Representation (politics); Period (music); History; Selection (genetic algorithm); Gender studies; First world war; British literature; Sociology; Literature; Political science; Art; Aesthetics; English literature; Law; Politics; Ancient history; Computer science","score_opus":0.048124870450156135,"score_gpt":0.342681067162671,"score_spread":0.2945561967125149,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220823893","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.995014,0.0006630626,0.000055376393,0.0012812499,0.00015196482,0.00007840465,0.000044426586,0.000006934119,0.002704586],"genre_scores_gemma":[0.99778503,0.00034492675,0.0003557488,0.00017712731,0.00017444466,0.000002266087,0.000008521807,0.000005291589,0.0011466228],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99870694,0.0002077283,0.0002839047,0.000100370526,0.00042200994,0.000279026],"domain_scores_gemma":[0.99893165,0.000054369604,0.0002873521,0.00004473626,0.0004706068,0.00021127162],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00058308075,0.00009051392,0.0002135625,0.000067606605,0.00065586844,0.00021047259,0.00018517963,0.00003299503,0.00011565282],"category_scores_gemma":[0.00013051525,0.00006209995,0.00016300344,0.00024159765,0.00019077705,0.00040457156,0.00006219591,0.00026372983,7.36667e-7],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000109801105,0.00028104472,0.0052373814,0.00002938497,0.0006471661,0.00007022126,0.97198504,0.0037671023,0.00206686,0.005080036,0.00818245,0.002543514],"study_design_scores_gemma":[0.00041346444,0.00035473655,0.006279246,0.000007666202,0.000034213248,0.0002944433,0.98411304,0.0011179477,0.00001367738,0.0020913184,0.005131936,0.00014834147],"about_ca_topic_score_codex":0.0012987034,"about_ca_topic_score_gemma":0.00007550316,"teacher_disagreement_score":0.012127971,"about_ca_system_score_codex":0.0003001181,"about_ca_system_score_gemma":0.00013660738,"threshold_uncertainty_score":0.50444764},"labels":[],"label_agreement":null},{"id":"W4224248308","doi":"10.22148/001c.34715","title":"Rhythms of Silence: Digital audio analysis of Swedish Radio Broadcasting, 1980-1989","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Radio, Podcasts, and Digital Media","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Silence; Broadcasting (networking); Public broadcasting; Modernization theory; Monopoly; Competition (biology); Telecommunications; Sociology; Political science; Computer science; Economics; Law; Computer security; Market economy; Aesthetics; Art","score_opus":0.03068121833925207,"score_gpt":0.30300412542167454,"score_spread":0.27232290708242246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224248308","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98449016,0.0008840691,0.000078625606,0.00041947947,0.00059676345,0.00007477632,0.0001238562,0.000011095675,0.013321167],"genre_scores_gemma":[0.9967427,0.0002052687,0.0001400227,0.00004451025,0.00047282688,9.196298e-7,0.000013496476,0.000008322411,0.0023719005],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.9967139,0.000121951394,0.0010653216,0.0001578863,0.0016165235,0.00032444624],"domain_scores_gemma":[0.99705493,0.00025751593,0.0015361203,0.00016219949,0.00072135334,0.00026785117],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00080486137,0.00015921387,0.0007926001,0.00043572174,0.0002651541,0.00012713221,0.0005478624,0.000056450575,0.00029651655],"category_scores_gemma":[0.0011251287,0.00012414533,0.0008338624,0.0023299868,0.00047446604,0.00083364494,0.00009027788,0.0003259457,0.0000019513598],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010026783,0.0028430722,0.38025627,0.0003083731,0.01703457,0.0010539521,0.25357223,0.1706766,0.0024433157,0.01647687,0.057576735,0.096755326],"study_design_scores_gemma":[0.0057768975,0.0052690394,0.13243747,0.0005466545,0.016857725,0.0009439052,0.536372,0.024475642,0.0016841751,0.005086439,0.26768503,0.0028650013],"about_ca_topic_score_codex":0.00016896431,"about_ca_topic_score_gemma":0.00007003652,"teacher_disagreement_score":0.28279978,"about_ca_system_score_codex":0.00021352537,"about_ca_system_score_gemma":0.0003266192,"threshold_uncertainty_score":0.50624996},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Contextualization; German; Context (archaeology); Historicism; Meaning (existential); Epistemology; Space (punctuation); Linguistics; Sociology; Philosophy; History; Interpretation (philosophy); Archaeology","score_opus":0.07670709901819155,"score_gpt":0.4049405778109269,"score_spread":0.32823347879273534,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4242882238","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97106004,0.0007545579,0.005141563,0.015562733,0.0003261647,0.000048113918,9.287316e-7,0.00000450079,0.007101404],"genre_scores_gemma":[0.99545586,0.00045086327,0.0018453054,0.00012279222,0.00067099795,1.788735e-7,1.835942e-7,0.0000015151178,0.0014522753],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993219,0.00010056643,0.00022156913,0.000059341997,0.00020802692,0.00008859498],"domain_scores_gemma":[0.998917,0.00016511373,0.0003196622,0.0000722201,0.0004535043,0.00007252951],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012550966,0.000053456097,0.0001940162,0.00002777652,0.00079516566,0.00028161492,0.00021242042,0.00003885748,0.000010600845],"category_scores_gemma":[0.0009835948,0.000027518412,0.000095530166,0.000037497175,0.0005040824,0.0002530473,0.00005421701,0.00013515905,3.1104955e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018423074,0.00010593716,0.020448646,0.00003519382,0.0009387586,0.000034786815,0.09773772,0.0040289345,0.000100184865,0.8134148,0.0014672863,0.06150351],"study_design_scores_gemma":[0.0070360126,0.00017124835,0.055229936,0.000143134,0.0016948893,0.00020707193,0.02800755,0.4657553,0.000018868705,0.31754866,0.12356044,0.00062688056],"about_ca_topic_score_codex":0.00029254658,"about_ca_topic_score_gemma":0.0002879444,"teacher_disagreement_score":0.49586615,"about_ca_system_score_codex":0.000015606827,"about_ca_system_score_gemma":0.000035956637,"threshold_uncertainty_score":0.6115852},"labels":[],"label_agreement":null},{"id":"W4244833923","doi":"10.22148/001c.22222","title":"Voice","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language, Metaphor, and Cognition","field":"Psychology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Style (visual arts); Vernacular; Passive voice; Linguistics; Reading (process); Criticism; Computer science; Art; Literature; Philosophy","score_opus":0.03719127276204112,"score_gpt":0.3307161208707724,"score_spread":0.29352484810873125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4244833923","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91888404,0.003712745,0.0010092737,0.0018688128,0.0017728985,0.000029816472,0.0000068880336,0.000013649462,0.07270185],"genre_scores_gemma":[0.9780489,0.00011930489,0.0005585279,0.00086414244,0.000915299,3.1729803e-7,0.0000063156176,0.000006578453,0.01948061],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9991911,0.00007733739,0.00031428356,0.00007629695,0.00020952919,0.00013145835],"domain_scores_gemma":[0.99892837,0.000041396062,0.00022877913,0.000099870485,0.00061870465,0.00008289311],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00015817024,0.000075029544,0.0001854167,0.000042169322,0.00003977037,0.000043028507,0.000100583595,0.000061092804,0.0017211937],"category_scores_gemma":[0.00011360654,0.000048204565,0.00020546927,0.00021135363,0.000024814733,0.00012688615,0.000013447077,0.00021641198,0.00021645892],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007957685,0.0025472012,0.011522978,0.00014729683,0.0062004738,0.01705021,0.05106996,0.00025538722,0.06278152,0.059143353,0.66015446,0.12833136],"study_design_scores_gemma":[0.0071006203,0.0010676443,0.07725433,0.000299504,0.0032626183,0.01761165,0.13521469,0.00023794283,0.023735287,0.021196788,0.7118347,0.0011842203],"about_ca_topic_score_codex":0.000006957295,"about_ca_topic_score_gemma":0.000013198649,"teacher_disagreement_score":0.12714715,"about_ca_system_score_codex":0.00002339815,"about_ca_system_score_gemma":0.00003502978,"threshold_uncertainty_score":0.99919134},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Literature; Philosophy; Art","score_opus":0.058045558012638984,"score_gpt":0.2682156559321219,"score_spread":0.21017009791948288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4247740782","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.912129,0.0002673526,0.001093253,0.011995167,0.00085846503,0.00003391397,0.000033616667,0.000021041562,0.07356818],"genre_scores_gemma":[0.9659431,0.00004384422,0.000044914468,0.00013103268,0.001186033,1.14710815e-7,5.1309405e-7,0.0000025591376,0.032647908],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993645,0.000045768997,0.00028682832,0.000044168915,0.00017928911,0.0000794257],"domain_scores_gemma":[0.9990539,0.000065634435,0.00025262023,0.000051932362,0.0005225108,0.00005340743],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00024556986,0.00006443178,0.00015932506,0.000052217285,0.00012180048,0.000054996362,0.00010448261,0.000014504961,0.0044403556],"category_scores_gemma":[0.000081713806,0.000026522042,0.00022143703,0.00003261611,0.00015721368,0.0003891046,0.000009096759,0.00006387698,0.000054273507],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007850607,0.000088839646,0.00060635735,0.000006965201,0.0004639525,0.000018884879,0.0069749053,0.000025206316,0.0024586956,0.950352,0.034778036,0.0041476614],"study_design_scores_gemma":[0.0006411591,0.00020893852,0.00045459246,0.00008956917,0.0002879893,0.00004593879,0.008970368,0.000031316147,0.0012084449,0.06893941,0.91892517,0.0001971379],"about_ca_topic_score_codex":0.000002788841,"about_ca_topic_score_gemma":0.000024075462,"teacher_disagreement_score":0.8841471,"about_ca_system_score_codex":0.000030379464,"about_ca_system_score_gemma":0.00001585847,"threshold_uncertainty_score":0.99646974},"labels":[],"label_agreement":null},{"id":"W4254847207","doi":"10.22148/001c.22331","title":"Against Conglomeration","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Poetry Analysis and Criticism","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Multiculturalism; Politics; The arts; Sociology; Government (linguistics); Endowment; Art history; Political science; Management; Media studies; Law; Art; Economics","score_opus":0.04628210556819579,"score_gpt":0.265659557897568,"score_spread":0.21937745232937222,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4254847207","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.84488994,0.0018909172,0.00024353925,0.0060272333,0.001165519,0.00002639709,0.00001521396,0.000014597861,0.14572662],"genre_scores_gemma":[0.98316884,0.00021421435,0.000088501896,0.00070691796,0.0011901133,1.1558848e-7,0.000009427344,0.0000036691579,0.014618173],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992113,0.000034010005,0.00037529165,0.000054677035,0.0002267725,0.00009794804],"domain_scores_gemma":[0.9985502,0.00002484131,0.0001934835,0.0000649526,0.0011021782,0.00006438577],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00012423773,0.00007133142,0.00020426337,0.000044147713,0.00013998982,0.00026522554,0.00007690241,0.000022105725,0.0012380397],"category_scores_gemma":[0.00009294851,0.000046118916,0.00024230401,0.000055467797,0.000066668974,0.00036758638,0.000011839443,0.00013524099,0.000026922691],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000045013778,0.00049628457,0.00046626257,0.00007084004,0.0015085468,0.000980592,0.029156448,0.0016632412,0.011653612,0.8129996,0.13028741,0.010672166],"study_design_scores_gemma":[0.0005212313,0.00010279797,0.00020303986,0.00007736882,0.00046672722,0.00012487287,0.028541911,0.0016959227,0.0036106962,0.0033816183,0.96104884,0.00022496165],"about_ca_topic_score_codex":0.000002609997,"about_ca_topic_score_gemma":0.00007029891,"teacher_disagreement_score":0.83076143,"about_ca_system_score_codex":0.000024851464,"about_ca_system_score_gemma":0.00004380895,"threshold_uncertainty_score":0.999675},"labels":[],"label_agreement":null},{"id":"W4281392358","doi":"10.22148/001c.35907","title":"What We Learned About the Humanities from a Study of Thousands of Newspaper Articles","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Digital humanities; Newspaper; Discipline; Cognitive reframing; Variety (cybernetics); Publics; Humanism; Sociology; Competition (biology); Humanities; Political science; Media studies; Social science; Art; Psychology; Social psychology; Politics; Computer science","score_opus":0.14411329786967156,"score_gpt":0.28644284727140595,"score_spread":0.1423295494017344,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281392358","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9901248,0.0021882073,4.4021945e-7,0.0005707425,0.00054895366,0.00009570994,0.000032246433,0.000006551703,0.0064323395],"genre_scores_gemma":[0.9843131,0.0002396779,0.0000057025713,0.00022181119,0.00042837404,0.000002286724,0.00000331317,0.000011841801,0.01477394],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99813527,0.0001412201,0.0007704893,0.000088473746,0.00071036996,0.0001541831],"domain_scores_gemma":[0.99839544,0.00016593128,0.00076470495,0.00017563096,0.00046519595,0.000033082735],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00037934296,0.00014050362,0.00039902344,0.000094952746,0.0005643346,0.0008328301,0.00048398302,0.000016192025,0.0023654297],"category_scores_gemma":[0.00003931235,0.00007781727,0.00026821287,0.000068423964,0.0003267094,0.0011634462,0.00014415986,0.00036226993,0.0000020993198],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046312757,0.001579591,0.001809249,0.000046935016,0.001188876,0.00005016095,0.90751725,0.0016489716,0.00031398833,0.070524044,0.009683993,0.005173812],"study_design_scores_gemma":[0.00070362474,0.0013892751,0.00084418984,0.000068304515,0.00027535766,0.000011173707,0.8387818,0.00005917775,0.00008532284,0.0058626137,0.15179648,0.00012267854],"about_ca_topic_score_codex":0.00024505545,"about_ca_topic_score_gemma":0.0014273555,"teacher_disagreement_score":0.14211248,"about_ca_system_score_codex":0.000036099933,"about_ca_system_score_gemma":0.000041033716,"threshold_uncertainty_score":0.99854654},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.08848718552736484,"score_gpt":0.371559383609055,"score_spread":0.28307219808169015,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283727265","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7615052,0.06983381,0.016050866,0.13543491,0.0037456863,0.0012240809,0.00020754279,0.000023058816,0.011974849],"genre_scores_gemma":[0.9943363,0.00022791953,0.003788868,0.00137568,0.00004716133,0.0000034172558,0.00001682961,0.000002111746,0.00020170394],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99832195,0.0004108332,0.000604532,0.00011048445,0.00045421664,0.00009796515],"domain_scores_gemma":[0.99902016,0.0002912967,0.00021831527,0.00033757323,0.000094875,0.000037773076],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001544739,0.0000729888,0.0002032613,0.000067873785,0.00011285356,0.00009477229,0.0013786058,0.000017547802,0.00001496651],"category_scores_gemma":[0.0002782511,0.000040456565,0.00004605034,0.00044918625,0.000057737925,0.00087454927,0.0006585378,0.0005687787,4.9631916e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022583043,0.002865263,0.018842151,0.00041956408,0.00007480842,0.00044681347,0.05839335,0.0023834466,0.0016925117,0.80151373,0.037671737,0.075470805],"study_design_scores_gemma":[0.0039286884,0.000769838,0.12882626,0.0039014893,0.00007816889,0.0015718294,0.022863239,0.30629063,0.00011241998,0.043144185,0.4874635,0.0010497514],"about_ca_topic_score_codex":0.000006315221,"about_ca_topic_score_gemma":0.00001430932,"teacher_disagreement_score":0.7583695,"about_ca_system_score_codex":0.000034398985,"about_ca_system_score_gemma":0.000121918725,"threshold_uncertainty_score":0.25618157},"labels":[],"label_agreement":null},{"id":"W4294203304","doi":"10.22148/001c.37588","title":"The Evolution of the Idiolect over the Lifetime: A Quantitative and Qualitative Study of French 19th Century Literature","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Agence Nationale de la Recherche","keywords":"Selection (genetic algorithm); Computer science; Variance (accounting); Task (project management); Feature (linguistics); Linguistics; Monotonic function; Natural language processing; Artificial intelligence; Literature; Mathematics; Philosophy; Art","score_opus":0.02960532437119118,"score_gpt":0.335989065498779,"score_spread":0.3063837411275878,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4294203304","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9868755,0.006013274,0.0038048967,0.0026344953,0.00040571208,0.00019034946,0.000016711198,0.000003719499,0.000055366876],"genre_scores_gemma":[0.99939364,0.000084239684,0.00028272124,0.000053572217,0.000035450845,0.0000012290571,4.683478e-7,0.0000021389976,0.00014653173],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9974196,0.0012526847,0.00047599684,0.0000946695,0.0006418761,0.00011516535],"domain_scores_gemma":[0.9977733,0.0005146111,0.0009201914,0.00019488546,0.00056295685,0.00003403683],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019333056,0.00009177339,0.0001934453,0.000048091304,0.0005070841,0.00008319516,0.00071760174,0.000024090674,0.000004394118],"category_scores_gemma":[0.0003376074,0.000035819026,0.00014178327,0.00094823324,0.0001222887,0.00022136405,0.00027110046,0.0005219941,1.6540487e-7],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016365759,0.0003493616,0.0040018437,0.00003180373,0.0006393254,0.000008428009,0.8533538,0.0041140094,0.0013989321,0.13160257,0.0036055283,0.00073076884],"study_design_scores_gemma":[0.0019482368,0.0039872285,0.090539455,0.00023058859,0.00036144562,0.0001934514,0.8265563,0.049978867,0.0004890566,0.02120169,0.00417963,0.00033401133],"about_ca_topic_score_codex":0.000037653175,"about_ca_topic_score_gemma":0.000017562255,"teacher_disagreement_score":0.11040088,"about_ca_system_score_codex":0.00009202706,"about_ca_system_score_gemma":0.000105790765,"threshold_uncertainty_score":0.39001325},"labels":[],"label_agreement":null},{"id":"W4307445554","doi":"10.22148/001c.38739","title":"Biodiversity is not declining in fiction","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language and cultural evolution","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Replicate; Matching (statistics); Taxon; Analytics; Computer science; Diversity (politics); Biodiversity; Data science; Artificial intelligence; Natural language processing; Information retrieval; Sociology; Statistics; Ecology; Mathematics; Biology","score_opus":0.05539905594160905,"score_gpt":0.32604144732637425,"score_spread":0.2706423913847652,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4307445554","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9927498,0.0001673829,0.000013438949,0.0043465644,0.00033853707,0.000032328087,0.000005884773,0.000006676732,0.0023393498],"genre_scores_gemma":[0.99773425,0.00014207837,0.0000939429,0.00065273116,0.0001934726,2.3884803e-7,0.0000021056835,9.376421e-7,0.0011802451],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99902415,0.0001162893,0.00022325729,0.000057459314,0.00045320173,0.00012565218],"domain_scores_gemma":[0.99948865,0.00002412531,0.0002346989,0.000032056138,0.00016444126,0.00005604641],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00056730193,0.000044863096,0.00011200399,0.00006627567,0.0004742064,0.000037950187,0.00016220118,0.000030831772,0.0006608574],"category_scores_gemma":[0.00011206642,0.000033105454,0.00011645342,0.00037672883,0.00004164918,0.0003544991,0.00003777034,0.0002481927,0.000007960026],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050189,0.0004491405,0.1172084,0.000020161846,0.00024951735,0.00047813365,0.49239367,0.0089283185,0.004600443,0.0030236295,0.35469925,0.017447442],"study_design_scores_gemma":[0.0013731583,0.00033921617,0.042079076,0.0000305939,0.00016228856,0.000075515105,0.47901848,0.0007895542,0.00037752735,0.0006396379,0.4747857,0.0003292524],"about_ca_topic_score_codex":0.0007574987,"about_ca_topic_score_gemma":0.00035407147,"teacher_disagreement_score":0.120086454,"about_ca_system_score_codex":0.00031839358,"about_ca_system_score_gemma":0.00006618431,"threshold_uncertainty_score":0.7235925},"labels":[],"label_agreement":null},{"id":"W4313316689","doi":"10.22148/001c.55593","title":"Foreword to the Special Issue “Theorytellings: Epistemic Narratives in the Digital Humanities”","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Scholarship; Neologism; Digital humanities; Field (mathematics); Epistemology; Sociology; Humanities; Philosophy; Linguistics; Political science; Mathematics","score_opus":0.05363414757406019,"score_gpt":0.25503167378603253,"score_spread":0.20139752621197235,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313316689","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6175773,0.00043778875,0.0000075249577,0.015777677,0.0019302802,0.00030031468,0.00018795201,0.000012841381,0.36376834],"genre_scores_gemma":[0.93698055,0.000012249745,0.00000609996,0.0025031178,0.008698454,0.000006136309,0.000008783484,0.000011887512,0.051772736],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983137,0.00012212362,0.0005739349,0.00010174793,0.0006432664,0.00024518024],"domain_scores_gemma":[0.9991387,0.00015362096,0.0003050486,0.00014802866,0.00021065873,0.000043932152],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00070781633,0.0001640442,0.0002470663,0.00011782725,0.0009503766,0.0014802443,0.0008968169,0.00001629723,0.002991722],"category_scores_gemma":[0.00007338924,0.00007931388,0.00025393435,0.000113310576,0.00022714578,0.0007856642,0.000111068664,0.0006474908,0.000038994745],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000118171214,0.00011826015,0.000052745014,0.000009483886,0.000069253605,0.000072405266,0.42882052,0.0005971991,0.0000014611165,0.41401377,0.15394282,0.0021838984],"study_design_scores_gemma":[0.00011431082,0.0002810214,0.000035313125,0.000015825955,0.000014771152,0.00006782445,0.34910887,0.000012602984,9.713235e-7,0.013812682,0.6364464,0.000089373265],"about_ca_topic_score_codex":0.000011629817,"about_ca_topic_score_gemma":0.00051743997,"teacher_disagreement_score":0.48250362,"about_ca_system_score_codex":0.00011121199,"about_ca_system_score_gemma":0.000037191945,"threshold_uncertainty_score":0.9995563},"labels":[],"label_agreement":null},{"id":"W4313318980","doi":"10.22148/001c.57197","title":"Grounding