{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":199,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":199,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"query_hash":"517981b04501","filters":{"venue":"Journal of Cultural Analytics"}},"results":[{"id":"W3085332162","doi":"10.22148/001c.17212","title":"Can GPT-3 Pass a Writer’s Turing Test?","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computability, Logic, AI Algorithms","field":"Computer Science","cited_by":205,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Computer science; Turing test; Heuristic; Field (mathematics); Grammar; Test (biology); Artificial intelligence; Scale (ratio); Language model; Natural language processing; Natural language; Turing; Language identification; Programming language; Linguistics; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.04380876233099833,"gpt":0.2608145662498129,"spread":0.2170058039188145,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002836183,0.0001964754,0.0004022609,0.00007608067,0.00009416886,0.0003850571,0.001358158,0.00006208047,0.00001681297],"category_scores_gemma":[0.0004986395,0.0001328432,0.0002944052,0.0006475557,0.00005331593,0.0005963298,0.0003543173,0.0004568165,0.00001314945],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001259191,"about_ca_system_score_gemma":0.0001116695,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001034533,"about_ca_topic_score_gemma":0.000008471317,"domain_scores_codex":[0.9981041,0.00006293324,0.0006687034,0.000245711,0.0006149987,0.0003035785],"domain_scores_gemma":[0.998153,0.0001741513,0.000439406,0.0002586436,0.0005841444,0.0003906303],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002274299,0.002864481,0.09566623,0.0009391286,0.002219222,0.00859012,0.09229141,0.1251046,0.04673479,0.02430968,0.1229658,0.4780872],"study_design_scores_gemma":[0.001723491,0.001855343,0.02270716,0.0001519001,0.0001801423,0.001491966,0.001415033,0.9344832,0.003535417,0.01067262,0.02080879,0.0009749455],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4494256,0.001013878,0.371239,0.17415,0.001887353,0.0003324081,0.0000171778,0.0003009899,0.001633574],"genre_scores_gemma":[0.951152,0.00003614496,0.04631538,0.001524865,0.000869702,3.979804e-7,5.968562e-7,0.000007576295,0.00009329316],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8093786,"threshold_uncertainty_score":0.5417189,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2401667852","doi":"10.22148/16.004","title":"The Science of Culture? Social Computing, Digital Humanities and Cultural Analytics","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":110,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Cultural analytics; Analytics; Data science; Digital humanities; Digital culture; Visualization; Cultural analysis; Computer science; Cultural studies; Software; Sociology; World Wide Web; Media studies; Social science; Semantic analytics; Anthropology; The Internet; Artificial intelligence","retraction":null,"screen_n_in":null,"score":{"opus":0.07116406461840949,"gpt":0.2735665621756626,"spread":0.2024024975572531,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication"],"consensus_categories":["sts"],"category_scores_codex":[0.0004596999,0.0002070568,0.0003598754,0.000108761,0.001312007,0.002492445,0.0005655771,0.00004058657,0.00006139894],"category_scores_gemma":[0.0002671843,0.0000859405,0.0002592265,0.00009698617,0.003635237,0.002609588,0.0001465991,0.0002185974,0.000005388834],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000863317,"about_ca_system_score_gemma":0.00007765137,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000007937118,"about_ca_topic_score_gemma":0.00005976284,"domain_scores_codex":[0.9980332,0.00002541162,0.000728008,0.0001359286,0.0007334893,0.0003439759],"domain_scores_gemma":[0.9970212,0.0001418731,0.0007433516,0.0001105045,0.001896048,0.00008703383],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001072078,0.0001144565,0.0005378423,0.00004996869,0.0003460145,0.00001945055,0.06197092,0.000009158838,0.0006709402,0.898228,0.02637475,0.01157122],"study_design_scores_gemma":[0.00196696,0.001346256,0.002465976,0.0004858896,0.0004440163,0.0002702865,0.2025598,0.0001946584,0.0009751098,0.03009796,0.7582834,0.0009096828],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9331735,0.0006360142,0.00001526693,0.0009417039,0.0005412005,0.00007050475,0.00006287686,0.00001764921,0.06454127],"genre_scores_gemma":[0.9799966,0.0001675191,0.00002028729,0.00008042079,0.001041716,2.006889e-7,0.000001608702,0.000009571347,0.01868205],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8681301,"threshold_uncertainty_score":0.9999881,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2786709099","doi":"10.22148/16.019","title":"The