Theory in Digital Data: A Methodological Approach for a Reflective Procedural Framework","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Computer science; Process (computing); Set (abstract data type); Frame (networking); Data science; Grounded theory; Scale (ratio); Epistemology; Artificial intelligence; Qualitative research; Sociology","score_opus":0.3585206163766748,"score_gpt":0.502581587722082,"score_spread":0.14406097134540713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313318980","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08220347,0.00044786814,0.90982574,0.0024634874,0.00034009063,0.0002774337,0.000037584658,0.00001814139,0.0043861875],"genre_scores_gemma":[0.79389215,0.000017858358,0.20472705,0.00019595379,0.0005360773,0.000006695975,0.00002345886,0.000005054451,0.0005956903],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9977883,0.00074637955,0.0004643134,0.00018415012,0.0005967219,0.00022011937],"domain_scores_gemma":[0.9968484,0.0022231513,0.00043025363,0.000091907925,0.0003205467,0.000085709915],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0067230253,0.00009466244,0.00035599712,0.00015773324,0.00049562356,0.00017316538,0.0006262531,0.000052509775,0.00004357213],"category_scores_gemma":[0.0066064456,0.00006501826,0.00022434056,0.0009238794,0.00012240834,0.0006505704,0.0001765682,0.00047822617,2.9955274e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012914886,0.00085841335,0.0052574286,0.00004533424,0.000840322,0.000053795167,0.046868756,0.039485402,0.00007967271,0.8483292,0.0029710038,0.053919178],"study_design_scores_gemma":[0.0005416151,0.0003059332,0.0014415315,0.000022824823,0.00024328442,0.000078754085,0.10454743,0.021050192,0.0000037408447,0.85481364,0.01668793,0.0002631364],"about_ca_topic_score_codex":0.000015773256,"about_ca_topic_score_gemma":0.000013025986,"teacher_disagreement_score":0.7116887,"about_ca_system_score_codex":0.00027878355,"about_ca_system_score_gemma":0.00020014399,"threshold_uncertainty_score":0.79090095},"labels":[],"label_agreement":null},{"id":"W4313363291","doi":"10.22148/001c.55507","title":"Conceptual Forays: A Corpus-based Study of “Theory” in Digital Humanities Journals","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Narratology; Subject (documents); Perspective (graphical); Digital humanities; Semantics (computer science); Literary theory; Interpretation (philosophy); Epistemology; Field (mathematics); Heuristic; Sociology; Computer science; Linguistics; Humanities; Artificial intelligence; Literary criticism; Mathematics; Philosophy; World Wide Web","score_opus":0.1228525614456737,"score_gpt":0.28428661952384965,"score_spread":0.16143405807817596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313363291","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9849203,0.0006436702,0.0000045118745,0.000059267764,0.00039944664,0.00014078437,0.00009781727,0.000009195943,0.0137249995],"genre_scores_gemma":[0.9928706,0.000007483472,0.0000055971113,0.00019901179,0.00032714242,0.0000036759711,0.000007943226,0.000016806112,0.0065617044],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.997522,0.000118945914,0.0011332766,0.00011209031,0.00086780335,0.00024590892],"domain_scores_gemma":[0.99802864,0.00018864298,0.000957643,0.00012710216,0.00062529003,0.000072668714],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00071274984,0.0001932293,0.0005416254,0.00037274265,0.0003750585,0.00072726735,0.00048875116,0.000022133914,0.0023901053],"category_scores_gemma":[0.00014074297,0.00014135716,0.00029525365,0.00011552394,0.0002810981,0.0013942274,0.00010281568,0.00058431504,0.0000026967807],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016070352,0.0055099064,0.015671097,0.00006629801,0.00083263003,0.000748167,0.40110293,0.009709148,0.000046028184,0.5558152,0.006225729,0.0026657914],"study_design_scores_gemma":[0.0028398978,0.0049438253,0.00071801955,0.000097385986,0.00015678669,0.00006718638,0.8950974,0.00009462388,0.00001982167,0.013966522,0.08166709,0.00033147333],"about_ca_topic_score_codex":0.000023877532,"about_ca_topic_score_gemma":0.000395476,"teacher_disagreement_score":0.5418487,"about_ca_system_score_codex":0.00014376699,"about_ca_system_score_gemma":0.00009141645,"threshold_uncertainty_score":0.99852186},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.14116809108927333,"score_gpt":0.3041771103153833,"score_spread":0.16300901922610997,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313363893","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9836583,0.004211089,0.0000018863445,0.0016053675,0.0013329567,0.000045435594,0.00012235221,0.000016886746,0.009005705],"genre_scores_gemma":[0.9846882,0.000018960953,0.00006215028,0.0010881051,0.0008175079,8.8779905e-7,0.000013341331,0.000007323318,0.013303526],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988353,0.00004893769,0.0004329101,0.000070283284,0.00048513417,0.00012742126],"domain_scores_gemma":[0.9984866,0.00006756457,0.0006499947,0.000054573567,0.00068631786,0.00005496082],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00017034604,0.000111012036,0.00025616126,0.00009397336,0.000594059,0.0004450637,0.00022688785,0.000009148191,0.0015676297],"category_scores_gemma":[0.00005509283,0.00007328399,0.00020559461,0.000044865355,0.0001540343,0.0008152916,0.00009615473,0.00030365368,0.000005024236],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018125298,0.00019526179,0.0013609374,0.00010488451,0.0012657325,0.00057072204,0.17494836,0.013080464,0.0000072901503,0.53612757,0.2685044,0.0036531226],"study_design_scores_gemma":[0.0005033113,0.00034463152,0.001106235,0.000053105134,0.00012474082,0.00032205798,0.13884689,0.00029149122,0.0000029507426,0.22279955,0.63534904,0.00025600469],"about_ca_topic_score_codex":0.0000019986821,"about_ca_topic_score_gemma":0.000022179305,"teacher_disagreement_score":0.36684462,"about_ca_system_score_codex":0.000103620834,"about_ca_system_score_gemma":0.00002315816,"threshold_uncertainty_score":0.99934506},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Operationalization; Realm; Epistemology; Process (computing); Construct (python library); Computer science; Sociology; Political science; Philosophy","score_opus":0.20341132908151102,"score_gpt":0.4160787014155996,"score_spread":0.21266737233408858,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313409859","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99522674,0.00004888878,0.002299206,0.00080081366,0.0003667583,0.000049918617,0.00010736511,0.000002448328,0.0010978422],"genre_scores_gemma":[0.9936478,0.0000026988762,0.00035971397,0.00014445101,0.00010199061,2.6773304e-7,0.000012946771,0.0000019208856,0.0057281693],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980307,0.000056307865,0.000453379,0.00015762514,0.0012294921,0.00007244764],"domain_scores_gemma":[0.99855995,0.00025816407,0.00041568629,0.00015804639,0.0005471263,0.000061033144],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007466531,0.00006275346,0.00018916253,0.00014874242,0.00017988845,0.0005128946,0.0003884746,0.000010393867,0.0008962275],"category_scores_gemma":[0.0009333392,0.000040970343,0.00006868798,0.0004119571,0.00006388861,0.00056947506,0.0003578818,0.000080189086,0.000021903246],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005259031,0.0006153151,0.029834136,0.00001870564,0.0004177375,0.00007687204,0.034054365,0.15091982,0.0068900296,0.01165716,0.67974627,0.085243694],"study_design_scores_gemma":[0.0026411996,0.0021746494,0.060609605,0.00013389334,0.00026277493,0.0002412825,0.13671455,0.10497566,0.0018032936,0.061957814,0.627681,0.000804256],"about_ca_topic_score_codex":0.000031075422,"about_ca_topic_score_gemma":0.000012417624,"teacher_disagreement_score":0.10266019,"about_ca_system_score_codex":0.000039728573,"about_ca_system_score_gemma":0.0000468719,"threshold_uncertainty_score":0.98130625},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Digital humanities; Hermeneutics; Humanities; Epistemology; Sociology; Philosophy","score_opus":0.15983666033118077,"score_gpt":0.3027010683654781,"score_spread":0.14286440803429734,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4319604386","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94758046,0.0012250408,0.000018829862,0.009104247,0.0006882446,0.0015505534,0.00042883892,0.00005106503,0.03935271],"genre_scores_gemma":[0.9843513,0.00019146022,0.000009955105,0.00022468151,0.00081950065,0.000036873338,0.000018425688,0.000022941456,0.014324894],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9976018,0.00010875655,0.0010657943,0.00015866426,0.0007081197,0.0003568365],"domain_scores_gemma":[0.99709994,0.000686295,0.0009756521,0.0003222404,0.00088998355,0.000025902382],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0013721574,0.0002735601,0.0005184115,0.0001629617,0.0006480043,0.0011286272,0.0010648873,0.000067610425,0.00005915983],"category_scores_gemma":[0.00027805273,0.000113501956,0.0005547445,0.00020691751,0.0012347074,0.0011616,0.00011267684,0.0006166645,0.0000026497592],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005345201,0.0004380915,0.0005175551,0.0007336034,0.000918724,0.00005781012,0.37037516,0.0007105683,0.00094201334,0.5725064,0.046834324,0.0054311906],"study_design_scores_gemma":[0.0010513611,0.0014137793,0.001953437,0.00078920374,0.00051444024,0.00025661834,0.4821534,0.0002650976,0.00052238716,0.037383575,0.47316843,0.00052828586],"about_ca_topic_score_codex":0.0000954247,"about_ca_topic_score_gemma":0.0014923752,"teacher_disagreement_score":0.5351229,"about_ca_system_score_codex":0.00007547786,"about_ca_system_score_gemma":0.00009454324,"threshold_uncertainty_score":0.99990827},"labels":[],"label_agreement":null},{"id":"W4320020468","doi":"10.22148/001c.57764","title":"Theory-Driven Statistics for the Digital Humanities: Presenting Pitfalls and a Practical Guide by the Example of the Reformation","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Data Analysis with R","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Credibility; Test (biology); Digital humanities; Statistical hypothesis testing; Bridge (graph theory); Protestantism; Logistic regression; Psychology; Data science; Computer science; Epistemology; Sociology; Political science; Statistics; Mathematics; Philosophy; Library science; Machine learning; Law","score_opus":0.058927451875183105,"score_gpt":0.32210626912233287,"score_spread":0.26317881724714975,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4320020468","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.043034922,0.00018839106,0.9413465,0.013747396,0.00035561755,0.00040663368,0.0002647705,0.0000274893,0.0006282443],"genre_scores_gemma":[0.97844553,0.00028151504,0.01474481,0.0006389774,0.00033384954,0.0000063910034,0.000034067914,0.000016427259,0.005498432],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986001,0.0000981947,0.00054160977,0.00008780719,0.0005128516,0.00015938542],"domain_scores_gemma":[0.99704546,0.0013583436,0.0007801916,0.0003144228,0.00046873032,0.000032839314],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010968902,0.00009091635,0.00016754409,0.000040734936,0.00030308665,0.00059033465,0.0007734401,0.000025091364,0.000002606635],"category_scores_gemma":[0.001322743,0.000033517295,0.00012087161,0.00034942632,0.00014962666,0.0010174933,0.00030985702,0.00017875702,0.000002046769],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004187014,0.0000923163,0.0012539846,0.00008898859,0.001177312,0.000012004589,0.013917673,0.00390124,0.0005530226,0.36789435,0.60018766,0.010879564],"study_design_scores_gemma":[0.000674113,0.0002111994,0.00455327,0.00008564604,0.00067435665,0.00030242963,0.011432327,0.7311493,0.00045418888,0.029863568,0.22035871,0.00024088257],"about_ca_topic_score_codex":0.000022509743,"about_ca_topic_score_gemma":0.000018468458,"teacher_disagreement_score":0.9354106,"about_ca_system_score_codex":0.000030171335,"about_ca_system_score_gemma":0.000058515125,"threshold_uncertainty_score":0.5692611},"labels":[],"label_agreement":null},{"id":"W4320058452","doi":"10.22148/001c.55508","title":"Sailing on Encrypted Seas: The Archive and Digital Memory in African and Diasporic Futurism","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Diverse Musicological Studies","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Digitization; Humanity; Scholarship; Faith; Futurist; Digital scholarship; Visual arts; Sociology; History; Art; Art history; Computer science; World Wide Web; Political science; Law; Telecommunications; Epistemology; Philosophy","score_opus":0.08162255771898413,"score_gpt":0.23602104120717565,"score_spread":0.15439848348819152,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4320058452","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9845558,0.00030698345,5.5199257e-7,0.0021381178,0.00015747517,0.000049071998,0.000013181464,0.000010069632,0.012768744],"genre_scores_gemma":[0.99769396,0.00036940104,0.000005992529,0.00017952315,0.0003948924,4.8057944e-7,0.0000013660933,0.0000031080378,0.0013512702],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99939495,0.000023829138,0.0002001453,0.00007170247,0.00017920013,0.00013014635],"domain_scores_gemma":[0.9994935,0.0002212196,0.00012581277,0.00003497409,0.00007478178,0.00004974686],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00013317422,0.00008817802,0.00018148888,0.00007816268,0.00016489335,0.00028038368,0.000083717016,0.000015693111,0.00003913051],"category_scores_gemma":[0.00013449207,0.000040130937,0.000057358604,0.000080022255,0.00025170535,0.00017633881,0.00007312915,0.00019481072,0.000008449448],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006104128,0.00042381245,0.011041527,0.00013014197,0.0012592283,0.0015277255,0.5600026,0.001457489,0.00018714281,0.21488525,0.080466084,0.12800857],"study_design_scores_gemma":[0.0016277998,0.0011273961,0.15835911,0.00020679404,0.00025147106,0.000060650156,0.56633794,0.0013775363,0.000012355227,0.030450659,0.23970146,0.0004868154],"about_ca_topic_score_codex":0.000019651494,"about_ca_topic_score_gemma":0.00008583836,"teacher_disagreement_score":0.1844346,"about_ca_system_score_codex":0.00002239038,"about_ca_system_score_gemma":0.0000057478114,"threshold_uncertainty_score":0.27037463},"labels":[],"label_agreement":null},{"id":"W4320484258","doi":"10.22148/001c.68188","title":"How Network Analysis Uncovers International Networks of Smuggling History: Criminals in Nagasaki, Japan circa 1667","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"China's Socioeconomic Reforms and Governance","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Washington University in St. Louis","keywords":"Magistrate; History; Criminology; Geography; Genealogy; Law; Political science; Sociology; Archaeology","score_opus":0.04108431609154059,"score_gpt":0.3001960300701222,"score_spread":0.25911171397858157,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4320484258","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98892516,0.00093568035,0.0004447254,0.0046132915,0.0016608726,0.00006727169,0.000006673521,0.000013950238,0.0033323932],"genre_scores_gemma":[0.9916997,0.0026848102,0.00015187537,0.00013718911,0.001098703,6.104976e-7,0.0000070801934,0.000007082331,0.0042129722],"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","domain_scores_codex":[0.99831444,0.000076538185,0.00061923184,0.00013780393,0.0005335747,0.00031839396],"domain_scores_gemma":[0.99810207,0.00013285982,0.0011873274,0.00010473176,0.00035387927,0.00011915434],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012828169,0.00012631099,0.0005150142,0.00023507066,0.00010630175,0.0000943698,0.00045962384,0.000116108786,0.000101519836],"category_scores_gemma":[0.0003199892,0.00009611811,0.000529864,0.00096160034,0.000171861,0.0006289942,0.000049576654,0.00030641715,0.0000024942165],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009953187,0.00007310458,0.23669747,0.000021217196,0.0017382854,0.000057590387,0.0285606,0.67917484,0.000040191728,0.008343635,0.041276567,0.0039169705],"study_design_scores_gemma":[0.0022254821,0.0002693368,0.28253913,0.0004968165,0.001997357,0.00001617594,0.08391691,0.12165539,0.000023117844,0.006454073,0.4994083,0.0009978979],"about_ca_topic_score_codex":0.0009854478,"about_ca_topic_score_gemma":0.0036670356,"teacher_disagreement_score":0.55751944,"about_ca_system_score_codex":0.001173981,"about_ca_system_score_gemma":0.00018710011,"threshold_uncertainty_score":0.39195827},"labels":[],"label_agreement":null},{"id":"W4321367095","doi":"10.22148/001c.68086","title":"‘A pretty sublime mix of WTF and OMG’. Four explorations into the practice of evaluation on online book reviewing platforms","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Games and Media","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Disappointment; Sketch; Reading (process); Literal (mathematical logic); Character (mathematics); Sublime; Style (visual arts); Computer science; World Wide Web; Literature; Psychology; Linguistics; Art; Philosophy","score_opus":0.12924748400913003,"score_gpt":0.4050283872097174,"score_spread":0.27578090320058735,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321367095","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95974314,0.0039806976,0.00004314086,0.020844521,0.00039998934,0.00036429468,0.000010426709,0.000010972971,0.0146027915],"genre_scores_gemma":[0.9796561,0.01488806,0.00052381057,0.00039769564,0.00036932,0.0000017216528,0.000009782502,0.0000055800124,0.0041479524],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99856,0.00007761878,0.00044851546,0.00005512194,0.0007609335,0.00009781349],"domain_scores_gemma":[0.99775857,0.00034870737,0.00069280557,0.00007126547,0.0010605962,0.00006808249],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0020068183,0.000060059945,0.00019087236,0.00006763725,0.00007975675,0.00005478899,0.00011692721,0.000036972764,0.000017438088],"category_scores_gemma":[0.0041200193,0.000031257718,0.0001033042,0.00043884627,0.00012629676,0.0010768562,0.000023369163,0.00013140088,0.0000023784962],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037190222,0.0006733354,0.0018002082,0.00034406112,0.0010355676,0.00003671328,0.3643628,0.0066820933,0.0024446691,0.025136271,0.061708122,0.53540426],"study_design_scores_gemma":[0.00088370545,0.000834941,0.005132154,0.00095892296,0.0010415433,0.000026335261,0.1724216,0.004829445,0.00039226795,0.005147205,0.80811137,0.00022047923],"about_ca_topic_score_codex":0.00002533601,"about_ca_topic_score_gemma":0.00012422359,"teacher_disagreement_score":0.7464033,"about_ca_system_score_codex":0.000051215404,"about_ca_system_score_gemma":0.00015304262,"threshold_uncertainty_score":0.49323457},"labels":[],"label_agreement":null},{"id":"W4328051182","doi":"10.22148/001c.55795","title":"From the Archive to the Computer: Michel Foucault and the Digital Humanities","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Digital humanities; Historiography; Context (archaeology); Michel foucault; Reading (process); Problematization; Epistemology; Field (mathematics); Humanities; Sociology; Digitization; Art history; Philosophy; History; Linguistics; Politics; Computer science; Archaeology; Political science","score_opus":0.06118782381022655,"score_gpt":0.24706904804020183,"score_spread":0.18588122422997527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4328051182","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8995785,0.0009067992,0.000053344676,0.04791759,0.0016846218,0.00031753932,0.00045384673,0.00005024013,0.0490375],"genre_scores_gemma":[0.973047,0.00009119906,0.00001586902,0.004067265,0.0048267664,0.0000022658212,0.000020006953,0.0000122444335,0.01791738],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99891824,0.00006230249,0.00037476214,0.00008781154,0.00036389148,0.00019298046],"domain_scores_gemma":[0.998589,0.0007595385,0.00020402679,0.00015407949,0.00023910512,0.00005429604],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.00035691736,0.00014635344,0.0002272295,0.00004082423,0.00078474067,0.0036522066,0.00056507246,0.000017228253,0.00008873628],"category_scores_gemma":[0.00010591345,0.000050269915,0.00022495481,0.000056936595,0.00065118034,0.0005311852,0.00018237492,0.0003318753,0.00007273067],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012167289,0.000022017586,0.0000876426,0.0000071835816,0.0003670406,0.00002349212,0.20650275,0.00019609721,0.0000019118256,0.2808667,0.50708425,0.0047192164],"study_design_scores_gemma":[0.00043632943,0.00011885324,0.00092173496,0.00005705691,0.000103188424,0.000028152805,0.101297684,0.0005540331,0.0000015900998,0.028120074,0.8682455,0.000115847615],"about_ca_topic_score_codex":0.00012313452,"about_ca_topic_score_gemma":0.00084593665,"teacher_disagreement_score":0.36116117,"about_ca_system_score_codex":0.000015716678,"about_ca_system_score_gemma":0.00001809664,"threshold_uncertainty_score":0.9973821},"labels":[],"label_agreement":null},{"id":"W4365998489","doi":"10.22148/001c.38966","title":"Heritage site-seeing through the visitor’s lens on Instagram","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Diverse Aspects of Tourism Research","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"University of Oxford","keywords":"Visitor pattern; Tourism; Heritage tourism; Context (archaeology); Social media; Cultural heritage; Narrative; Object (grammar); Destinations; Historic site; Geography; Visual arts; Cultural heritage management; Archaeology; Computer science; World Wide Web; Art","score_opus":0.06141230905664609,"score_gpt":0.35667579298470764,"score_spread":0.29526348392806157,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4365998489","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77656573,0.0005094863,0.000028126851,0.05640073,0.0014378068,0.00023135221,0.000017680755,0.00003454283,0.16477455],"genre_scores_gemma":[0.99100393,0.00025105476,0.00017209469,0.0007177259,0.0010825783,9.962853e-7,9.972221e-7,0.000007570005,0.0067630596],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971718,0.00037217105,0.00029186715,0.000103513914,0.0017402554,0.0003203595],"domain_scores_gemma":[0.9989278,0.00021626831,0.00030700935,0.0001396475,0.0003165533,0.000092729584],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0011644269,0.00009611236,0.00018599535,0.00006789397,0.0015571347,0.0002596574,0.00077147014,0.000035536454,0.000691386],"category_scores_gemma":[0.00043498242,0.000057585123,0.00024681634,0.00049394084,0.0002127289,0.00058467255,0.00017775089,0.0007685815,0.000038231112],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045130806,0.0008749514,0.003313153,0.000025126137,0.0007506032,0.0031998036,0.16344002,0.040691443,0.0010599087,0.12852931,0.6497427,0.00792164],"study_design_scores_gemma":[0.0003320829,0.0003715317,0.0004497938,0.000016094295,0.000053880132,0.000020903033,0.115712345,0.00016408414,0.00005696316,0.0016402649,0.88104975,0.00013230012],"about_ca_topic_score_codex":0.0006060402,"about_ca_topic_score_gemma":0.00017804351,"teacher_disagreement_score":0.23130703,"about_ca_system_score_codex":0.00036078776,"about_ca_system_score_gemma":0.00018244353,"threshold_uncertainty_score":0.9997427},"labels":[],"label_agreement":null},{"id":"W4366714876","doi":"10.22148/001c.74293","title":"Searching maps by words: how machine learning changes the way we explore map collections","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Geographic Information Systems Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Arts and Humanities Research Council; Alan Turing Institute; Austrian Institute of Technology; University of Minnesota; University of Southern California","keywords":"Interactivity; Mind map; Computer science; Curiosity; Interface (matter); Information retrieval; World Wide Web; Visualization; Beauty; Annotation; Artificial intelligence; Art","score_opus":0.08158655033373287,"score_gpt":0.3221743146279093,"score_spread":0.24058776429417644,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366714876","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1726082,0.006688251,0.0005312781,0.7918218,0.0036996957,0.00076871517,0.000053927644,0.0003589935,0.023469191],"genre_scores_gemma":[0.9108387,0.006676155,0.0000723703,0.00014720773,0.0007177538,0.000006579232,0.00000706497,0.0000087677245,0.081525356],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99836737,0.00021526191,0.00027933958,0.00006306132,0.0007782215,0.0002967683],"domain_scores_gemma":[0.9986662,0.00024108589,0.00042256515,0.00006311821,0.0005125951,0.0000944051],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0015943871,0.000101394406,0.00020602785,0.00023789595,0.0023981454,0.00040987955,0.00027601924,0.000051492214,0.00003166941],"category_scores_gemma":[0.0004639359,0.00005598645,0.0001578198,0.0016889261,0.00018618719,0.0004007946,0.00006187849,0.00044400545,0.000026389558],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021261645,0.00002415205,0.0046360246,0.00006372181,0.0005565126,0.000016697288,0.39960524,0.0017693645,0.00020310187,0.0027041235,0.58768487,0.0027149415],"study_design_scores_gemma":[0.0001391014,0.00004806541,0.00014663291,0.000060737737,0.00003695893,0.0000074404516,0.32132822,0.0005311062,0.00001590049,0.00034237964,0.6772656,0.00007787422],"about_ca_topic_score_codex":0.00025299026,"about_ca_topic_score_gemma":0.0014907107,"teacher_disagreement_score":0.79167455,"about_ca_system_score_codex":0.0000848864,"about_ca_system_score_gemma":0.00003930152,"threshold_uncertainty_score":0.9989006},"labels":[],"label_agreement":null},{"id":"W4379467399","doi":"10.22148/001c.74602","title":"Text-mining metadata: What can titles tell us of the history of modern and contemporary art?","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Metadata; Reading (process); Digitization; Contemporary art; Computer science; Scholarship; The arts; Topic model; World Wide Web; Visual arts; History; Information retrieval; Art; Linguistics; Art history","score_opus":0.12504934775014093,"score_gpt":0.3525318226868839,"score_spread":0.22748247493674295,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379467399","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9811861,0.007038036,0.0003426118,0.0037265506,0.0007601864,0.0000652109,0.000008509886,0.000008352142,0.006864443],"genre_scores_gemma":[0.9784748,0.0007653999,0.0005402717,0.00007779386,0.00011848865,1.3604318e-7,0.0000017882688,0.000003177735,0.020018155],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99887645,0.00016796589,0.0003728279,0.000060305785,0.0004422151,0.00008022657],"domain_scores_gemma":[0.9987398,0.0002495827,0.0005424018,0.00006274703,0.00034883682,0.000056631],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00085353333,0.00005996249,0.00026036362,0.00008942784,0.00006966204,0.00004290119,0.0001859949,0.00003405546,0.00002642958],"category_scores_gemma":[0.00024441784,0.000034169225,0.00018158091,0.00031741254,0.00025914592,0.00047957696,0.00004579465,0.000092534494,8.1562206e-7],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030985422,0.00047215694,0.113075465,0.00040526924,0.0048152753,0.00012290227,0.28857112,0.029660791,0.01925834,0.03165047,0.3867188,0.124939546],"study_design_scores_gemma":[0.0018825334,0.00040230647,0.1216696,0.0010841995,0.00221687,0.00005691987,0.12688264,0.046841677,0.0011358894,0.049384594,0.64759856,0.00084421894],"about_ca_topic_score_codex":0.00015004366,"about_ca_topic_score_gemma":0.00038469658,"teacher_disagreement_score":0.26087973,"about_ca_system_score_codex":0.00007300614,"about_ca_system_score_gemma":0.00028918177,"threshold_uncertainty_score":0.13933806},"labels":[],"label_agreement":null},{"id":"W4381708035","doi":"10.22148/001c.73249","title":"An Ever-Expanding World Literary Genre: Defining Magic Realism on Wikipedia","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Literature, Magical Realism, García Márquez","field":"Arts and Humanities","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Magic realism; MAGIC (telescope); Realism; Literature; Scholarship; Phenomenon; Philosophical realism; History; Epistemology; Art; Philosophy; Political science; Law","score_opus":0.056729037839126216,"score_gpt":0.3107540225573207,"score_spread":0.2540249847181945,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4381708035","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90250576,0.0011913909,0.000015702191,0.0033699411,0.0025947376,0.00016456723,0.00017101024,0.00022441414,0.08976251],"genre_scores_gemma":[0.96294105,0.000215553,0.0002489941,0.0019226527,0.0038504181,0.0000012228471,0.00012848595,0.000039380244,0.030652234],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99776465,0.000111834735,0.00076846045,0.00022006758,0.0007053504,0.0004296361],"domain_scores_gemma":[0.9982481,0.00022187407,0.0004927852,0.00023533647,0.0004932439,0.0003086448],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004947734,0.0002812921,0.00045157768,0.00048675414,0.00026330375,0.0010327285,0.00041372786,0.00007536576,0.00058137346],"category_scores_gemma":[0.00015833453,0.00017591876,0.00032418713,0.0003664706,0.00013721432,0.0011490472,0.000052709074,0.0006956051,0.00018884508],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040923015,0.00030591385,0.0015270398,0.00014121452,0.00047643934,0.001774534,0.113227114,0.0011338617,0.0006804812,0.5902778,0.2805724,0.009473979],"study_design_scores_gemma":[0.0015976881,0.001122026,0.004891376,0.0008899905,0.00042054802,0.00019235311,0.009689136,0.0039955354,0.00020825616,0.053147633,0.922839,0.0010064432],"about_ca_topic_score_codex":0.000028462913,"about_ca_topic_score_gemma":0.00010958378,"teacher_disagreement_score":0.64226663,"about_ca_system_score_codex":0.00010812034,"about_ca_system_score_gemma":0.000048934824,"threshold_uncertainty_score":0.9958624},"labels":[],"label_agreement":null},{"id":"W4382314620","doi":"10.22148/001c.73258","title":"Escritor / Qillqaq: The Representation of Peruvian Literature in the Spanish and Quechua Wikipedias","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Wikis in Education and Collaboration","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Indigenous; Representation (politics); Politics; Relation (database); Population; Product (mathematics); Sociology; Linguistics; Geography; Ethnology; Anthropology; Political science; Computer science; Demography","score_opus":0.04461075900244968,"score_gpt":0.3820155821576306,"score_spread":0.33740482315518094,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382314620","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9471729,0.00025020246,0.0000055656333,0.04978048,0.00048607978,0.00009889436,0.0000020575424,0.000004661366,0.0021991543],"genre_scores_gemma":[0.99620074,0.0013402198,0.00006105067,0.00025143012,0.0007518344,0.0000011478287,0.0000027323672,0.0000018422855,0.0013889928],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9990101,0.0002351901,0.00024345542,0.000047394835,0.00038293097,0.000080923186],"domain_scores_gemma":[0.9990351,0.00015893656,0.000217622,0.000066651475,0.0004921775,0.000029495133],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008820136,0.00004273809,0.000091250666,0.000064716536,0.00016525474,0.00019280842,0.00017761752,0.00004593939,0.000014440236],"category_scores_gemma":[0.00073230395,0.000020599191,0.000048416132,0.0012756631,0.00012488771,0.00033617686,0.000009099526,0.00016081352,0.0000020652424],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004905496,0.0001090735,0.05863532,0.000029821153,0.000061903265,0.000021594577,0.76911736,0.00039156384,0.0011346757,0.04098604,0.12202733,0.0074362536],"study_design_scores_gemma":[0.00033258955,0.00009507872,0.20976202,0.00010612276,0.0000819149,0.000020700141,0.6809482,0.00043229735,0.0002049004,0.008308193,0.09959481,0.000113165785],"about_ca_topic_score_codex":0.00017145295,"about_ca_topic_score_gemma":0.0010032589,"teacher_disagreement_score":0.1511267,"about_ca_system_score_codex":0.000031538308,"about_ca_system_score_gemma":0.00009860153,"threshold_uncertainty_score":0.1859256},"labels":[],"label_agreement":null},{"id":"W4382315002","doi":"10.22148/001c.68341","title":"Italian Nostalgia: National and Global Identities of the Italian Novel","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Italian Fascism and Post-war Society","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"National identity; Resistance (ecology); Distribution (mathematics); Consumption (sociology); Sociology; Identity (music); Media studies; History; Political science; Aesthetics; Social science; Art; Law","score_opus":0.03774417578911504,"score_gpt":0.32238682909809296,"score_spread":0.2846426533089779,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382315002","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9682722,0.00024764493,0.00001907946,0.009817341,0.0009830326,0.00008460053,0.00015950917,0.000015004192,0.020401604],"genre_scores_gemma":[0.9912082,0.00015419818,0.000111735855,0.00017812315,0.00033072516,2.1035946e-7,0.00000283404,0.0000031838379,0.008010786],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9985129,0.000049881088,0.00032196185,0.00006600588,0.00087652256,0.00017268417],"domain_scores_gemma":[0.99881977,0.000057269124,0.00032022066,0.00005454513,0.000658291,0.00008988667],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000647669,0.00007874739,0.00016198347,0.000031800355,0.00031613198,0.000111675996,0.0002801329,0.000075550764,0.000045380766],"category_scores_gemma":[0.0003958639,0.000048347618,0.00023077142,0.0006546577,0.00042854925,0.00040361646,0.00006131667,0.00011614335,0.00001122899],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000045393717,0.00017148972,0.05839474,0.00013813582,0.00075296074,0.000031389376,0.08169487,0.00048046024,0.0016101908,0.21917608,0.63665324,0.0008510406],"study_design_scores_gemma":[0.0022712238,0.0002855159,0.42983416,0.00044698865,0.0005994144,0.00024115508,0.3454281,0.0006199669,0.00056912145,0.051143695,0.1677676,0.00079307234],"about_ca_topic_score_codex":0.00011101503,"about_ca_topic_score_gemma":0.00021653624,"teacher_disagreement_score":0.46888563,"about_ca_system_score_codex":0.000106063926,"about_ca_system_score_gemma":0.00020694187,"threshold_uncertainty_score":0.24314636},"labels":[],"label_agreement":null},{"id":"W4382337864","doi":"10.22148/001c.74068","title":"Quantifying the Gap: The Gender Gap in French Writers’ Wikidata","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Wikis in Education and Collaboration","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Gender gap; Representation (politics); Information gap; Diversity (politics); Computer science; Sociology; Political science; Anthropology","score_opus":0.26256984528829574,"score_gpt":0.43209060713089065,"score_spread":0.1695207618425949,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382337864","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8921962,0.00046112924,0.00007848542,0.09329965,0.0017197876,0.00015005686,0.0000039777237,0.000019779915,0.012070937],"genre_scores_gemma":[0.99577975,0.0005949985,0.00007393504,0.0005657133,0.0007350423,0.0000011500229,0.000002683895,0.0000030144474,0.0022436983],"study_design_codex":"not_applicable","study_design_gemma":"qualitative","domain_scores_codex":[0.99874675,0.00022537343,0.00030739422,0.00006078856,0.0004969852,0.00016267966],"domain_scores_gemma":[0.9990643,0.00017741827,0.0002369844,0.000110209294,0.00036270826,0.000048363894],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016994118,0.000057362384,0.00009605698,0.00007315065,0.00042265657,0.0002577287,0.0004238268,0.00004293849,0.00010116403],"category_scores_gemma":[0.0005740244,0.000026967937,0.00006991874,0.0013424596,0.00013174402,0.00035085162,0.000026566313,0.00023585626,0.000034598517],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019672581,0.00011392376,0.049563527,0.000016848302,0.00014127926,0.000027363287,0.33938757,0.0031680404,0.0005059608,0.07527425,0.5256871,0.006094505],"study_design_scores_gemma":[0.0003252968,0.000034505443,0.11536164,0.00005677413,0.0000839783,0.000014276189,0.4790056,0.0024783048,0.00003928604,0.00891591,0.39350697,0.00017743971],"about_ca_topic_score_codex":0.00019512672,"about_ca_topic_score_gemma":0.002356224,"teacher_disagreement_score":0.13961807,"about_ca_system_score_codex":0.00009057,"about_ca_system_score_gemma":0.00022484738,"threshold_uncertainty_score":0.32507753},"labels":[],"label_agreement":null},{"id":"W4382598314","doi":"10.22148/001c.74598","title":"Preface: World Literature in an Expanding Digital Space","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Translation Studies and Practices","field":"Arts and Humanities","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Deutsche Forschungsgemeinschaft","keywords":"Encyclopedia; Multilingualism; Data science; Space (punctuation); Computer science; Sociology; Library science","score_opus":0.0832909468269947,"score_gpt":0.32988617093724826,"score_spread":0.24659522411025356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382598314","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9478276,0.0018087316,0.000010852502,0.006474125,0.00079945056,0.000070208705,0.000040226827,0.00004169718,0.042927135],"genre_scores_gemma":[0.9882449,0.00034960243,0.000036418613,0.00006279351,0.00077991287,3.2683877e-7,0.000009643874,0.000005826566,0.010510564],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.99929607,0.000021079219,0.00028776296,0.000063484535,0.00020216264,0.000129444],"domain_scores_gemma":[0.99941105,0.00007455408,0.0002041703,0.000048475635,0.0002093409,0.000052395088],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00015303552,0.00008661758,0.00017111245,0.00019794783,0.00012332237,0.0008660601,0.00009275842,0.000019292653,0.0001283617],"category_scores_gemma":[0.000023175007,0.00005214464,0.000091878166,0.00023594373,0.000042148306,0.002160954,0.000013576562,0.00022798977,0.000020614842],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00072608027,0.00046544368,0.05394762,0.00027386472,0.00083997595,0.00091851683,0.6819204,0.010341333,0.0004910193,0.080879346,0.11932255,0.04987383],"study_design_scores_gemma":[0.0004124774,0.00011921365,0.0024623703,0.00016072778,0.00004343534,0.000015516682,0.0276942,0.0005202974,0.0000140328,0.0011866294,0.96722674,0.00014436599],"about_ca_topic_score_codex":0.0000044228395,"about_ca_topic_score_gemma":0.00056227646,"teacher_disagreement_score":0.8479042,"about_ca_system_score_codex":0.000017466184,"about_ca_system_score_gemma":0.000010139014,"threshold_uncertainty_score":0.83514374},"labels":[],"label_agreement":null},{"id":"W4382629066","doi":"10.22148/001c.70251","title":"The Wikipedia Republic of Literary Characters","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Wikis in Education and Collaboration","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Encyclopedia; Narrative; Opposition (politics); Taxonomy (biology); Linguistics; Theme (computing); Computer science; Character (mathematics); Literature; World Wide Web; Art; Philosophy; Political science","score_opus":0.03925478040940007,"score_gpt":0.36423273205712353,"score_spread":0.3249779516477235,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382629066","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9573368,0.00022966499,0.000004848644,0.03618467,0.0013750356,0.000046481528,0.00000180396,0.000013017002,0.004807675],"genre_scores_gemma":[0.9869038,0.0014261926,0.000047134894,0.00015664827,0.00088582915,4.970473e-7,0.0000023669447,0.0000023686857,0.010575199],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990489,0.00009490326,0.00031766354,0.000035821315,0.0003958344,0.00010688707],"domain_scores_gemma":[0.9984504,0.00013847202,0.00040912846,0.00006529442,0.00087149907,0.00006520437],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00093902694,0.000037631926,0.00008937273,0.000067924244,0.00025364134,0.00013606287,0.00020741462,0.000033737662,0.000031391028],"category_scores_gemma":[0.0007624625,0.000021248947,0.000080586,0.00092696096,0.00012258408,0.00040032476,0.000010772916,0.00009797213,0.000010712355],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000097075594,0.00015086978,0.014531274,0.000027425578,0.0002724182,0.000016112557,0.19793138,0.00016324062,0.0014958779,0.17210679,0.53987604,0.07333146],"study_design_scores_gemma":[0.00018406771,0.000069304704,0.045649633,0.00003846272,0.000049139493,0.0000049177916,0.065485716,0.00014620811,0.00017844248,0.007836924,0.88026404,0.000093164825],"about_ca_topic_score_codex":0.000012340666,"about_ca_topic_score_gemma":0.000053963737,"teacher_disagreement_score":0.34038797,"about_ca_system_score_codex":0.000048376696,"about_ca_system_score_gemma":0.00026677843,"threshold_uncertainty_score":0.19508298},"labels":[],"label_agreement":null},{"id":"W4385719959","doi":"10.22148/001c.84475","title":"Examining the representation of landscape and its emotional value in German-Swiss fiction between 1840 and 1940","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Culinary Culture and Tourism","field":"Agricultural and Biological Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"German; Representation (politics); Space (punctuation); Sentiment analysis; Value (mathematics); Perspective (graphical); Valence (chemistry); Relation (database); Linguistics; Sociology; Psychology; Computer science; Artificial intelligence; Philosophy; Politics; Political science; Law","score_opus":0.07550319640186247,"score_gpt":0.2976491677200317,"score_spread":0.2221459713181692,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385719959","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9967487,0.00032264905,0.0000011043421,0.0026412543,0.00005277494,0.000048469392,0.000007617892,0.0000054243114,0.00017199498],"genre_scores_gemma":[0.9987686,0.0004961829,0.000015909052,0.00002372587,0.00042094136,3.88154e-7,0.000019401377,5.1075716e-7,0.00025429158],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992339,0.0000634877,0.00030810156,0.000086961096,0.00021148205,0.00009607156],"domain_scores_gemma":[0.9994043,0.00018516456,0.00022929799,0.000018594408,0.00011748984,0.000045123266],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00029666285,0.00007108889,0.00017426582,0.000026354384,0.0001005671,0.000037424757,0.00007894478,0.00005468617,0.000018165749],"category_scores_gemma":[0.00006893373,0.000022503333,0.00005160835,0.00043273278,0.00003370628,0.00021233915,0.000038770602,0.00014212058,0.0000013627784],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016824686,0.00013503032,0.78475064,0.0000834943,0.00033074434,0.00011898463,0.007814315,0.0028341778,0.1446273,0.0013571508,0.0191577,0.03862222],"study_design_scores_gemma":[0.00019113503,0.00016569567,0.99387854,0.00004945338,0.000046287416,0.00006311748,0.0018903338,0.0015254428,0.00052354985,0.00070663064,0.00089766877,0.000062160274],"about_ca_topic_score_codex":0.000022776889,"about_ca_topic_score_gemma":0.000021411597,"teacher_disagreement_score":0.2091279,"about_ca_system_score_codex":0.000007722674,"about_ca_system_score_gemma":0.0000040829195,"threshold_uncertainty_score":0.09176593},"labels":[],"label_agreement":null},{"id":"W4386250985","doi":"10.22148/001c.84860","title":"Algorithmic Maps and the Political Geography of Early-modern Japan","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Geographic Information Systems Studies","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Universitat