Transformation of Gender in English-Language Fiction","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Literature: history, themes, analysis","field":"Arts and Humanities","cited_by":95,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"History; Literature; English language; Linguistics; Psychology; Sociology; Gender studies; Art; Philosophy","retraction":null,"screen_n_in":null,"score":{"opus":0.02661860338798629,"gpt":0.2437616383731717,"spread":0.2171430349851854,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003451349,0.00007549121,0.0001669631,0.0001192203,0.0001501548,0.0001024258,0.0001438517,0.00003280108,0.0002134893],"category_scores_gemma":[0.00006703194,0.00003946801,0.0001691065,0.00008753468,0.0001705854,0.0004824158,0.000005239082,0.0001628657,0.000003427253],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000721106,"about_ca_system_score_gemma":0.00002025305,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002822738,"about_ca_topic_score_gemma":0.0005922425,"domain_scores_codex":[0.9990587,0.00004553299,0.0004991266,0.00004341007,0.000253944,0.00009924905],"domain_scores_gemma":[0.9987434,0.00003360162,0.0003590022,0.00008418247,0.0007508374,0.00002899056],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00008963648,0.00005869952,0.00007781186,0.00002840377,0.000224963,0.000003757268,0.961338,0.000125314,0.0006189604,0.02141788,0.01439361,0.001622981],"study_design_scores_gemma":[0.000952005,0.0003408342,0.000407484,0.0000945302,0.0004669773,0.00001939208,0.23368,0.00320827,0.001719154,0.001536226,0.7573453,0.0002297959],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9232108,0.002441172,0.0001804052,0.0005103002,0.001542603,0.00008754307,0.00001408788,0.0000143424,0.07199872],"genre_scores_gemma":[0.9964088,0.0001016395,0.00003426503,0.00005728977,0.001310515,2.899423e-7,0.000003629371,0.000004259028,0.002079379],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7429517,"threshold_uncertainty_score":0.2337558,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3153865529","doi":"10.22148/001c.22221","title":"The Goodreads “Classics”: A Computational Study of Readers, Amazon, and Crowdsourced Amateur Criticism","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Games and Media","field":"Social Sciences","cited_by":55,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Amateur; Publishing; Criticism; Literary criticism; Reading (process); Point (geometry); Literature; Computer science; History; Sociology; Art; Linguistics; Philosophy","retraction":null,"screen_n_in":null,"score":{"opus":0.03654861629295647,"gpt":0.3283822056129931,"spread":0.2918335893200366,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004424581,0.00007111691,0.0002076032,0.00002706506,0.0001583733,0.0002163855,0.0001451076,0.00003635557,0.00001095786],"category_scores_gemma":[0.0005549587,0.00004191069,0.00009058225,0.000222946,0.0002681639,0.0002103384,0.00003488514,0.0001422533,9.760194e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003093059,"about_ca_system_score_gemma":0.0001548105,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002344007,"about_ca_topic_score_gemma":0.0004097253,"domain_scores_codex":[0.9986924,0.0001165701,0.0004003541,0.0000708513,0.0005691059,0.0001507402],"domain_scores_gemma":[0.9983179,0.0002752552,0.0002807009,0.00005811903,0.0009363429,0.0001316936],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.0002970296,0.003645896,0.03596193,0.0001466395,0.002174828,0.0008235689,0.7062479,0.005351393,0.0007994872,0.08301898,0.03957188,0.1219605],"study_design_scores_gemma":[0.001693184,0.0005959484,0.02170076,0.0001093206,0.0002960827,0.0001106823,0.8038211,0.0006226653,0.0001011742,0.004962782,0.1657831,0.0002031885],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9874661,0.0004687541,0.00006165027,0.004212162,0.0001983781,0.00006309587,0.000003722541,0.000004797103,0.007521317],"genre_scores_gemma":[0.9910288,0.000230426,0.0002264046,0.0001210405,0.0002008564,2.919718e-7,0.000001032592,0.000003892486,0.008187277],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1262112,"threshold_uncertainty_score":0.208661,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2942659792","doi":"10.22148/16.036","title":"Data