Autònoma de Barcelona; Center for Spatial and Textual Analysis, Stanford University; Waseda University; International Research Center for Japanese Studies","keywords":"Politics; Contiguity; Historiography; Political geography; Rank (graph theory); Geography; State (computer science); Economic geography; Space (punctuation); Population; Cartography; Genealogy; History; Sociology; Political science; Demography; Law; Mathematics; Archaeology; Computer science; Combinatorics","score_opus":0.026960302746373403,"score_gpt":0.30473644063718186,"score_spread":0.27777613789080846,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386250985","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9795342,0.0003417338,0.000057821882,0.011092499,0.00030078972,0.00013334824,0.000009079869,0.000018528453,0.008512002],"genre_scores_gemma":[0.99892056,0.00038995262,0.000076933706,0.0000967203,0.00023482362,8.9084125e-7,6.4591654e-7,0.0000024384383,0.0002770464],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9985092,0.00010096378,0.00050394074,0.000044455606,0.00060149754,0.00023992202],"domain_scores_gemma":[0.9985911,0.00019672034,0.0003813565,0.000062801846,0.0006686688,0.00009930194],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011814833,0.00007195303,0.00026825425,0.00018348696,0.00031250893,0.00007231273,0.00018987509,0.000051248153,0.000005734805],"category_scores_gemma":[0.00023548378,0.000037538768,0.00021926922,0.00062679587,0.00085920026,0.00025532703,0.000044265,0.00014884472,0.000005752682],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000114090966,0.0000392167,0.26633912,0.00008760425,0.0010599892,0.000011578202,0.15963656,0.00009168796,0.000033152108,0.5632544,0.005994871,0.0033377153],"study_design_scores_gemma":[0.0035917102,0.00029937815,0.354029,0.0002255788,0.0005615936,0.00009385887,0.47949606,0.0014666261,0.000036365785,0.13606311,0.023732292,0.00040441434],"about_ca_topic_score_codex":0.0006203304,"about_ca_topic_score_gemma":0.00008948716,"teacher_disagreement_score":0.42719132,"about_ca_system_score_codex":0.000017527651,"about_ca_system_score_gemma":0.00003772892,"threshold_uncertainty_score":0.3165759},"labels":[],"label_agreement":null},{"id":"W4386925245","doi":"10.22148/001c.87560","title":"Understanding Peanuts and Schulzian Symmetry: Panel Detection, Caption Detection, and Gag Panels in 17,897 Comic Strips Through Distant Viewing.","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Advanced Image and Video Retrieval Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Comic strip; Comics; Newspaper; Thumbnail; Visual arts; Audience measurement; Computer science; Media studies; Art; Sociology; Advertising; Artificial intelligence; Image (mathematics)","score_opus":0.1672736653302643,"score_gpt":0.32001831376422835,"score_spread":0.15274464843396404,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386925245","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2062094,0.00059880805,0.79193497,0.00076181703,0.0002222777,0.0001114011,0.0000023966966,0.00008878569,0.00007011716],"genre_scores_gemma":[0.9937961,0.0033290666,0.0025955678,0.000084354135,0.00009626191,9.069171e-7,9.0059194e-7,0.00000798387,0.00008882243],"study_design_codex":"bench_or_experimental","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99862117,0.00009465379,0.0005174589,0.00021688061,0.00030354975,0.00024629684],"domain_scores_gemma":[0.9991099,0.00012477906,0.0003572004,0.00013543916,0.00016940749,0.00010327972],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000594076,0.00016685101,0.00030453617,0.00026486526,0.00024307262,0.00025121952,0.00020427923,0.00008046185,0.0000015517085],"category_scores_gemma":[0.00016761752,0.00012452542,0.00007494088,0.0010182859,0.000086462234,0.001549833,0.00010694529,0.00042962204,0.0000011553064],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009389412,0.0005274078,0.01941327,0.0011527766,0.0011887217,0.0024230133,0.05476857,0.0037399668,0.56239563,0.03052539,0.0051069693,0.31781936],"study_design_scores_gemma":[0.006781271,0.0036245775,0.13581178,0.0022353528,0.0004436155,0.0043629403,0.03564377,0.18590221,0.17623185,0.42345113,0.022704829,0.0028066777],"about_ca_topic_score_codex":0.000019155934,"about_ca_topic_score_gemma":0.00002861267,"teacher_disagreement_score":0.7893394,"about_ca_system_score_codex":0.00027065582,"about_ca_system_score_gemma":0.000024136281,"threshold_uncertainty_score":0.5077999},"labels":[],"label_agreement":null},{"id":"W4387706663","doi":"10.22148/001c.88113","title":"Operationalizing Canonicity: A Quantitative Study of French 19th and 20th Century Literature","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Agence Nationale de la Recherche","keywords":"Operationalization; Grammar; Sociology; Norm (philosophy); Politics; Literary criticism; Cultural transmission in animals; Sociocultural evolution; Linguistics; Epistemology; Philosophy; Law; Political science; Anthropology","score_opus":0.08131779061149383,"score_gpt":0.42113231399091483,"score_spread":0.339814523379421,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387706663","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99694484,0.0009201833,0.00017193021,0.0009996205,0.00014409013,0.00007239414,0.0000061585774,0.0000072504968,0.0007335044],"genre_scores_gemma":[0.99585503,0.0006481355,0.0022112383,0.00003885163,0.0001906655,4.0822863e-7,0.0000053859285,0.0000031846037,0.0010470779],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9984577,0.00030552002,0.00041715344,0.0000940797,0.0006036469,0.00012191147],"domain_scores_gemma":[0.998137,0.00031612246,0.000346108,0.000043299402,0.0010663692,0.00009110182],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010800761,0.00008141985,0.0002796114,0.00020590561,0.00023230913,0.00012488481,0.00013709598,0.00004064823,0.000033656594],"category_scores_gemma":[0.0004773877,0.00005413534,0.00011122546,0.0012535373,0.0000852544,0.0003158152,0.00003559888,0.00016528233,0.0000018005933],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001333718,0.0010060584,0.08071912,0.000103210754,0.0026081589,0.00027307347,0.76120293,0.033840112,0.0018782677,0.09583929,0.010173178,0.012223248],"study_design_scores_gemma":[0.0032430328,0.0024100398,0.30612355,0.00060935796,0.001507719,0.000064226486,0.5902189,0.030346042,0.000085675136,0.023131581,0.041447192,0.0008126604],"about_ca_topic_score_codex":0.0004631082,"about_ca_topic_score_gemma":0.001063548,"teacher_disagreement_score":0.22540443,"about_ca_system_score_codex":0.000042769836,"about_ca_system_score_gemma":0.00014875368,"threshold_uncertainty_score":0.22075751},"labels":[],"label_agreement":null},{"id":"W4388698876","doi":"10.22148/001c.87937","title":"Time Maps: Theory and Method","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Narrative Theory and Analysis","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Computer science; Schema (genetic algorithms); Chronotope; Narrative network; Heuristic; Narrative inquiry; Artificial intelligence; Art; Information retrieval; Literature; Narrative criticism","score_opus":0.04047170319887512,"score_gpt":0.3014801139038926,"score_spread":0.2610084107050175,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388698876","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9101396,0.001164266,0.00081945,0.0061133127,0.0006340004,0.0001070984,0.0000648445,0.00010706875,0.08085034],"genre_scores_gemma":[0.76132727,0.00031973742,0.00061229436,0.000561822,0.001998153,6.2577533e-7,0.000013725135,0.000018198269,0.23514819],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9991867,0.00020109916,0.00027590976,0.000062908475,0.0001597312,0.00011367124],"domain_scores_gemma":[0.9991817,0.00023431485,0.00020993211,0.000054336782,0.00025327725,0.00006646562],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0012512442,0.00008980849,0.0002490812,0.0001408831,0.00018001892,0.000118803284,0.00010373884,0.000021574187,0.0013945827],"category_scores_gemma":[0.00013085458,0.000050420338,0.00016797964,0.00010711948,0.00017690346,0.00027829802,0.000025217549,0.00013683597,0.00012853021],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013199325,0.00003667371,0.000029886345,0.000024251976,0.0007730319,0.0000790273,0.050710883,0.00030191298,0.0016774365,0.8934765,0.04810797,0.0046504717],"study_design_scores_gemma":[0.0007841297,0.00038561018,0.00010646098,0.0001151589,0.0010641785,0.0001268095,0.119962305,0.003019574,0.00067820115,0.4491281,0.4242191,0.0004103907],"about_ca_topic_score_codex":0.0000015724269,"about_ca_topic_score_gemma":0.0000053346544,"teacher_disagreement_score":0.44434837,"about_ca_system_score_codex":0.000011303587,"about_ca_system_score_gemma":0.000011426147,"threshold_uncertainty_score":0.9995183},"labels":[],"label_agreement":null},{"id":"W4391810377","doi":"10.22148/001c.91289","title":"\"Authentic and Amazing\": authenticity as an evaluative category in online consumer restaurant reviews","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Sport and Mega-Event Impacts","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Advertising; Psychology; Aesthetics; Art; Business","score_opus":0.10147950542829795,"score_gpt":0.43019515091578675,"score_spread":0.3287156454874888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391810377","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9879381,0.00883875,0.000012531282,0.0019158337,0.00042154707,0.00019112973,0.0000037972013,0.0000106991165,0.00066757726],"genre_scores_gemma":[0.9903649,0.0064034397,0.00015001725,0.00014935246,0.0003046372,8.638035e-7,0.0000043769933,0.0000067284045,0.0026156928],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983359,0.00024939812,0.00056888716,0.00013041162,0.0004743224,0.00024105993],"domain_scores_gemma":[0.9990715,0.00010191152,0.00025139464,0.000077564415,0.00024222283,0.00025545474],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021721735,0.00012540256,0.00035852136,0.00016119283,0.00011752623,0.00017969952,0.00015746096,0.00007431794,0.00016630338],"category_scores_gemma":[0.0008912413,0.00007787789,0.00013552657,0.00039719636,0.00020979527,0.0006534776,0.000019326575,0.00033608216,0.000019576642],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004901888,0.0016208629,0.02764764,0.00067591,0.0009273973,0.002265525,0.716862,0.00017809596,0.0053236107,0.07721372,0.026878752,0.13991629],"study_design_scores_gemma":[0.0014731649,0.0007760299,0.028277762,0.002544497,0.0014236306,0.00029291422,0.072461404,0.003957676,0.00018371842,0.027507706,0.86036223,0.0007392971],"about_ca_topic_score_codex":0.0005607646,"about_ca_topic_score_gemma":0.0026562032,"teacher_disagreement_score":0.83348346,"about_ca_system_score_codex":0.00015087482,"about_ca_system_score_gemma":0.0003244767,"threshold_uncertainty_score":0.31757683},"labels":[],"label_agreement":null},{"id":"W4392387227","doi":"10.22148/001c.92531","title":"Validation of the Story World Absorption Scale through annotation of online book reviews","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Media Influence and Health","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Annotation; Scale (ratio); Absorption (acoustics); Computer science; Data science; World Wide Web; Information retrieval; Artificial intelligence; Geography; Cartography; Physics; Optics","score_opus":0.10285769558743348,"score_gpt":0.342681603623478,"score_spread":0.23982390803604453,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392387227","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97289073,0.015696518,0.00007584091,0.0060593802,0.0016637945,0.00021742671,0.000039038514,0.000008710918,0.003348545],"genre_scores_gemma":[0.9736967,0.006678548,0.00045829508,0.0012851366,0.00230204,0.000001428758,0.000019265588,0.000011242772,0.015547307],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99894136,0.00006859419,0.00061607757,0.000047286896,0.00026023755,0.00006645163],"domain_scores_gemma":[0.99888027,0.000044972963,0.00058730005,0.00007465063,0.00038935585,0.00002343221],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003801824,0.000065529646,0.00022355557,0.000070307426,0.00006687194,0.000032866414,0.000101262616,0.000020066129,0.0003468472],"category_scores_gemma":[0.000055458466,0.000032504988,0.00018784672,0.00009621184,0.00010959768,0.0005506279,0.000008752414,0.00020092019,0.000007864962],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014693402,0.0005004824,0.0018141798,0.004294158,0.00038786756,0.000008155976,0.27551073,0.0021682077,0.014832028,0.10991113,0.5580377,0.032388434],"study_design_scores_gemma":[0.00015964074,0.00015698039,0.0011565129,0.0015996114,0.00025176266,0.000007366204,0.003969531,0.000370379,0.0019881716,0.0018738791,0.98838675,0.00007940456],"about_ca_topic_score_codex":0.000019243043,"about_ca_topic_score_gemma":0.00018233049,"teacher_disagreement_score":0.43034905,"about_ca_system_score_codex":0.000045139666,"about_ca_system_score_gemma":0.00006917188,"threshold_uncertainty_score":0.37977335},"labels":[],"label_agreement":null},{"id":"W4393084754","doi":"10.22148/001c.92535","title":"Topic models indicate textual aboutness and pragmatics: Valuation practices in Islamophobic discourse","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Pragmatics; Valuation (finance); Linguistics; Sociology; Computer science; Political science; Business; Philosophy; Accounting","score_opus":0.12967615831363632,"score_gpt":0.46276630122179513,"score_spread":0.3330901429081588,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393084754","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97292584,0.0025283021,0.011827238,0.008944092,0.00028396308,0.000092117705,0.0000020494792,0.000013223968,0.0033831554],"genre_scores_gemma":[0.9942654,0.0008856787,0.0032542504,0.000050501265,0.00041702276,9.140536e-7,0.0000021390917,0.0000039484735,0.0011201763],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99841815,0.00031098136,0.0004564758,0.0001115019,0.000566852,0.00013605121],"domain_scores_gemma":[0.9986612,0.00045362843,0.0004949858,0.000043654127,0.00026693247,0.00007960486],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021363222,0.000085086176,0.00021775265,0.0002310951,0.00012557849,0.00031281306,0.0001432196,0.00005909199,0.000045154622],"category_scores_gemma":[0.0005616756,0.000056555895,0.00010366524,0.00058389734,0.00012889417,0.0011463964,0.00002440526,0.00024313171,0.000003834739],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008059273,0.00036100898,0.008393591,0.00025859696,0.00090647873,0.00028280015,0.19298242,0.0817836,0.00044280567,0.46382475,0.0015883065,0.24909504],"study_design_scores_gemma":[0.00055917015,0.00013307769,0.011726847,0.00042503452,0.0008301075,0.00007331301,0.06923393,0.51271695,0.000032375377,0.39572248,0.008192368,0.00035432618],"about_ca_topic_score_codex":0.00020489171,"about_ca_topic_score_gemma":0.00062415213,"teacher_disagreement_score":0.43093336,"about_ca_system_score_codex":0.000095965544,"about_ca_system_score_gemma":0.0002460178,"threshold_uncertainty_score":0.30164635},"labels":[],"label_agreement":null},{"id":"W4394612139","doi":"10.22148/001c.115371","title":"What We Didn’t Know a Recipe Could Be: Political Commentary, Machine Learning Models, and the Fluidity of Form in Nineteenth-Century Newspaper Recipes","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Recipe; Newspaper; Politics; Artificial intelligence; History; Sociology; Computer science; Law; Media studies; Political science; Ancient history","score_opus":0.08343540009799033,"score_gpt":0.2788673364551353,"score_spread":0.19543193635714495,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394612139","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7254777,0.11766071,0.000022713977,0.11627608,0.0026475168,0.00038914065,0.0001362175,0.000047537724,0.037342403],"genre_scores_gemma":[0.9752772,0.018986674,0.000021459338,0.0010382305,0.0005916878,0.0000011781998,0.000016150692,0.000013732032,0.004053678],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99849856,0.00009363768,0.0006502488,0.00011351048,0.00038106577,0.00026298314],"domain_scores_gemma":[0.9991415,0.0002891854,0.0001696896,0.000087024884,0.00020467254,0.00010794301],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0005481668,0.00017928176,0.00040738742,0.000114771145,0.00016810208,0.0012135371,0.00020488034,0.00005045913,0.00026629044],"category_scores_gemma":[0.00006327397,0.00009275311,0.00023748487,0.000058329017,0.0004547073,0.0022594417,0.00009170315,0.0007125134,0.0000013691521],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043937686,0.00016283614,0.0002520214,0.00022080795,0.00037813137,0.00006849067,0.0600639,0.00023778756,0.000019718835,0.90673804,0.023454212,0.007964685],"study_design_scores_gemma":[0.00141598,0.00042521316,0.00005598716,0.0011675833,0.00028598274,0.00009605674,0.07221102,0.021885522,0.000021816819,0.038949832,0.86322975,0.00025527226],"about_ca_topic_score_codex":0.00021274219,"about_ca_topic_score_gemma":0.0010880595,"teacher_disagreement_score":0.8677882,"about_ca_system_score_codex":0.00008968394,"about_ca_system_score_gemma":0.000027397888,"threshold_uncertainty_score":0.9998233},"labels":[],"label_agreement":null},{"id":"W4396567945","doi":"10.22148/001c.115481","title":"Vectors of Violence: Legitimation and Distribution of State Power in the _People’s Liberation Army Daily_ (_Jiefangjun Bao_), 1956-1989","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Chinese history and philosophy","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Legitimation; Memoir; Power (physics); State (computer science); Politics; Representation (politics); China; Reading (process); Political science; Extension (predicate logic); Sociology; History; Law","score_opus":0.019711204644525283,"score_gpt":0.3001765817800995,"score_spread":0.2804653771355742,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396567945","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99592316,0.0008808454,0.00017984779,0.0017388812,0.00027444938,0.00007239237,0.000025633035,0.000003793745,0.00090096693],"genre_scores_gemma":[0.99928564,0.00039398522,0.00003449446,0.000028664706,0.00018473943,4.377223e-7,0.000011306477,0.0000022391323,0.000058498354],"study_design_codex":"qualitative","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99895144,0.00014194124,0.00039601608,0.000061337174,0.00037724,0.00007204538],"domain_scores_gemma":[0.9993222,0.00011484512,0.00024941773,0.00004812543,0.00022982564,0.000035597983],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00078718783,0.00006637887,0.00015978907,0.0000718057,0.000094021314,0.000060600654,0.000110694426,0.000044181317,0.000016612905],"category_scores_gemma":[0.00015323304,0.00003938345,0.00009126051,0.00040751864,0.00015575273,0.000488721,0.000008751083,0.00016695683,9.730757e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005699299,0.000751037,0.024005307,0.00081703794,0.0004524706,0.00012540804,0.58928096,0.010009717,0.010778075,0.3322918,0.017127916,0.013790335],"study_design_scores_gemma":[0.0027144218,0.002215078,0.23498507,0.005221212,0.0010664221,0.00017659133,0.04111217,0.013027709,0.0032833142,0.6085807,0.08635679,0.0012604992],"about_ca_topic_score_codex":0.00015297942,"about_ca_topic_score_gemma":0.00020341293,"teacher_disagreement_score":0.5481688,"about_ca_system_score_codex":0.00008181371,"about_ca_system_score_gemma":0.00009627976,"threshold_uncertainty_score":0.16060103},"labels":[],"label_agreement":null},{"id":"W4396687966","doi":"10.22148/001c.116142","title":"Gender (im)balance in the Russian cinema: on the screen and behind the camera","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Cinema and Media Studies","field":"Economics, Econometrics and Finance","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Movie theater; Attendance; Narrative; Character (mathematics); Representation (politics); Prejudice (legal term); Test (biology); Psychology; Stereotype (UML); Film industry; Sociology; Gender studies; Social psychology; Art; Political science; Literature; Law; Mathematics","score_opus":0.0632649657141904,"score_gpt":0.26300683281109904,"score_spread":0.19974186709690867,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396687966","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.730109,0.032482978,0.00007936685,0.20998374,0.0006807052,0.0001953655,0.000038399772,0.0000065490417,0.026423924],"genre_scores_gemma":[0.99411887,0.0029892356,0.00002791632,0.0014577257,0.00049389503,0.0000017648374,7.0091494e-7,0.00000532754,0.0009045398],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992594,0.000022477694,0.00039868258,0.00009917156,0.000078078105,0.0001421732],"domain_scores_gemma":[0.99941564,0.000232407,0.00017759252,0.00012299378,0.00002205179,0.000029347684],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007523905,0.00010079489,0.00022648714,0.0000709121,0.00013503533,0.00015994831,0.00023064896,0.000031462965,0.00004340007],"category_scores_gemma":[0.0001242997,0.000037853442,0.00011898537,0.00019348392,0.00009803546,0.00010916959,0.000028498624,0.00039467355,0.000024175963],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000054552216,0.00010369093,0.04002546,0.0000749607,0.00070616265,0.00017825625,0.043197628,0.00047503455,0.000020163876,0.76920193,0.14373055,0.0022316193],"study_design_scores_gemma":[0.0009990847,0.00037584387,0.39631054,0.00027053693,0.00019876257,0.00045173397,0.02387409,0.009116513,0.000028126573,0.06458077,0.50336975,0.0004242386],"about_ca_topic_score_codex":0.000039878058,"about_ca_topic_score_gemma":0.00006872301,"teacher_disagreement_score":0.70462114,"about_ca_system_score_codex":0.00003124403,"about_ca_system_score_gemma":0.000013607757,"threshold_uncertainty_score":0.17146821},"labels":[],"label_agreement":null},{"id":"W4399118105","doi":"10.22148/001c.116905","title":"The World of Shakespeare and Company: An Introduction","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Shakespeare, Adaptation, and Literary Criticism","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Business; Management; Economics","score_opus":0.042666983910807886,"score_gpt":0.28038459366443785,"score_spread":0.23771760975362996,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399118105","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9444396,0.012320176,0.00006074835,0.022765364,0.0035212436,0.00011510231,0.000037538,0.000041396426,0.016698837],"genre_scores_gemma":[0.97770154,0.00014604349,0.000075597505,0.00006387891,0.0033415563,2.0728807e-7,0.000005305949,0.0000064728883,0.018659376],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992504,0.000039697006,0.00035926563,0.000067301786,0.00019555441,0.00008775013],"domain_scores_gemma":[0.9992807,0.00008324644,0.00011354411,0.00006780469,0.00040268127,0.000052020987],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023775073,0.00007567996,0.00014160924,0.000091635775,0.0001855396,0.00046931146,0.000093175724,0.000013218284,0.00028312576],"category_scores_gemma":[0.000031870928,0.000038938255,0.00008301678,0.00006807696,0.00020567067,0.00064717326,0.000008186738,0.00015626781,0.0000026698308],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000054622928,0.000040072548,0.000096422686,0.00008123447,0.00018343997,0.000017128248,0.06535382,0.000095866926,0.00020330976,0.8802716,0.04571957,0.007882926],"study_design_scores_gemma":[0.00010453213,0.00014727958,0.00048594733,0.00006697012,0.00011849614,0.000060109985,0.021302644,0.0021277377,0.000068272275,0.0032456429,0.9722017,0.00007067967],"about_ca_topic_score_codex":0.000011831468,"about_ca_topic_score_gemma":0.00041353478,"teacher_disagreement_score":0.92648214,"about_ca_system_score_codex":0.000012615864,"about_ca_system_score_gemma":0.000023326767,"threshold_uncertainty_score":0.45255813},"labels":[],"label_agreement":null},{"id":"W4399118211","doi":"10.22148/001c.116922","title":"Lending Books on the Left and Right Banks: Borrowing Practices at the American Library in Paris and Shakespeare and Company","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Historical Art and Culture Studies","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"New Left; Left and right; History; Business; Management; Political science; Economic history; Economics; Law; Engineering; Politics","score_opus":0.0413849591910085,"score_gpt":0.26820668807450876,"score_spread":0.22682172888350027,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399118211","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90832686,0.019964468,5.1301095e-7,0.037453193,0.00021892566,0.00006601048,0.000008318524,0.000014629441,0.033947095],"genre_scores_gemma":[0.9873663,0.0027195963,0.000011495507,0.0005793126,0.00049013353,4.990471e-7,7.23513e-7,0.0000064943274,0.008825451],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99934983,0.00005890106,0.00022116599,0.00009737625,0.00015819613,0.00011454296],"domain_scores_gemma":[0.99935263,0.00030349486,0.00021953766,0.000044055083,0.00003557495,0.000044705153],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00016505866,0.00011315198,0.0002097139,0.000042049018,0.0005439804,0.00052213675,0.00007159287,0.000013593237,0.00018214503],"category_scores_gemma":[0.000048431408,0.00004276305,0.000056137713,0.0000436132,0.0004057754,0.00046935098,0.00006567609,0.00031950782,0.0000029853643],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022751528,0.00008044424,0.0135964835,0.00020276596,0.0008767056,0.000330815,0.39234903,0.000018999002,0.0003563157,0.17560697,0.40695977,0.009394194],"study_design_scores_gemma":[0.00007299147,0.00012696057,0.002262993,0.00015182726,0.000109579814,0.00006724908,0.028036317,0.00022984507,0.000022692146,0.00048709183,0.968341,0.000091465015],"about_ca_topic_score_codex":0.000030595256,"about_ca_topic_score_gemma":0.00054612634,"teacher_disagreement_score":0.5613812,"about_ca_system_score_codex":0.000021590173,"about_ca_system_score_gemma":0.000009093585,"threshold_uncertainty_score":0.5034977},"labels":[],"label_agreement":null},{"id":"W4399118217","doi":"10.22148/001c.117123","title":"Black Internationalism and Shakespeare and Company","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Shakespeare, Adaptation, and Literary Criticism","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Internationalism (politics); Management; Political science; Art; Economic history; History; Economics; Law; Politics","score_opus":0.05067961012807186,"score_gpt":0.27761187415690725,"score_spread":0.22693226402883537,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399118217","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96715814,0.008137247,0.00008013103,0.0047443868,0.0008549981,0.00004981237,0.00004238366,0.000026497772,0.0189064],"genre_scores_gemma":[0.9854612,0.00039989833,0.00010499124,0.00028994525,0.0012320159,1.6981119e-7,0.0000062936865,0.0000072526127,0.012498246],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993316,0.000019656429,0.0002899789,0.00007977244,0.00019382045,0.00008516457],"domain_scores_gemma":[0.9994522,0.00007382391,0.0000695108,0.000033982335,0.00029621655,0.00007431717],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00012129502,0.000091579,0.00015011516,0.000103070604,0.000087260574,0.00070613937,0.00006596801,0.000021375967,0.00045761],"category_scores_gemma":[0.00002869267,0.000058749756,0.00006911643,0.000029994797,0.00020491723,0.00062722777,0.000015431477,0.00015087683,0.000010263539],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022889033,0.000027525877,0.00028434148,0.00013717855,0.00025206717,0.00012782041,0.159155,0.00003128989,0.000068688336,0.81065345,0.025539007,0.0037007725],"study_design_scores_gemma":[0.00030310056,0.00012722462,0.0015397246,0.00031239717,0.00016443816,0.00028935235,0.025993576,0.0078626275,0.00001236164,0.0106354775,0.9525961,0.00016358867],"about_ca_topic_score_codex":0.000020658377,"about_ca_topic_score_gemma":0.000077745404,"teacher_disagreement_score":0.92705715,"about_ca_system_score_codex":0.000013429924,"about_ca_system_score_gemma":0.000018910627,"threshold_uncertainty_score":0.6809318},"labels":[],"label_agreement":null},{"id":"W4399120936","doi":"10.22148/001c.116926","title":"Missing Data, Speculative Reading","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Human Mobility and Location-Based Analysis","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Reading (process); Missing data; Computer science; Linguistics; Philosophy","score_opus":0.10435306227255364,"score_gpt":0.40938981243681555,"score_spread":0.3050367501642619,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399120936","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.56974,0.015696667,0.123226844,0.12476836,0.0045875846,0.0005065933,0.00010474422,0.00032613293,0.16104306],"genre_scores_gemma":[0.9944641,0.00021865437,0.00075145444,0.0000924102,0.001163177,8.637517e-8,0.000009273985,0.000004104434,0.003296728],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.998816,0.00012635865,0.0003548204,0.00011821287,0.00044445603,0.00014011329],"domain_scores_gemma":[0.9990718,0.00019374058,0.00014798943,0.00013538315,0.00033010202,0.0001209872],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013481397,0.00006831146,0.00017278107,0.0001128231,0.00029847154,0.00038057787,0.00034419322,0.000047621437,0.00037767878],"category_scores_gemma":[0.00057574897,0.000047007572,0.00014960454,0.0005113038,0.00015822881,0.0007313657,0.000026810236,0.00022380389,0.000029841989],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009248295,0.0003978009,0.00748047,0.0003626348,0.003747966,0.00092497753,0.2656345,0.013180192,0.0031753217,0.1339441,0.14596662,0.42509294],"study_design_scores_gemma":[0.00024215893,0.00009127898,0.0015005755,0.00075063854,0.0014016886,0.000038175138,0.06977383,0.091980286,0.00022640997,0.018638713,0.8149202,0.00043607186],"about_ca_topic_score_codex":0.000399672,"about_ca_topic_score_gemma":0.0008197741,"teacher_disagreement_score":0.66895354,"about_ca_system_score_codex":0.0001742371,"about_ca_system_score_gemma":0.00028576484,"threshold_uncertainty_score":0.41353178},"labels":[],"label_agreement":null},{"id":"W4399139472","doi":"10.22148/001c.116910","title":"Unpacking Reed Peggram’s Library","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Library Science and Administration","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Unpacking; Computer science; Linguistics; Philosophy","score_opus":0.05111554788119449,"score_gpt":0.34206987738666567,"score_spread":0.2909543295054712,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399139472","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.72144425,0.002624648,0.00036319616,0.09783074,0.003269383,0.000120932695,0.0000073019996,0.00017337539,0.1741662],"genre_scores_gemma":[0.98153776,0.00051962957,0.00071515515,0.00030524968,0.0014626095,1.0621575e-7,0.000001932993,0.000004257583,0.015453278],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990138,0.00005757008,0.00027078885,0.000074303774,0.00042901994,0.00015453529],"domain_scores_gemma":[0.99955916,0.00006710915,0.00011552076,0.000044088356,0.000072513845,0.00014159734],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00034352147,0.000060370603,0.00010766837,0.00008845087,0.00019001443,0.0008382499,0.00024968834,0.000048987407,0.00034231748],"category_scores_gemma":[0.00009735583,0.00003800389,0.00014542323,0.0006031628,0.00013252287,0.0042702793,0.000016538957,0.00017760324,0.00003181431],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000050324958,0.00011281637,0.011700075,0.00008445581,0.00022445641,0.0009045395,0.058219723,0.0003526192,0.0029488169,0.4935564,0.38988033,0.041965414],"study_design_scores_gemma":[0.00009118501,0.00013698533,0.001225817,0.00018512258,0.00006045338,0.00005074153,0.017897502,0.0009524272,0.00064163964,0.015251621,0.9633553,0.00015123235],"about_ca_topic_score_codex":0.000014335056,"about_ca_topic_score_gemma":0.000017396835,"teacher_disagreement_score":0.57347494,"about_ca_system_score_codex":0.000027685324,"about_ca_system_score_gemma":0.00041682573,"threshold_uncertainty_score":0.80832636},"labels":[],"label_agreement":null},{"id":"W4399140793","doi":"10.22148/001c.116913","title":"Sylvia Beach's Final Book","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"American and British Literature Analysis","field":"Arts and Humanities","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"History; Art","score_opus":0.029726383477319884,"score_gpt":0.2542185891902544,"score_spread":0.22449220571293454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399140793","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13127352,0.2627904,0.001378804,0.030573927,0.0077385115,0.00024909494,0.00020408236,0.00031234362,0.56547934],"genre_scores_gemma":[0.84486425,0.0006900206,0.00019035683,0.0015102588,0.0041257385,4.224306e-7,0.000015895615,0.000013993414,0.14858906],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990252,0.000024331275,0.0004101503,0.00009298842,0.00029493458,0.00015234097],"domain_scores_gemma":[0.99930245,0.00003776511,0.00014845202,0.00006789425,0.00035263374,0.00009082757],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00013077846,0.00012022183,0.0002732768,0.00014551458,0.00010990867,0.00073879125,0.0001565249,0.000029489247,0.0027094672],"category_scores_gemma":[0.000025551002,0.00007286554,0.00042644065,0.000106731015,0.00017582509,0.00068507396,0.000015175568,0.0003159696,0.00006807677],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020112078,0.00004971972,0.00002743002,0.000053342053,0.0006849568,0.000605519,0.011618649,0.000047244397,0.00010246853,0.053994633,0.92200154,0.010794396],"study_design_scores_gemma":[0.000083427076,0.00012798111,0.00002056202,0.00014649787,0.0003744082,0.00022306667,0.0029775111,0.0006645513,0.000028128437,0.000568532,0.9946654,0.00011997375],"about_ca_topic_score_codex":0.000017429647,"about_ca_topic_score_gemma":0.000024266887,"teacher_disagreement_score":0.71359074,"about_ca_system_score_codex":0.000044820204,"about_ca_system_score_gemma":0.00004900744,"threshold_uncertainty_score":0.9982022},"labels":[],"label_agreement":null},{"id":"W4399140855","doi":"10.22148/001c.116908","title":"Virginia Woolf’s Common Readers in Paris","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Autobiographical and Biographical Writing","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"History; Literature; Art; Psychoanalysis; Psychology","score_opus":0.033713958388853923,"score_gpt":0.26834544890270917,"score_spread":0.23463149051385523,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399140855","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95340836,0.0052779205,0.000033392942,0.018805118,0.0011261824,0.00008559558,0.000024451545,0.000071684604,0.021167316],"genre_scores_gemma":[0.9971359,0.0006161062,0.000058648682,0.0003653309,0.0008209135,4.3826822e-7,0.000003576828,0.000007791278,0.000991314],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99884355,0.000036707028,0.0005643861,0.00009834894,0.00025777964,0.00019922464],"domain_scores_gemma":[0.99950856,0.000095544856,0.00012409838,0.000055617977,0.00012503458,0.000091159396],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00026774078,0.00012461477,0.00027573528,0.00031632232,0.0000866557,0.0003534041,0.00015191753,0.000062874664,0.0003649569],"category_scores_gemma":[0.000023173328,0.00007000833,0.0003642004,0.00029691146,0.0002404322,0.0003442483,0.000022730595,0.0005439336,0.00001885348],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002755443,0.0007848736,0.027737306,0.00068536063,0.0016280896,0.002748657,0.082843095,0.00033947677,0.0011189856,0.618969,0.21928366,0.04358594],"study_design_scores_gemma":[0.00044896448,0.0004221428,0.0034007893,0.0009374616,0.0002636378,0.000113275615,0.028456824,0.0028327086,0.00007519237,0.01907112,0.9435664,0.0004114981],"about_ca_topic_score_codex":0.00010643065,"about_ca_topic_score_gemma":0.0004186197,"teacher_disagreement_score":0.72428274,"about_ca_system_score_codex":0.000024412298,"about_ca_system_score_gemma":0.000016924354,"threshold_uncertainty_score":0.39960223},"labels":[],"label_agreement":null},{"id":"W4399140907","doi":"10.22148/001c.116919","title":"The Afterlives of Shakespeare and Company in Online Social Readership","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Shakespeare, Adaptation, and Literary Criticism","field":"Arts and Humanities","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Audience measurement; Social media; Advertising; Art; Sociology; Media studies; Business; World Wide Web; Computer science","score_opus":0.08475425153316005,"score_gpt":0.29684889344107507,"score_spread":0.21209464190791502,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399140907","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.982346,0.00801276,0.0000056277318,0.0057073566,0.00039831307,0.000044114102,0.00004402765,0.000008418449,0.0034333772],"genre_scores_gemma":[0.9953092,0.0002473203,0.00004728214,0.00009371528,0.0009933697,2.1186426e-7,0.0000049579553,0.00000579099,0.003298161],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.99918425,0.000047078112,0.0004132944,0.000057621473,0.00018993292,0.000107844964],"domain_scores_gemma":[0.9994468,0.00015284208,0.00010617464,0.0000351329,0.0002263601,0.000032672906],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019408824,0.00008013727,0.00017896447,0.00008123141,0.00012349153,0.0002716152,0.00009895955,0.000024269099,0.00008906944],"category_scores_gemma":[0.00003832917,0.0000429314,0.000103159306,0.000056188077,0.0002560902,0.00032863594,0.000009944802,0.00019460371,0.0000013954685],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000140376,0.000118094045,0.00091180194,0.0002678162,0.0002678006,0.00011021932,0.60497963,0.000029851491,0.00013662374,0.37342113,0.009977815,0.009638813],"study_design_scores_gemma":[0.0006863292,0.0003484086,0.020750776,0.0007646391,0.00025108212,0.00010470742,0.4725919,0.0031627263,0.00003040379,0.011057464,0.48998532,0.00026624504],"about_ca_topic_score_codex":0.000026819667,"about_ca_topic_score_gemma":0.00065140903,"teacher_disagreement_score":0.4800075,"about_ca_system_score_codex":0.000014817788,"about_ca_system_score_gemma":0.000031579068,"threshold_uncertainty_score":0.26191914},"labels":[],"label_agreement":null},{"id":"W4399140931","doi":"10.22148/001c.116915","title":"A Counterfactual Canon","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Games and Media","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Counterfactual thinking; Canon; Psychology; Art; Literature; Social psychology","score_opus":0.03572198473454398,"score_gpt":0.34128535197763826,"score_spread":0.3055633672430943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399140931","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.422313,0.0056712623,0.0002496956,0.010143383,0.0038550354,0.00007645659,0.0000121999665,0.000050440118,0.5576285],"genre_scores_gemma":[0.9368829,0.0005636692,0.00007798896,0.00013753613,0.0010154272,1.00691345e-7,5.1788294e-7,0.0000034937234,0.061318338],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9992223,0.000017907449,0.00019515496,0.000045195913,0.00038342018,0.00013606976],"domain_scores_gemma":[0.99951863,0.000052380467,0.00006665609,0.000028109986,0.00019855596,0.00013567093],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00028168398,0.000050891224,0.00011310067,0.00004769337,0.00003240257,0.000299121,0.0001286163,0.00003671892,0.00015668814],"category_scores_gemma":[0.00015347333,0.00003065413,0.0001316572,0.00018279046,0.000094653,0.00044957327,0.000009596365,0.00013594606,0.000037101352],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053331114,0.0001076955,0.0005416419,0.000077239376,0.00048274416,0.00076676044,0.09983476,0.00024796432,0.0007835859,0.06321508,0.2812515,0.5526377],"study_design_scores_gemma":[0.00005130167,0.000056626362,0.00006977235,0.0000736539,0.000044738117,0.00002719377,0.0060274387,0.00009270444,0.000045252633,0.0004784221,0.9929826,0.00005029724],"about_ca_topic_score_codex":0.000028961571,"about_ca_topic_score_gemma":0.00021409657,"teacher_disagreement_score":0.7117311,"about_ca_system_score_codex":0.00010117083,"about_ca_system_score_gemma":0.00024939363,"threshold_uncertainty_score":0.28844306},"labels":[],"label_agreement":null},{"id":"W4399730290","doi":"10.22148/001c.116225","title":"A Topic Modelling of Muslim Religious Reform in the Colonial Age: A Computational and Digital Study of _al-Manār_ (1898-1935)","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Education and Islamic Studies","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Qatar National Research Fund; Fonds National de la Recherche Luxembourg","keywords":"Arabic; Islam; Colonialism; Order (exchange); Sociology; History; Political science; Law; Theology; Philosophy; Linguistics","score_opus":0.06530625548821499,"score_gpt":0.35803555031730283,"score_spread":0.2927292948290878,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399730290","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9950346,0.00043485532,0.0000635975,0.0024747497,0.00014348318,0.00008323274,0.0000020396794,0.0000025174652,0.0017609464],"genre_scores_gemma":[0.99917954,0.00033784943,0.00009435862,0.000036371013,0.00015210322,5.9641195e-7,6.1235374e-7,0.0000018491846,0.00019674005],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99905807,0.000049409216,0.00038316683,0.00005579418,0.00037407444,0.000079459416],"domain_scores_gemma":[0.99939185,0.00015490895,0.00016579858,0.0000331987,0.00023217798,0.000022047549],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00040902838,0.000053067328,0.00017513434,0.00008129677,0.00007526257,0.0000830981,0.00012079405,0.000025621657,0.0000030114982],"category_scores_gemma":[0.00008170787,0.00002974748,0.00006259015,0.00029255296,0.00013021774,0.00017910505,0.000017933491,0.0001329177,3.7063919e-7],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033290136,0.0003690729,0.0067793163,0.000050838386,0.00020734734,0.000043787386,0.96846783,0.012509929,0.0000070527867,0.008864947,0.00132083,0.0013457766],"study_design_scores_gemma":[0.00074879255,0.00066700095,0.011644216,0.00019952453,0.00016496537,0.00003083771,0.9525992,0.0162928,0.000002929254,0.014949665,0.002569416,0.00013065149],"about_ca_topic_score_codex":0.0004679198,"about_ca_topic_score_gemma":0.0005632305,"teacher_disagreement_score":0.015868612,"about_ca_system_score_codex":0.00008355443,"about_ca_system_score_gemma":0.00010105574,"threshold_uncertainty_score":0.121306695},"labels":[],"label_agreement":null},{"id":"W4399732070","doi":"10.22148/001c.116223","title":"Racio-national Imaginary and Discursive Formation of Arabo-Islamic Identity in al-Manār and al-Risālah: A Topic Modeling Study","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Qatar National Research Fund; Fonds National de la Recherche Luxembourg; Qatar Foundation","keywords":"Conceptualization; Intelligentsia; Islam; The Imaginary; Modernity; Identity (music); Politics; Sociology; Gender studies; National identity; Religious studies; Media studies; Linguistics; Political science; Law; Psychoanalysis; Philosophy; Theology; Aesthetics; Psychology","score_opus":0.05478203777936157,"score_gpt":0.4321214697361947,"score_spread":0.3773394319568331,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399732070","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98259836,0.0011299537,0.011753357,0.0040166965,0.00009961576,0.00007788973,0.000002508253,0.0000040991204,0.0003175313],"genre_scores_gemma":[0.9980551,0.00042242324,0.0012548936,0.00005676046,0.00009490801,6.529112e-7,0.000001865505,0.0000022144238,0.000111193614],"study_design_codex":"qualitative","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99866515,0.00016820908,0.00043745016,0.000091472684,0.0005493592,0.00008832725],"domain_scores_gemma":[0.9991751,0.00014597835,0.00017011852,0.00002606532,0.00044357256,0.000039218856],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013025461,0.00007032733,0.0002125008,0.00023134841,0.00010586489,0.00013996087,0.000086349195,0.000023955427,0.000010301724],"category_scores_gemma":[0.00019248463,0.000049830713,0.000083793435,0.0004482828,0.000076854136,0.0013705429,0.000035007095,0.00015626969,5.4535735e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028331953,0.0014495836,0.10034247,0.000508773,0.0017827214,0.0003867402,0.4311638,0.18342933,0.0030368338,0.18990916,0.00281667,0.08489063],"study_design_scores_gemma":[0.00048760383,0.0001190309,0.02166399,0.00014903,0.00023871854,0.00003833049,0.033482444,0.7694843,0.000009159143,0.17379434,0.00038854443,0.00014450702],"about_ca_topic_score_codex":0.00018103239,"about_ca_topic_score_gemma":0.00073136674,"teacher_disagreement_score":0.586055,"about_ca_system_score_codex":0.000091339476,"about_ca_system_score_gemma":0.000080793965,"threshold_uncertainty_score":0.20320375},"labels":[],"label_agreement":null},{"id":"W4399732979","doi":"10.22148/001c.116818","title":"The Potential and Limits of Arabic Digital Humanities","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Education and Islamic Studies","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Ephemera; Data science; Metadata; Newspaper; Subject (documents); Field (mathematics); Analytics; Computer science; Sociology; World Wide Web; Media studies; Visual arts; Art","score_opus":0.04905456536908647,"score_gpt":0.33601463540090143,"score_spread":0.28696007003181495,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399732979","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9478662,0.009383167,0.000023100205,0.012439254,0.0011645514,0.00003471781,0.0000033885233,0.000008296588,0.029077293],"genre_scores_gemma":[0.9848656,0.004018214,0.0000149874095,0.00003081101,0.00047809954,1.08785976e-7,1.7250358e-7,0.0000015895072,0.010590437],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99946725,0.000021033342,0.00019017459,0.000030327632,0.00021693866,0.000074258205],"domain_scores_gemma":[0.9994827,0.000112967726,0.00009493757,0.000022172308,0.00026456235,0.000022656435],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002091854,0.000037370795,0.00008460086,0.0000299328,0.00023734209,0.00028535613,0.00008609318,0.000017554778,0.000014012016],"category_scores_gemma":[0.00019669428,0.000018619017,0.0000748867,0.00009801896,0.00032358235,0.0002757252,0.000010863208,0.00008266487,0.000002167196],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000049388334,0.00011556913,0.0049032182,0.0001136777,0.001186521,0.00003846979,0.2498187,0.000067674715,0.0006248544,0.5380708,0.13498561,0.07002551],"study_design_scores_gemma":[0.00017312434,0.00016549561,0.01003446,0.0001829798,0.00021938577,0.000051509858,0.4020225,0.00019459547,0.000090063084,0.021737747,0.56498283,0.00014530375],"about_ca_topic_score_codex":0.000014798275,"about_ca_topic_score_gemma":0.000057394067,"teacher_disagreement_score":0.51633304,"about_ca_system_score_codex":0.000022566563,"about_ca_system_score_gemma":0.00007193161,"threshold_uncertainty_score":0.27516958},"labels":[],"label_agreement":null},{"id":"W4399732990","doi":"10.22148/001c.116372","title":"Neither Corpus Nor Edition: Building a Pipeline to Make Data Analysis Possible on Medieval Arabic Commentary Traditions","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Computer science; Macro; Suite; Python (programming language); Macro level; Software; Representation (politics); Pipeline (software); Arabic; Natural language processing; Information retrieval; Artificial intelligence; Programming language; Linguistics; History","score_opus":0.04664510234599157,"score_gpt":0.3418212453957643,"score_spread":0.29517614304977274,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399732990","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008780443,0.0028754398,0.9241002,0.06262686,0.0010073748,0.00011219291,0.00016433575,0.00020920082,0.00012395624],"genre_scores_gemma":[0.38705817,0.00023161268,0.6003385,0.00893847,0.0026696287,0.0000027729786,0.00011876993,0.000022850414,0.0006192107],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99807036,0.00006281732,0.0005814262,0.00033640972,0.00071120216,0.00023781354],"domain_scores_gemma":[0.99850905,0.00015506668,0.00023667757,0.00055791857,0.0003301773,0.00021108135],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00058597734,0.00019890415,0.00036716272,0.0006242357,0.00013955262,0.00068405963,0.0017264453,0.00006692384,0.00003681645],"category_scores_gemma":[0.000107974454,0.0001275649,0.00024995842,0.0022879723,0.00004239302,0.0011770837,0.00020379433,0.0005194322,0.00001003793],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000100151876,0.00042211014,0.00029832506,0.00013056249,0.002510901,0.0015588958,0.0024153984,0.004503796,0.006473224,0.016807126,0.8913528,0.07342674],"study_design_scores_gemma":[0.00088929484,0.0014685305,0.00076379295,0.0018384143,0.004510544,0.0017362728,0.00045410157,0.7957964,0.01291506,0.045175344,0.13309276,0.001359483],"about_ca_topic_score_codex":0.000023047329,"about_ca_topic_score_gemma":0.0000519764,"teacher_disagreement_score":0.7912926,"about_ca_system_score_codex":0.00017572613,"about_ca_system_score_gemma":0.00006806203,"threshold_uncertainty_score":0.65964025},"labels":[],"label_agreement":null},{"id":"W4399732994","doi":"10.22148/001c.116368","title":"Exploring Gender Differences in Fatwa through Machine Learning","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Popularity; Context (archaeology); Computer science; Preprocessor; Margin (machine learning); Artificial intelligence; Machine learning; Thematic analysis; Data science; Psychology; Social psychology; Qualitative research; Sociology; Social science","score_opus":0.2739161805313298,"score_gpt":0.31646283693417326,"score_spread":0.04254665640284344,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399732994","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6044366,0.002984619,0.38945028,0.001522206,0.00086106546,0.00002710974,2.957082e-7,0.00004668175,0.00067115756],"genre_scores_gemma":[0.98161125,0.0007835067,0.017082509,0.000050125207,0.00023833792,5.042508e-7,2.0099525e-7,0.000004239251,0.00022932848],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988988,0.000046787176,0.00040516997,0.00013936343,0.00033678568,0.00017308748],"domain_scores_gemma":[0.99959254,0.00006831164,0.00010641317,0.000094280585,0.00008761384,0.000050843806],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00026754837,0.00010262284,0.00020569646,0.00011929757,0.00004600418,0.0002840882,0.00042819662,0.000028100621,0.000011145333],"category_scores_gemma":[0.00006217996,0.00006343313,0.00011560345,0.00037412616,0.000015048829,0.0017077245,0.00009557515,0.000491954,0.000007298593],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004768592,0.00028395458,0.09135508,0.00065164745,0.0009876194,0.0041105063,0.15873213,0.2506318,0.00610314,0.27134275,0.0010858431,0.21466783],"study_design_scores_gemma":[0.00018854972,0.00007861608,0.0033045677,0.00023282388,0.000026438167,0.00022839553,0.0010346244,0.98418146,0.00030039327,0.006535312,0.0037077703,0.00018103511],"about_ca_topic_score_codex":0.000023213684,"about_ca_topic_score_gemma":0.00000779897,"teacher_disagreement_score":0.73354965,"about_ca_system_score_codex":0.000069245274,"about_ca_system_score_gemma":0.000038388436,"threshold_uncertainty_score":0.2739469},"labels":[],"label_agreement":null},{"id":"W4399733006","doi":"10.22148/001c.116239","title":"Two translations of Mahfouz’s _Awlad Haratina_ (Children of our Alley): A computational-stylistic analysis","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Variety (cybernetics); Readability; Linguistics; Style (visual arts); Context (archaeology); Alley; Computer science; Sentence; Natural language processing; Artificial intelligence; Literature; History; Art; Philosophy","score_opus":0.041846418159801196,"score_gpt":0.34727249025675117,"score_spread":0.30542607209694994,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399733006","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.084356554,0.001069511,0.9112709,0.0027435925,0.0001785936,0.00006898542,0.0000482083,0.000023929375,0.00023973097],"genre_scores_gemma":[0.9500728,0.000038414008,0.04956883,0.000038742488,0.00007887987,3.3508792e-7,0.000020046602,0.000004140786,0.0001777778],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981393,0.00010759919,0.00091614056,0.00014642772,0.000551296,0.00013923136],"domain_scores_gemma":[0.99835455,0.00013723552,0.0005067322,0.00014872575,0.00074984133,0.00010290563],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005676265,0.00012766958,0.000430294,0.00043632396,0.00006322094,0.0001091922,0.00045359237,0.000051890012,0.000024565019],"category_scores_gemma":[0.00007879105,0.000091035035,0.0005519194,0.0018371012,0.000040426712,0.0004625078,0.000037490718,0.00024821106,0.0000056414456],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003643363,0.0001934079,0.010498525,0.00013567667,0.0045219082,0.000037774756,0.0043790927,0.7997981,0.0011666914,0.1731524,0.0015671216,0.0045129103],"study_design_scores_gemma":[0.00048958545,0.00017384556,0.026918367,0.0001594465,0.0016445243,0.0001304132,0.00042735552,0.96120757,0.00071043964,0.0076297065,0.00029283515,0.0002159335],"about_ca_topic_score_codex":0.000022172984,"about_ca_topic_score_gemma":0.000015407419,"teacher_disagreement_score":0.8657163,"about_ca_system_score_codex":0.000039210932,"about_ca_system_score_gemma":0.00016321942,"threshold_uncertainty_score":0.37123013},"labels":[],"label_agreement":null},{"id":"W4399733018","doi":"10.22148/001c.117486","title":"Poet-Composer Collaborations in Modern Egyptian Song: A Social Network Analysis Approach to Music History","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Musical; Narrative; Context (archaeology); Social network analysis; Suite; Centrality; Social network (sociolinguistics); Literature; History; Art; Computer science; World Wide Web; Archaeology; Social media","score_opus":0.0414461494911035,"score_gpt":0.2857449999835815,"score_spread":0.244298850492478,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399733018","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4667948,0.0031528918,0.50192976,0.002507381,0.0007303308,0.00044689214,0.00004761275,0.00013670835,0.024253648],"genre_scores_gemma":[0.9904661,0.000009743984,0.006792069,0.00009277902,0.0017768181,0.000006449338,0.000038919585,0.00001768477,0.00079941336],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99820757,0.00010151903,0.00076339336,0.00023215635,0.00038261217,0.00031273454],"domain_scores_gemma":[0.99898845,0.000056290428,0.00026834733,0.00015779324,0.0003859676,0.00014316893],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004086745,0.00021685062,0.0006396955,0.00057670125,0.000104367115,0.00018180526,0.00031384858,0.00005997827,0.00018902699],"category_scores_gemma":[0.0000062229296,0.0001709801,0.0007085388,0.0031354523,0.000044563818,0.00028993102,0.00004914048,0.00043042802,0.000006608664],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000081121965,0.000520434,0.01025754,0.000033840548,0.0069447113,0.00003797482,0.0138497325,0.6018003,0.0005162456,0.04392866,0.31278363,0.009245798],"study_design_scores_gemma":[0.00044056654,0.00014109972,0.0052073803,0.00011017628,0.005414324,0.000008896293,0.004137405,0.8680818,0.00002398174,0.016432857,0.09923904,0.0007624911],"about_ca_topic_score_codex":0.00007828084,"about_ca_topic_score_gemma":0.00016412194,"teacher_disagreement_score":0.5236713,"about_ca_system_score_codex":0.000417926,"about_ca_system_score_gemma":0.00016660511,"threshold_uncertainty_score":0.6972366},"labels":[],"label_agreement":null},{"id":"W4400779700","doi":"10.22148/001c.118152","title":"A Digital Archaeology of Early Hispanic Film Culture: Film Magazines and the Male Fan Reader","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Cinema and Media Studies","field":"Economics, Econometrics and Finance","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Movie theater; Fandom; Reading (process); Popular culture; Affordance; Interactivity; Film studies; Appropriation; Media studies; Visual arts; Art; Sociology; History; Multimedia; Computer science; Linguistics","score_opus":0.02592223816153074,"score_gpt":0.23650642669449048,"score_spread":0.21058418853295974,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400779700","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92851686,0.04811964,0.00038562258,0.010817808,0.00078527426,0.00011241511,0.000102967984,0.000009910539,0.011149466],"genre_scores_gemma":[0.9766065,0.0043006204,0.00012865801,0.000089066554,0.0002959161,0.0000011702853,0.0000026495904,0.000007286608,0.01856812],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99906063,0.000009266671,0.0006096199,0.000116409174,0.00005709566,0.0001469921],"domain_scores_gemma":[0.99931115,0.00010616506,0.0003252094,0.00008705564,0.000113580885,0.00005683985],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002231641,0.000115593764,0.0004794472,0.00010951777,0.000037930946,0.00007649597,0.00014301781,0.00005566858,0.000079334895],"category_scores_gemma":[0.00024171703,0.000062960506,0.00023091403,0.00019024481,0.00033661496,0.0002646065,0.00006811168,0.00021894102,0.000026759877],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00087040424,0.00026389284,0.1612308,0.0009555673,0.0051032165,0.00045406708,0.10637652,0.0003821094,0.00053526054,0.4906011,0.22454284,0.008684216],"study_design_scores_gemma":[0.0058282763,0.0018687786,0.075275995,0.0005767964,0.0007339941,0.0030703852,0.013641028,0.011041383,0.00018050836,0.08525763,0.8016009,0.00092437945],"about_ca_topic_score_codex":0.000012497702,"about_ca_topic_score_gemma":0.000016913546,"teacher_disagreement_score":0.577058,"about_ca_system_score_codex":0.000021798382,"about_ca_system_score_gemma":0.00001506139,"threshold_uncertainty_score":0.2567455},"labels":[],"label_agreement":null},{"id":"W4400779769","doi":"10.22148/001c.120944","title":"Digital Film Historiography: Challenges of/and Interdisciplinarity","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Historiography; History; Archaeology","score_opus":0.07071047823287271,"score_gpt":0.2724238292079205,"score_spread":0.2017133509750478,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400779769","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5401804,0.03008756,0.000011727173,0.0019583148,0.0014539239,0.000051951043,0.00007978351,0.000032003034,0.42614433],"genre_scores_gemma":[0.9867063,0.0005709881,0.00001258279,0.000016732356,0.00064428675,2.0203274e-7,0.0000021291685,0.0000073084143,0.01203945],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99922985,0.000008808435,0.00036690387,0.00007680524,0.00021422657,0.0001033849],"domain_scores_gemma":[0.9994156,0.000057901976,0.00013564326,0.00005791694,0.00026541,0.000067531444],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00013253992,0.00010655277,0.00022810136,0.00015064125,0.00007974012,0.0007495687,0.00013076715,0.000027537404,0.00014402301],"category_scores_gemma":[0.000027060603,0.000065152286,0.00026200534,0.00003224727,0.00024083491,0.0013427922,0.000051758077,0.000213867,0.000004408099],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000047895697,0.00009990857,0.00015212977,0.00035256604,0.0004560768,0.00009955073,0.063146755,0.000011915877,0.000010398488,0.8833488,0.014760798,0.03751315],"study_design_scores_gemma":[0.00017763047,0.00060140004,0.00033987346,0.0005003074,0.00017750416,0.00012046098,0.0442405,0.00022026208,0.000011826885,0.036676124,0.91672474,0.00020934141],"about_ca_topic_score_codex":0.0000036396495,"about_ca_topic_score_gemma":0.00006676009,"teacher_disagreement_score":0.90196395,"about_ca_system_score_codex":0.000028202605,"about_ca_system_score_gemma":0.000017976996,"threshold_uncertainty_score":0.7228108},"labels":[],"label_agreement":null},{"id":"W4400779789","doi":"10.22148/001c.118497","title":"Revisiting Weimar Film Reviewers’ Sentiments: Integrating Lexicon-Based Sentiment Analysis with Large Language Models","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Sentiment Analysis and Opinion Mining","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Lexicon; Sentiment analysis; Natural language processing; Computer science; Linguistics; Artificial intelligence; Philosophy","score_opus":0.0232972015043026,"score_gpt":0.3005641542562081,"score_spread":0.2772669527519055,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400779789","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.036281142,0.009152654,0.95062613,0.0024799588,0.0002934148,0.00011482032,0.000009038299,0.000080174905,0.00096265593],"genre_scores_gemma":[0.9277602,0.00027286515,0.06997344,0.00049635506,0.0003896819,0.0000018653006,0.000028314133,0.00001821458,0.0010590799],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971406,0.00014734888,0.001016605,0.00036701618,0.00096361566,0.00036482455],"domain_scores_gemma":[0.9981003,0.00013090455,0.00069521746,0.00036353077,0.0005366334,0.00017346014],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013503974,0.00028636437,0.00071695383,0.00062996417,0.00016593843,0.001019127,0.00062578527,0.000057161385,0.00014756314],"category_scores_gemma":[0.000053622545,0.00017068959,0.0009097575,0.0026455403,0.000030969353,0.0013355557,0.00010813833,0.0003785112,0.000018748304],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013460938,0.0008903712,0.01644062,0.0015282629,0.044621885,0.003807701,0.030197756,0.7446261,0.0055066072,0.06764058,0.024731258,0.05987425],"study_design_scores_gemma":[0.00032884936,0.000077719684,0.00006691324,0.00093100325,0.0019954205,0.000035104746,0.0020568385,0.99169147,0.0007039455,0.000089013374,0.001768669,0.0002550303],"about_ca_topic_score_codex":0.000011846146,"about_ca_topic_score_gemma":0.000005984494,"teacher_disagreement_score":0.891479,"about_ca_system_score_codex":0.00016382129,"about_ca_system_score_gemma":0.00014518689,"threshold_uncertainty_score":0.9827465},"labels":[],"label_agreement":null},{"id":"W4400779800","doi":"10.22148/001c.118494","title":"A Digital Trail of Rupture. The German Film Exile 1933-1945 in the Data of Günter Peter Straschek","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Italian Fascism and Post-war Society","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"German; Film director; Subject (documents); Art history; Estate; Filmmaking; Nazism; Treasure; History; Social connectedness; Art; Political science; Library science; Computer science; Law; Psychology; Movie theater; Archaeology","score_opus":0.05298380408686617,"score_gpt":0.3425772735089551,"score_spread":0.28959346942208897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400779800","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9711493,0.0014665169,0.000043622633,0.007605954,0.0004958975,0.00015290521,0.0003523678,0.000007723413,0.018725665],"genre_scores_gemma":[0.99726796,0.00016078973,0.00006536657,0.00020610233,0.00040489325,2.9182053e-7,0.000013878539,0.000004804772,0.0018758818],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.99849707,0.00011591647,0.00047555362,0.00009467579,0.00063941366,0.00017739224],"domain_scores_gemma":[0.9990927,0.00024201466,0.00023355286,0.00021862208,0.00016375518,0.00004935152],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011387164,0.00009515505,0.00021088304,0.00003813709,0.00010790034,0.00029904264,0.0010093169,0.000069969734,0.00010018525],"category_scores_gemma":[0.0001561946,0.000042254425,0.00026940758,0.00039449584,0.00043709503,0.0007826381,0.00006369871,0.00035360412,0.0000060230445],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006859485,0.0002989183,0.0033835946,0.00020175317,0.00067889167,0.00016415905,0.6480538,0.00011724612,0.0006280347,0.008211615,0.3305046,0.0076887626],"study_design_scores_gemma":[0.0007030598,0.00035331975,0.017579924,0.0005772108,0.000537506,0.00016552124,0.24896944,0.0017059367,0.00021611102,0.004025006,0.7247757,0.00039127542],"about_ca_topic_score_codex":0.00006314069,"about_ca_topic_score_gemma":0.00024590126,"teacher_disagreement_score":0.3990844,"about_ca_system_score_codex":0.000028126595,"about_ca_system_score_gemma":0.00018440494,"threshold_uncertainty_score":0.2883675},"labels":[],"label_agreement":null},{"id":"W4400779908","doi":"10.22148/001c.118589","title":"Quantifying Women’s Marginalisation in Ibero-American Film Culture During the First Half of the Twentieth Century: A Network-Science Proposal","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social and Cultural Dynamics","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Novelty; Situated; Perspective (graphical); Sociology; Field (mathematics); Social network analysis; Phenomenon; Social network (sociolinguistics); Public sphere; Position (finance); Gender studies; Social science; Epistemology; Psychology; Visual arts; Social psychology; Political science; Computer science; Social media; Art; Mathematics","score_opus":0.01990972322105852,"score_gpt":0.30292276167201027,"score_spread":0.28301303845095177,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400779908","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9870839,0.0006580291,0.000009057057,0.010227034,0.0011626604,0.00016398232,0.0000038348944,0.000014328034,0.0006771514],"genre_scores_gemma":[0.9972887,0.0008621393,0.000114532784,0.00010211727,0.0006542176,0.0000028146992,6.240658e-7,0.000006239281,0.00096860493],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99784875,0.00014944626,0.0004716978,0.00014333453,0.00096574536,0.00042105268],"domain_scores_gemma":[0.99885285,0.00006215924,0.00047868412,0.00010423726,0.0004114311,0.00009062896],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012604991,0.00013344266,0.00025295492,0.000058926784,0.0008838184,0.00033969164,0.00068570406,0.000060922524,0.000028731183],"category_scores_gemma":[0.00025716313,0.00005972457,0.00022827538,0.002362362,0.0010813294,0.00061711774,0.00008606814,0.00049162534,0.000001648024],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018821398,0.00020309526,0.109284386,0.00032754435,0.00039036217,0.00006816835,0.75104654,0.023266668,0.0030975565,0.10160959,0.005359389,0.005158482],"study_design_scores_gemma":[0.0007185834,0.00027418256,0.23434263,0.0014572594,0.0002985647,0.00009404715,0.58544207,0.010553743,0.00022261392,0.0058905114,0.16002573,0.0006800884],"about_ca_topic_score_codex":0.00073224184,"about_ca_topic_score_gemma":0.0050643277,"teacher_disagreement_score":0.16560452,"about_ca_system_score_codex":0.00068386784,"about_ca_system_score_gemma":0.00035942034,"threshold_uncertainty_score":0.6797706},"labels":[],"label_agreement":null},{"id":"W4400779983","doi":"10.22148/001c.118495","title":"Soviet View of the World. Exploring Long-Term Visual Patterns in “Novosti dnia” Newsreel Journal (1945-1992)","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Italian Fascism and Post-war Society","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Soviet union; Period (music); Visual culture; World War II; Term (time); Political science; History; Media studies; Visual arts; Sociology; Law; Politics; Art; Aesthetics","score_opus":0.06607763407418345,"score_gpt":0.35072104309476015,"score_spread":0.2846434090205767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400779983","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9869636,0.0027634376,0.0001093354,0.005989729,0.0026217126,0.00011534519,0.000012360997,0.000012051245,0.001412427],"genre_scores_gemma":[0.99267304,0.0031305328,0.00009056777,0.00024151868,0.0016450102,7.908943e-7,0.0000012725811,0.000013626426,0.0022036224],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9973778,0.00025879205,0.00088183983,0.00012842934,0.00097096636,0.0003821755],"domain_scores_gemma":[0.9987379,0.00015568586,0.0004734608,0.000117087824,0.00031795885,0.00019792876],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014459535,0.00017790985,0.00039695253,0.00016193127,0.00030755578,0.00038538643,0.00064756337,0.00009145518,0.00026424252],"category_scores_gemma":[0.0001756983,0.00010278533,0.0006403823,0.0010560398,0.00022365768,0.001010575,0.00008843226,0.00084890437,0.000011328461],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015138344,0.0006325978,0.60336304,0.0009878447,0.0013913307,0.0014010812,0.23083137,0.00084488955,0.0018785837,0.0059127156,0.06491423,0.08769092],"study_design_scores_gemma":[0.0036710615,0.0010365039,0.5610354,0.017492099,0.0021157018,0.0013809532,0.16987355,0.001217925,0.0024109646,0.004048816,0.23351121,0.0022058003],"about_ca_topic_score_codex":0.00014222975,"about_ca_topic_score_gemma":0.0026581816,"teacher_disagreement_score":0.16859698,"about_ca_system_score_codex":0.00021556078,"about_ca_system_score_gemma":0.00038086408,"threshold_uncertainty_score":0.41914642},"labels":[],"label_agreement":null},{"id":"W4400780080","doi":"10.22148/001c.118499","title":"Approaching a National Film History through Data. Network Analysis in German Film History","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"European history and politics","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"German; German history; History; Archaeology","score_opus":0.12789732649818666,"score_gpt":0.35547066922394693,"score_spread":0.22757334272576027,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400780080","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0675662,0.09138088,0.0034912676,0.003916154,0.011297465,0.00030392696,0.00012290574,0.00018480123,0.8217364],"genre_scores_gemma":[0.933615,0.0013905464,0.002662891,0.0014469868,0.0057395967,0.0000010010642,0.000098901364,0.000028544153,0.0550165],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975902,0.0003990888,0.00065217866,0.00021888944,0.0008148029,0.00032483842],"domain_scores_gemma":[0.99883366,0.0002326987,0.00032361152,0.00019155302,0.0002738352,0.00014462319],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021815228,0.00014573189,0.00035369507,0.0002677698,0.00016549128,0.00006946357,0.00066586374,0.000095556796,0.0006315427],"category_scores_gemma":[0.00036443208,0.00011915659,0.00030326366,0.0006446535,0.0002726488,0.001175429,0.00007299465,0.00063606683,0.000030463436],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000018420018,0.000070275826,0.0006446855,0.00004254651,0.000655365,0.00018952636,0.045471545,0.021553008,0.0000075167195,0.13811722,0.7930771,0.00015277907],"study_design_scores_gemma":[0.00009066009,0.000019553205,0.00038159496,0.00006087039,0.000495248,0.000013435927,0.0010289757,0.019597886,2.120675e-7,0.0010442716,0.97710866,0.00015866019],"about_ca_topic_score_codex":0.00029122067,"about_ca_topic_score_gemma":0.0009691839,"teacher_disagreement_score":0.8660488,"about_ca_system_score_codex":0.0034815664,"about_ca_system_score_gemma":0.0010380372,"threshold_uncertainty_score":0.9104175},"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,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_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","score_opus":0.06055083490596529,"score_gpt":0.3632976391978363,"score_spread":0.302746804291871,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402197367","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99347067,0.0016011769,0.00088964857,0.0010740431,0.00020352258,0.00006862668,0.00013537449,0.000015629897,0.0025412964],"genre_scores_gemma":[0.9974979,0.00015031786,0.000047250523,0.0000946925,0.00045366888,3.9791038e-7,0.00004088126,0.0000025472414,0.0017123445],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99928963,0.000046863635,0.00023527956,0.000058939066,0.0002854064,0.00008389419],"domain_scores_gemma":[0.9993974,0.00008153163,0.00010667245,0.00001760731,0.00034148528,0.00005530995],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00016315345,0.000059362323,0.00012288612,0.00003962534,0.00010444082,0.000248488,0.00006150744,0.00003778172,0.00015850966],"category_scores_gemma":[0.00017848953,0.00003754871,0.00007492314,0.00018201169,0.000083723346,0.0007764521,0.000015465557,0.00007782328,0.000010821773],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012193861,0.0005843615,0.0024547738,0.00042942466,0.0030569641,0.00047997973,0.5493865,0.00054907426,0.052074503,0.16156745,0.07003352,0.15816405],"study_design_scores_gemma":[0.0018072361,0.0014054609,0.014781052,0.0017999193,0.001896055,0.00009434714,0.7991282,0.0013890904,0.0075651673,0.04487829,0.12437135,0.0008838629],"about_ca_topic_score_codex":0.000033226795,"about_ca_topic_score_gemma":0.00008743265,"teacher_disagreement_score":0.24974166,"about_ca_system_score_codex":0.000058726313,"about_ca_system_score_gemma":0.000042283293,"threshold_uncertainty_score":0.23961754},"labels":[],"label_agreement":null},{"id":"W4408609774","doi":"10.22148/001c.131682","title":"Racial and Ethnic Representation in Literature Taught in US High Schools","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Critical Race Theory in Education","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Ethnic group; Representation (politics); Mathematics education; Sociology; Psychology; Political science; Anthropology; Law; Politics","score_opus":0.02654759242506276,"score_gpt":0.4352521767272682,"score_spread":0.4087045843022054,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408609774","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97695696,0.0006756991,0.000024669516,0.020776896,0.00035656276,0.00006693381,0.0000010327327,0.0000030713381,0.0011381683],"genre_scores_gemma":[0.9974222,0.0007113114,0.00029089343,0.00020533195,0.00026912545,8.1213716e-7,0.0000016862937,0.0000017260271,0.001096884],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.998976,0.00023629004,0.00034974466,0.00007976205,0.00022271519,0.00013547332],"domain_scores_gemma":[0.9992902,0.00019379333,0.00011626269,0.00005005953,0.0002877525,0.00006189395],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00081574114,0.000056920824,0.00016147918,0.00020499302,0.00006603176,0.00012895942,0.00011466347,0.00009653429,0.000027224549],"category_scores_gemma":[0.0024966602,0.000044266966,0.000043651507,0.0010272334,0.00009451589,0.0005383953,0.00001572948,0.0004122024,0.0000015943555],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005958142,0.00063281035,0.14341067,0.00012526255,0.00011005575,0.00021000089,0.31369704,0.002062663,0.0047000735,0.4776488,0.0057625645,0.051044248],"study_design_scores_gemma":[0.0015832239,0.00010302695,0.48694748,0.0010250397,0.00010851416,0.000016912822,0.15973711,0.00033494798,0.0004997563,0.34410015,0.005285126,0.00025872415],"about_ca_topic_score_codex":0.00021549306,"about_ca_topic_score_gemma":0.0036793924,"teacher_disagreement_score":0.3435368,"about_ca_system_score_codex":0.00021187027,"about_ca_system_score_gemma":0.00013487544,"threshold_uncertainty_score":0.29889157},"labels":[],"label_agreement":null},{"id":"W4408609775","doi":"10.22148/001c.128008","title":"Quantifying the Presence of Ancient Greek and Latin Classics in Early Modern Britain","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Historical and Literary Studies","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Classics; History; Ancient Greek; Ancient history; Art","score_opus":0.06673526972641931,"score_gpt":0.3292331578435328,"score_spread":0.2624978881171135,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408609775","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99110895,0.0027663556,0.00017476162,0.003404383,0.00014228155,0.00004541672,0.0000013702379,0.000002056316,0.0023544265],"genre_scores_gemma":[0.997123,0.00040677766,0.00019852752,0.000112649446,0.000061131,2.7185757e-7,6.795969e-8,0.0000010106891,0.0020965787],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99913377,0.000101275786,0.0003151413,0.00005616854,0.00027359027,0.00012005219],"domain_scores_gemma":[0.99930406,0.00020651966,0.00017917145,0.00004196364,0.00023032381,0.0000379485],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005342596,0.000050487735,0.00017468838,0.000051374933,0.00018422386,0.000037452966,0.0001579748,0.000031714913,0.0000023015418],"category_scores_gemma":[0.00041212965,0.000027583363,0.00006334416,0.00040427214,0.00023748385,0.00018985075,0.00003668864,0.00016752076,2.2653033e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000100353216,0.0001714057,0.48995903,0.00009115588,0.00016634009,0.000058558733,0.41712382,0.0009293595,0.0009915509,0.05855773,0.010210083,0.02164061],"study_design_scores_gemma":[0.0010664624,0.00029693244,0.55441934,0.00076737185,0.00019960251,0.000009836392,0.057488844,0.010553831,0.00009164535,0.014535365,0.36024597,0.0003247734],"about_ca_topic_score_codex":0.00045373235,"about_ca_topic_score_gemma":0.0012264053,"teacher_disagreement_score":0.35963497,"about_ca_system_score_codex":0.000062361905,"about_ca_system_score_gemma":0.00006657156,"threshold_uncertainty_score":0.14169197},"labels":[],"label_agreement":null},{"id":"W4409678925","doi":"10.22148/001c.128010","title":"The “Mapping German fiction in translation” dataset: Data collection, scope, and data quality","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Scope (computer science); German; Computer science; Translation (biology); Data quality; Quality (philosophy); Data collection; Information retrieval; Data science; History; Statistics; Engineering; Operations management; Mathematics","score_opus":0.10180379937609223,"score_gpt":0.4145446143603592,"score_spread":0.312740814984267,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409678925","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005491335,0.031454235,0.93476254,0.026851578,0.00059738033,0.0002480942,0.00018220229,0.00007743648,0.0003351955],"genre_scores_gemma":[0.38266876,0.006791061,0.6076771,0.0011716764,0.00031925074,0.0000019201398,0.00072438834,0.000011053612,0.00063475],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988386,0.00010898774,0.0005085457,0.00020153465,0.0002336047,0.00010871489],"domain_scores_gemma":[0.99853426,0.00018196848,0.0002968801,0.0007666526,0.00018931467,0.000030893752],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016651682,0.00007890882,0.00015102187,0.00012194511,0.00023151097,0.0004255852,0.0020423564,0.00004320404,8.8530965e-7],"category_scores_gemma":[0.00037978086,0.000046353256,0.000016042954,0.00079580303,0.000052892545,0.002107254,0.000523468,0.00028326287,2.4854646e-7],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011308745,0.00015974422,0.0028178282,0.00021747463,0.00031313076,0.000052607826,0.0025290966,0.00012023693,0.003924206,0.017999157,0.5959398,0.3758136],"study_design_scores_gemma":[0.0014485485,0.00009060881,0.00899726,0.0007590202,0.00014641033,0.00027764516,0.0008072426,0.5656325,0.0013473599,0.02724032,0.39275458,0.0004984748],"about_ca_topic_score_codex":0.000066584435,"about_ca_topic_score_gemma":0.00021860936,"teacher_disagreement_score":0.5655123,"about_ca_system_score_codex":0.00004536223,"about_ca_system_score_gemma":0.00011402781,"threshold_uncertainty_score":0.4103928},"labels":[],"label_agreement":null},{"id":"W4410984062","doi":"10.22148/001c.138350","title":"Online Reviews as Communal Formations. A Corpus Linguistic Approach","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Communication and Language","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Linguistics; Corpus linguistics; History; Computer science; Natural language processing; Philosophy","score_opus":0.04906321785311471,"score_gpt":0.33377345979714335,"score_spread":0.2847102419440286,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410984062","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014592907,0.02106002,0.3789395,0.008878114,0.00067051064,0.00038722943,0.00001055146,0.000106453524,0.5753547],"genre_scores_gemma":[0.8940931,0.0017300693,0.09408761,0.0036833615,0.00008754321,0.0000015716369,0.000021442285,0.0000044535795,0.0062908563],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99890417,0.00008587104,0.00063193607,0.00006302862,0.0002044738,0.00011051377],"domain_scores_gemma":[0.99856526,0.000072768,0.0004378973,0.00040920792,0.0004419744,0.00007289811],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00038432537,0.00009723885,0.00026510045,0.00012887378,0.00010317697,0.000265121,0.0012790953,0.00003303881,0.000007655106],"category_scores_gemma":[0.00056403095,0.00006176773,0.00018700045,0.00052981253,0.000040580497,0.00049130426,0.0002049531,0.0002989747,0.000016488593],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000346062,0.0014441578,0.00025672963,0.0002159893,0.0002974273,0.00003790453,0.005967131,0.000990921,0.00016238465,0.8453979,0.038309857,0.10688503],"study_design_scores_gemma":[0.0006329104,0.00014414173,0.0006825602,0.000311325,0.000091402144,0.00032569256,0.0008325156,0.05575103,0.00007335016,0.00860664,0.9323414,0.00020701702],"about_ca_topic_score_codex":0.000006680213,"about_ca_topic_score_gemma":0.0000061664464,"teacher_disagreement_score":0.8940316,"about_ca_system_score_codex":0.000053574408,"about_ca_system_score_gemma":0.00008474045,"threshold_uncertainty_score":0.25565678},"labels":[],"label_agreement":null},{"id":"W4413346466","doi":"10.22148/001c.142979","title":"The Eclectic Reader","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social and Cultural Dynamics","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Computer science; Psychology","score_opus":0.02647182788071664,"score_gpt":0.35404789215881044,"score_spread":0.3275760642780938,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413346466","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5297739,0.00553238,0.0004824463,0.16614394,0.0055773403,0.0002855765,0.0000029403468,0.00006742416,0.29213408],"genre_scores_gemma":[0.94879353,0.0013751772,0.00011477811,0.00054938084,0.00061625405,5.153995e-7,3.4952896e-7,0.000002553157,0.048547484],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99902606,0.000106067586,0.00029026854,0.000048843067,0.00033731794,0.00019146642],"domain_scores_gemma":[0.99884194,0.00017357069,0.00020972583,0.000056688445,0.0006499767,0.00006806766],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00052710634,0.00006485373,0.00013474037,0.000027320693,0.0010065604,0.00020830559,0.00032723343,0.00006721326,0.00002244967],"category_scores_gemma":[0.00070338533,0.00003166522,0.00018672271,0.0005048838,0.00026502425,0.00020483807,0.000020189906,0.0002242117,0.000008197729],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000103376886,0.000101795755,0.007273528,0.000017732425,0.00065798307,0.000028498107,0.029258227,0.00027538027,0.00039466293,0.73566383,0.17258859,0.053636402],"study_design_scores_gemma":[0.0002989288,0.000052711515,0.007470348,0.00006290496,0.0002144116,0.000009214634,0.08690044,0.00022531282,0.00004528393,0.05135372,0.8532266,0.00014009165],"about_ca_topic_score_codex":0.00012378625,"about_ca_topic_score_gemma":0.0009451496,"teacher_disagreement_score":0.68431014,"about_ca_system_score_codex":0.00014685633,"about_ca_system_score_gemma":0.00018613548,"threshold_uncertainty_score":0.7741751},"labels":[],"label_agreement":null},{"id":"W4413916071","doi":"10.22148/001c.142973","title":"Geotropes: Situating Postcolonial Bestsellers in the Global Literary Marketplace","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Postcolonial and Cultural Literary Studies","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Hospitality and Tourism Research Centre, Hong Kong Polytechnic University","keywords":"Art; Literature; Sociology; History; Aesthetics; Art history","score_opus":0.020252630385960747,"score_gpt":0.2724122058863573,"score_spread":0.2521595755003965,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413916071","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9178341,0.00289435,0.000003577919,0.018645693,0.00077440526,0.00013628854,0.000038446342,0.00001282721,0.059660334],"genre_scores_gemma":[0.98868537,0.00012740676,0.00007893486,0.002559359,0.0007154293,0.0000018141966,0.0000055835935,0.0000030573049,0.007823038],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986579,0.00013307497,0.00058681815,0.00010471005,0.00028206044,0.00023546211],"domain_scores_gemma":[0.999157,0.00012368248,0.00025940727,0.00008632554,0.0003350925,0.000038460432],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003468953,0.00017194143,0.0003072668,0.000064303014,0.000334379,0.0003800599,0.00039028548,0.000040079434,0.00012323749],"category_scores_gemma":[0.00011769202,0.00008239276,0.00022876648,0.00020336096,0.00016902128,0.00051914057,0.00005887328,0.00028997194,0.000012442493],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016393587,0.0008149404,0.029818323,0.00026187513,0.0014962362,0.00077678595,0.43153286,0.0005533228,0.00016359758,0.28486252,0.22918497,0.018895214],"study_design_scores_gemma":[0.0014791366,0.00052512414,0.0129142795,0.0005408556,0.0003438523,0.0001488756,0.0721267,0.0004258438,0.000013499735,0.016392719,0.8947216,0.00036751022],"about_ca_topic_score_codex":0.00007895332,"about_ca_topic_score_gemma":0.0007209457,"teacher_disagreement_score":0.66553664,"about_ca_system_score_codex":0.000108046646,"about_ca_system_score_gemma":0.000056008575,"threshold_uncertainty_score":0.36649263},"labels":[],"label_agreement":null},{"id":"W4414090048","doi":"10.22148/001c.143671","title":"Beyond Plot: How Sentiment Analysis Reshapes Our Understanding of Narrative Structure","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language, Metaphor, and Cognition","field":"Psychology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Sentiment analysis; Narrative structure; Discourse analysis; Perspective (graphical)","score_opus":0.03671393494212861,"score_gpt":0.33310064768971726,"score_spread":0.29638671274758865,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414090048","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9790827,0.0014046436,0.007266581,0.0047686156,0.0009154863,0.00011990213,0.000057531317,0.000012895342,0.0063716127],"genre_scores_gemma":[0.99380624,0.000033141147,0.00033648865,0.00019343033,0.00017412806,4.36282e-7,0.000012740131,0.0000055618216,0.0054378416],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9987184,0.00014048403,0.00048033948,0.00014027608,0.0003483575,0.00017214398],"domain_scores_gemma":[0.9985954,0.00004830044,0.00067852193,0.00013529237,0.00047469957,0.00006775067],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00022691692,0.00015251651,0.00050823856,0.00045934942,0.00009994859,0.0000673273,0.00017878765,0.00011170596,0.00029724676],"category_scores_gemma":[0.00006407118,0.00009683225,0.0004629259,0.00087564375,0.00004915735,0.00017371099,0.00002461851,0.00027688176,0.0000015575374],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029633932,0.001583823,0.049042422,0.0004670478,0.14025721,0.0008876047,0.34340572,0.0058187083,0.06854765,0.17699452,0.20382805,0.006203841],"study_design_scores_gemma":[0.0028235032,0.0005002226,0.016159732,0.00020312077,0.014529891,0.000092800874,0.91386646,0.0008396116,0.008226425,0.041227326,0.0010874566,0.0004434625],"about_ca_topic_score_codex":0.0000145544345,"about_ca_topic_score_gemma":0.00010463913,"teacher_disagreement_score":0.57046074,"about_ca_system_score_codex":0.0001382559,"about_ca_system_score_gemma":0.00003975724,"threshold_uncertainty_score":0.39487046},"labels":[],"label_agreement":null},{"id":"W4414090244","doi":"10.22148/001c.137088","title":"Parler Games: A Narrative Framework Analysis of Parler Conspiracy Theories and the January 6th Insurrection","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Narrative; Event (particle physics); Conceptual framework; Narrative history","score_opus":0.020757351768840193,"score_gpt":0.35017000774101864,"score_spread":0.32941265597217845,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414090244","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9581114,0.001015644,0.003688013,0.008774722,0.0003899356,0.00014511049,0.0000065549657,0.0000093726185,0.027859226],"genre_scores_gemma":[0.9966879,0.0012525647,0.00016198303,0.0007033791,0.00008087049,3.0592182e-7,0.0000011977384,0.0000012478874,0.0011105499],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9986748,0.00026040492,0.00051927916,0.000052366526,0.00037357793,0.00011957886],"domain_scores_gemma":[0.99794734,0.00053873577,0.00063583895,0.00008043016,0.00072646857,0.00007117882],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012752034,0.000084755724,0.00037337875,0.00022388925,0.00032364854,0.00015405809,0.00014893664,0.00008293245,0.00011357911],"category_scores_gemma":[0.0022326247,0.00004239175,0.00022667932,0.0013969097,0.00064370944,0.00056749396,0.000020635209,0.00024258358,8.4360784e-7],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003623113,0.0000629463,0.006877324,0.00002124196,0.003173225,0.0000015682482,0.6785027,0.0015525427,0.000015091452,0.29505163,0.0075318213,0.0068476577],"study_design_scores_gemma":[0.0013224698,0.00011708697,0.038575657,0.00024754796,0.002967822,0.000004861924,0.8852651,0.00286389,0.0001052245,0.045503158,0.022830449,0.00019674073],"about_ca_topic_score_codex":0.000112235655,"about_ca_topic_score_gemma":0.00016489376,"teacher_disagreement_score":0.24954847,"about_ca_system_score_codex":0.00004853206,"about_ca_system_score_gemma":0.00015129951,"threshold_uncertainty_score":0.26728216},"labels":[],"label_agreement":null},{"id":"W4414090284","doi":"10.22148/001c.133923","title":"Beyond Computational Formalism or, Architecture Matters","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Semantic Web and Ontologies","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Dartmouth College","keywords":"Formalism (music); Architecture; Calculus (dental); Representation (politics)","score_opus":0.017236449138540406,"score_gpt":0.2841832181358266,"score_spread":0.2669467689972862,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414090284","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039463233,0.0003706139,0.8975532,0.059008617,0.0009961753,0.000057480614,0.0000017293571,0.000038271923,0.0025106515],"genre_scores_gemma":[0.78793395,0.00006135253,0.19621558,0.012735725,0.00021333183,5.649136e-7,0.0000022054996,0.0000043809787,0.0028329387],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9990706,0.00002900372,0.00037299452,0.00008947541,0.00028411954,0.00015382623],"domain_scores_gemma":[0.999184,0.00012979869,0.00022362948,0.000117025476,0.00029154235,0.000053995092],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00016701502,0.00010021411,0.00021645435,0.00015515416,0.00008444569,0.00017382493,0.00064946053,0.000044179193,0.000012555517],"category_scores_gemma":[0.00006710748,0.000054445605,0.00014305457,0.00035618458,0.000053065793,0.00039602892,0.00010077467,0.00019482331,0.000004744191],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021836063,0.0002457591,0.0031181641,0.00018420916,0.0010340895,0.00069211435,0.0062918626,0.1790002,0.00061699224,0.3331897,0.3952655,0.080143064],"study_design_scores_gemma":[0.004110254,0.00081675925,0.03841488,0.0005539215,0.00039983387,0.0038778083,0.003158294,0.22358127,0.0019592282,0.58556473,0.13662596,0.0009370414],"about_ca_topic_score_codex":0.0000037105913,"about_ca_topic_score_gemma":0.000008200259,"teacher_disagreement_score":0.74847066,"about_ca_system_score_codex":0.000037614023,"about_ca_system_score_gemma":0.000111247275,"threshold_uncertainty_score":0.22202274},"labels":[],"label_agreement":null},{"id":"W4414110760","doi":"10.22148/001c.143993","title":"Computation and Form, Reconsidered","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Dartmouth College","keywords":"Generative grammar; Sketch; Comics; Computation; Poetry; Key (lock)","score_opus":0.07278789245797943,"score_gpt":0.279315459305353,"score_spread":0.20652756684737358,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414110760","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.79350966,0.0005853457,0.00007819128,0.0016670763,0.00056099636,0.00004215198,0.0000061532983,0.000010617277,0.2035398],"genre_scores_gemma":[0.9808836,0.000046542533,0.00011630438,0.00041170098,0.00019698667,1.3700813e-7,0.0000018673368,0.0000023514228,0.018340502],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994934,0.000011369457,0.00028224473,0.000040857227,0.00009723686,0.00007488852],"domain_scores_gemma":[0.99937755,0.000049070903,0.00015010114,0.000027568009,0.0003620287,0.00003365623],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000113557595,0.00006265746,0.00014546496,0.000080265636,0.00014008734,0.00054007646,0.000056770914,0.000017500784,0.00010998535],"category_scores_gemma":[0.00004847724,0.0000402485,0.00007488298,0.00002260398,0.00009631884,0.0006531278,0.000016520944,0.00011331394,0.0000028048303],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007346784,0.00007600769,0.00072687754,0.0000883716,0.0003230481,0.000029994644,0.02098086,0.00009256559,0.00003625135,0.87182266,0.08082894,0.024920952],"study_design_scores_gemma":[0.0012221795,0.00038962296,0.0018668156,0.00033904114,0.0002868406,0.00011241101,0.049212646,0.0012515547,0.00008691332,0.28768706,0.65725946,0.0002854189],"about_ca_topic_score_codex":0.0000068606896,"about_ca_topic_score_gemma":0.00011447092,"teacher_disagreement_score":0.5841356,"about_ca_system_score_codex":0.00002096142,"about_ca_system_score_gemma":0.000021419273,"threshold_uncertainty_score":0.520797},"labels":[],"label_agreement":null},{"id":"W4415033376","doi":"10.22148/001c.144825","title":"Generative Aesthetics: On formal stuckness in AI verse","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Generative grammar; Feature (linguistics); Interpretation (philosophy); Set (abstract data type); Natural language","score_opus":0.04955622214745402,"score_gpt":0.28333752314994526,"score_spread":0.23378130100249123,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415033376","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7960294,0.00017017507,0.00002229157,0.0024741304,0.0007945702,0.00006655348,0.000012052299,0.000008372214,0.20042247],"genre_scores_gemma":[0.97469974,0.000033069402,0.000020729296,0.0016433161,0.00033715597,6.731736e-7,0.0000027943343,0.0000047283884,0.023257824],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99907506,0.000029457453,0.00041900834,0.00007002904,0.00022485669,0.00018158783],"domain_scores_gemma":[0.9992909,0.00003398879,0.00015622911,0.00007278791,0.00040034982,0.000045769037],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019774615,0.00012153201,0.00023811437,0.00018575137,0.0001534119,0.00050759665,0.00018743331,0.000032805805,0.0002530184],"category_scores_gemma":[0.000041002164,0.00007670695,0.00015603342,0.000058055357,0.00011940035,0.0009928554,0.000029108995,0.00033609662,0.000016534104],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016624105,0.000199555,0.00030816862,0.000022083956,0.00009921645,0.00009020652,0.017782627,0.0007733041,0.000009545298,0.9607886,0.015866278,0.003894195],"study_design_scores_gemma":[0.002173615,0.0011328991,0.001127562,0.00053339737,0.00015308084,0.0000345022,0.075826675,0.0009091836,0.0002792363,0.03517308,0.88224375,0.00041303426],"about_ca_topic_score_codex":0.000011817566,"about_ca_topic_score_gemma":0.0004317642,"teacher_disagreement_score":0.9256155,"about_ca_system_score_codex":0.00008768178,"about_ca_system_score_gemma":0.000050188173,"threshold_uncertainty_score":0.48947665},"labels":[],"label_agreement":null},{"id":"W4415058363","doi":"10.22148/001c.143615","title":"Pranks and the Viral Canon: On Top-Creator Content from YouTube to TikTok","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social Media and Politics","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"University of Illinois at Urbana-Champaign","keywords":"Content analysis; Content (measure theory); The Internet; Social media; Context (archaeology)","score_opus":0.042192918308567647,"score_gpt":0.3408522479251427,"score_spread":0.29865932961657504,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415058363","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9427764,0.00051614345,0.00002996405,0.04482772,0.0011982432,0.00013235903,0.000012295428,0.000006792491,0.0105001135],"genre_scores_gemma":[0.98977095,0.0002441808,0.000108853645,0.00409838,0.001284732,0.0000012832163,7.6268003e-7,0.0000028242528,0.0044880584],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989206,0.00016595585,0.0002890018,0.000071862516,0.0003670947,0.00018548095],"domain_scores_gemma":[0.99881583,0.0004741549,0.00014367058,0.00007005949,0.000324711,0.00017157393],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003107197,0.00008601326,0.00026018804,0.000044417087,0.00028134757,0.00013265968,0.00021482754,0.000076976416,0.000035210593],"category_scores_gemma":[0.0011847519,0.000045041757,0.00013241102,0.00017421112,0.00028592153,0.00009591377,0.00002629296,0.00022336631,0.0000066981665],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013233052,0.00017793447,0.043559063,0.000021303924,0.0011123458,0.00007256289,0.25922856,0.00007310071,0.00081627426,0.58226216,0.09028218,0.0210712],"study_design_scores_gemma":[0.00465336,0.00046684855,0.013976525,0.0003759655,0.0010940193,0.0000071421955,0.25333914,0.000111738525,0.0018763543,0.03684795,0.68684137,0.00040959005],"about_ca_topic_score_codex":0.0020957873,"about_ca_topic_score_gemma":0.0007491487,"teacher_disagreement_score":0.5965592,"about_ca_system_score_codex":0.00012788843,"about_ca_system_score_gemma":0.00020051036,"threshold_uncertainty_score":0.31682175},"labels":[],"label_agreement":null},{"id":"W4415166832","doi":"10.22148/001c.144827","title":"Seeing Formalism or Formal Viewing: Computational Formalism for the Analysis of Visual Media Forms and Contexts","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Image Retrieval and Classification Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Formalism (music); Visual media; Computational model; Visualization; Context (archaeology)","score_opus":0.03178069413648718,"score_gpt":0.3372798080262438,"score_spread":0.3054991138897566,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415166832","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018223085,0.0006759141,0.9786339,0.0020972495,0.00012212404,0.00014245749,0.000009908583,0.00001654766,0.00007882437],"genre_scores_gemma":[0.97911793,0.00026638186,0.019957768,0.00040612562,0.000047344314,0.000003482857,0.000007402322,0.0000030867484,0.00019046149],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984816,0.00002665091,0.0008048408,0.00010866062,0.00038073008,0.00019755222],"domain_scores_gemma":[0.99741846,0.0006070676,0.00068762474,0.00012501456,0.001097357,0.0000644946],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00081048446,0.00013051355,0.00041747995,0.000336025,0.00022232308,0.00017078426,0.0005261571,0.000063763386,0.000005295971],"category_scores_gemma":[0.00022590523,0.00006308469,0.00034983992,0.0011346401,0.00010398596,0.0013352206,0.00012375975,0.00014860263,1.5731283e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00081793155,0.0003997624,0.0022750907,0.00046090578,0.008532737,0.00002096382,0.015536633,0.0069261817,0.0018120381,0.49453223,0.0047115562,0.46397397],"study_design_scores_gemma":[0.0010512945,0.0002854892,0.013339093,0.00010851175,0.0015643089,0.00006471936,0.001330026,0.9600583,0.0045994287,0.009986231,0.0074167904,0.00019585161],"about_ca_topic_score_codex":0.000003830553,"about_ca_topic_score_gemma":0.000008206093,"teacher_disagreement_score":0.9608949,"about_ca_system_score_codex":0.00004469651,"about_ca_system_score_gemma":0.000116729476,"threshold_uncertainty_score":0.25725192},"labels":[],"label_agreement":null},{"id":"W4415400747","doi":"10.22148/001c.144809","title":"Tug-of-War of Emotion: Measuring and Modeling Sentiment Cycles in Chinese-Language Pop Song Lyrics, 1967-2023","year":2025,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Music History and Culture","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Popular music; Term (time); Sound recording and reproduction","score_opus":0.03933005308812216,"score_gpt":0.26311685740249796,"score_spread":0.2237868043143758,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415400747","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9895861,0.0027892946,0.00008950006,0.00017526458,0.00030122258,0.000043945733,0.000009391348,0.0000039521283,0.0070013283],"genre_scores_gemma":[0.9967965,0.0001086788,0.00013582042,0.00004325478,0.00024874485,2.458496e-7,0.0000028482648,0.0000038006358,0.0026600999],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9990098,0.000033361124,0.0005546318,0.00007684559,0.00022405248,0.000101298705],"domain_scores_gemma":[0.9993061,0.000017537552,0.00029126887,0.000068552465,0.00027803224,0.00003852706],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00024250861,0.00011046537,0.00033127048,0.00014644892,0.00010070566,0.00003056195,0.00010507439,0.00004066752,0.00016701722],"category_scores_gemma":[0.000045459652,0.00007065781,0.00014021594,0.00009349575,0.00008076178,0.00021584435,0.000027712058,0.00018652726,7.26417e-7],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000532827,0.0010525245,0.013384294,0.0024992405,0.0017782851,0.00020149912,0.7996908,0.09140515,0.016256219,0.048960242,0.015315822,0.008923104],"study_design_scores_gemma":[0.0080449,0.0008505127,0.008837206,0.008295295,0.002104339,0.00019390095,0.6996396,0.20997286,0.0035625726,0.011195713,0.04556277,0.0017402888],"about_ca_topic_score_codex":0.000073753356,"about_ca_topic_score_gemma":0.0004979389,"teacher_disagreement_score":0.118567705,"about_ca_system_score_codex":0.000048153343,"about_ca_system_score_gemma":0.000031371623,"threshold_uncertainty_score":0.2881342},"labels":[],"label_agreement":null},{"id":"W7126085874","doi":"10.22148/jca.769","title":"Heritage site-seeing through the visitor’s lens on Instagram","year":2022,"lang":"","type":"article","venue":"Journal of Cultural Analytics","topic":"Museums and Cultural Heritage","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Lens (geology); Photography; Reflection (computer programming); Social media; Through-the-lens metering","score_opus":0.055713432268587375,"score_gpt":0.2678691530227232,"score_spread":0.2121557207541358,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7126085874","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.86143225,0.009178848,0.000016484792,0.02022988,0.008632423,0.0005524235,0.00039135435,0.000050142018,0.09951621],"genre_scores_gemma":[0.96214396,0.0012255296,0.000049934755,0.0032657578,0.004645321,0.0000045103134,0.00001618426,0.00004326562,0.02860555],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9955521,0.00039351292,0.0015020029,0.0003090855,0.0016003136,0.0006429892],"domain_scores_gemma":[0.99701583,0.00018302043,0.0015700014,0.0003909458,0.00068069453,0.00015949641],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0006563003,0.0005461078,0.00080145,0.00009345927,0.0038941724,0.0012222963,0.0009796489,0.000096775286,0.00841816],"category_scores_gemma":[0.00007780576,0.00029066467,0.0011748204,0.0002936809,0.00039426354,0.001262573,0.000324587,0.0021597855,0.000077142555],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009921938,0.0015731531,0.00019378612,0.0002253369,0.0020187558,0.0010501122,0.372764,0.02298684,0.00094586733,0.2906317,0.30136102,0.0052572475],"study_design_scores_gemma":[0.0007181457,0.0017973961,0.00010462543,0.00012704778,0.00041515965,0.00020594422,0.14251062,0.000641371,0.000036355577,0.0004941821,0.8525261,0.00042303046],"about_ca_topic_score_codex":0.00020775107,"about_ca_topic_score_gemma":0.00021982365,"teacher_disagreement_score":0.5511651,"about_ca_system_score_codex":0.00041812556,"about_ca_system_score_gemma":0.00012442203,"threshold_uncertainty_score":0.9999545},"labels":[],"label_agreement":null},{"id":"W7126089735","doi":"10.22148/jca.1186","title":"The Identity Issue","year":2018,"lang":"","type":"article","venue":"Journal of Cultural Analytics","topic":"Identity, Memory, and Therapy","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Identity (music); Ask price; Social identity theory; Cultural identity; Field (mathematics)","score_opus":0.041114354418516505,"score_gpt":0.3870495380924138,"score_spread":0.3459351836738973,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7126089735","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.83821976,0.026758688,0.0007191993,0.024838237,0.053464957,0.00035977105,0.000018687555,0.00002482954,0.05559587],"genre_scores_gemma":[0.9033661,0.009981415,0.000100471625,0.001003927,0.019080384,5.6097787e-7,9.78196e-7,0.000029721969,0.06643644],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9961478,0.00045728748,0.0015258488,0.0002359786,0.0009817077,0.0006513411],"domain_scores_gemma":[0.9948315,0.00018437697,0.0017901305,0.00053312397,0.0023596403,0.00030123015],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0020543754,0.00034637615,0.000622509,0.00015578927,0.0011071743,0.0008737502,0.0014155112,0.00027275938,0.00361712],"category_scores_gemma":[0.0003464748,0.00019598728,0.00080592197,0.0006624621,0.0010298188,0.0008284124,0.000099840465,0.0009453875,0.0021408396],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017332277,0.0011289517,0.0059654657,0.00008626619,0.0062420783,0.000516045,0.023873964,0.00002855685,0.0030441186,0.031733565,0.89609385,0.029553913],"study_design_scores_gemma":[0.0019197242,0.0016602963,0.015482362,0.00011840353,0.00092017744,0.0008902869,0.011037769,0.00016107113,0.00068959466,0.010758272,0.9559742,0.00038782254],"about_ca_topic_score_codex":0.000095082454,"about_ca_topic_score_gemma":0.00057627086,"teacher_disagreement_score":0.06514634,"about_ca_system_score_codex":0.00016474849,"about_ca_system_score_gemma":0.00015294168,"threshold_uncertainty_score":0.9986361},"labels":[],"label_agreement":null},{"id":"W7126108520","doi":"10.22148/jca.1197","title":"Archival Circulation on the Web: The Vine-Tweets Dataset","year":2019,"lang":"","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Circulation (fluid dynamics); Feature (linguistics); Component (thermodynamics); Identification (biology)","score_opus":0.083460553295701,"score_gpt":0.2694264287423221,"score_spread":0.18596587544662108,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7126108520","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.87859213,0.00042765663,0.0000017294974,0.011149261,0.0019468968,0.0002926858,0.0009840714,0.0000071380864,0.106598414],"genre_scores_gemma":[0.97792417,0.0002703981,0.0000026638938,0.0022062315,0.002677774,7.4658857e-7,0.00011239404,0.000021052954,0.01678459],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971585,0.00021694174,0.0010008157,0.0001826911,0.0010548965,0.00038614313],"domain_scores_gemma":[0.99726593,0.00055987085,0.0010321264,0.00047711047,0.0005366077,0.00012837414],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0009696147,0.0003427456,0.0004446637,0.00010328971,0.00086381857,0.0031617149,0.0010031768,0.00005960169,0.0043433006],"category_scores_gemma":[0.0001577623,0.00014180137,0.00055117294,0.00009518594,0.0004382276,0.0012633775,0.00012681172,0.0011479736,0.00045304385],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034640502,0.0003204598,0.0006034528,0.00008826782,0.0010030518,0.000057190297,0.015177206,0.0022740955,0.00012812423,0.68082184,0.29726517,0.0019147382],"study_design_scores_gemma":[0.0006400134,0.00081492006,0.0035831616,0.00036302826,0.0004536644,0.000121088364,0.02154258,0.0025769926,0.000017717655,0.0068307808,0.9626967,0.00035939398],"about_ca_topic_score_codex":0.000017383118,"about_ca_topic_score_gemma":0.00006211125,"teacher_disagreement_score":0.6739911,"about_ca_system_score_codex":0.00009860913,"about_ca_system_score_gemma":0.000112584195,"threshold_uncertainty_score":0.9978731},"labels":[],"label_agreement":null},{"id":"W7140893723","doi":"10.22148/11039","title":"Beyond Poet Voice: Sampling the (Non-) Performance Styles of 100 American Poets","year":2018,"lang":"","type":"article","venue":"Journal of Cultural Analytics","topic":"Poetry Analysis and Criticism","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"University of California, Davis; Northwestern University","keywords":"Sampling (signal processing); Style (visual arts); Narrative; Sample (material)","score_opus":0.04519147385284388,"score_gpt":0.3072623438947864,"score_spread":0.2620708700419425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7140893723","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97769433,0.0016927314,0.00013775636,0.002791419,0.0008891304,0.0000883811,0.000059098104,0.00000590289,0.016641276],"genre_scores_gemma":[0.98666817,0.002398456,0.0002053277,0.0004901004,0.006107804,5.868675e-7,0.0000048475713,0.00002372791,0.0041010045],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.99618095,0.00014324709,0.0018848627,0.00023141516,0.0009902122,0.0005693393],"domain_scores_gemma":[0.9936514,0.0002807549,0.0025563412,0.00038775126,0.0028918097,0.00023193158],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0009885905,0.0004287967,0.0011358592,0.00024659673,0.0010069803,0.00054571265,0.0009078615,0.00008991273,0.0011134649],"category_scores_gemma":[0.00019508762,0.00023190632,0.0009478083,0.00044244836,0.002914372,0.0008458262,0.00013116206,0.0007383138,0.000056085642],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00326023,0.0045679584,0.017500712,0.0020907794,0.025519209,0.00021245149,0.49818957,0.0067141014,0.016321763,0.17142928,0.135432,0.11876194],"study_design_scores_gemma":[0.0024621051,0.010024031,0.023497453,0.0021739185,0.011023054,0.00044598704,0.40283033,0.030865448,0.0073848986,0.0030364862,0.5042728,0.0019835029],"about_ca_topic_score_codex":0.00014631545,"about_ca_topic_score_gemma":0.00021977621,"teacher_disagreement_score":0.36884078,"about_ca_system_score_codex":0.00009327978,"about_ca_system_score_gemma":0.00018300614,"threshold_uncertainty_score":0.99979967},"labels":[],"label_agreement":null}]}