Is the New What? Popular Metaphors & Professional Ethics in Emerging Data Culture","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":50,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Big data; Data science; Field (mathematics); Data collection; Computer science; Conjunction (astronomy); Engineering ethics; Knowledge management; Sociology; Social science; Data mining; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.2958110940572425,"gpt":0.4945049306874361,"spread":0.1986938366301936,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["research_integrity"],"consensus_categories":[],"category_scores_codex":[0.00694487,0.0001773406,0.0004021524,0.00007862064,0.0006283489,0.0008953108,0.00312831,0.0003929548,0.0003128769],"category_scores_gemma":[0.00474091,0.00009664485,0.0001533131,0.0006598193,0.0002306752,0.004951225,0.0006152902,0.002567883,0.00005551444],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001132934,"about_ca_system_score_gemma":0.001227621,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001500716,"about_ca_topic_score_gemma":0.01955016,"domain_scores_codex":[0.996368,0.0005570363,0.0006779697,0.0002618216,0.001738906,0.0003962567],"domain_scores_gemma":[0.9969367,0.0003458056,0.0006977452,0.0008510137,0.0008987014,0.000270069],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00005642416,0.0001418081,0.004452419,0.00004573833,0.0004262293,0.00001775826,0.3111159,0.0002385491,0.0003119674,0.01363021,0.6670938,0.00246919],"study_design_scores_gemma":[0.000277,0.00001859257,0.0004106037,0.0002801595,0.0001430312,0.000007259849,0.08163743,0.0001787588,0.000004122609,0.008923724,0.9079359,0.0001833869],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.2551705,0.008317033,0.0001471279,0.7281649,0.004839561,0.0004819777,0.0001627062,0.00002400725,0.002692232],"genre_scores_gemma":[0.7531652,0.07811005,0.01279268,0.03250051,0.008068154,0.000001069273,0.0007115022,0.00007862686,0.1145722],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6956643,"threshold_uncertainty_score":0.9997332,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2587326830","doi":"10.22148/16.006","title":"There Will Be Numbers","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":40,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Realm; TRACE (psycholinguistics); Mythology; The Internet; Point (geometry); Social media; Natural (archaeology); Computer science; Sociology; Epistemology; Data science; History; World Wide Web; Literature; Art; Philosophy; Mathematics; Linguistics; Archaeology","retraction":null,"screen_n_in":null,"score":{"opus":0.06962991266336825,"gpt":0.2568047155385784,"spread":0.1871748028752102,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0001454455,0.0001118227,0.0001958426,0.00004746917,0.0001346691,0.0004037806,0.0002233098,0.00002754248,0.003614834],"category_scores_gemma":[0.00006138227,0.00004921652,0.0002336105,0.00002036461,0.0001936367,0.001639725,0.00002318813,0.0001229751,0.0000437028],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004956303,"about_ca_system_score_gemma":0.00002342372,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001018907,"about_ca_topic_score_gemma":0.000164396,"domain_scores_codex":[0.9991056,0.00001882758,0.0003501653,0.00006206395,0.0002963612,0.0001669638],"domain_scores_gemma":[0.9990394,0.00005634878,0.0002623926,0.0000834101,0.000462048,0.00009642711],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00008815996,0.0001200108,0.0007632457,0.000022619,0.0003583076,0.0001207185,0.01663945,0.000006902297,0.0002130852,0.8136779,0.1588738,0.00911576],"study_design_scores_gemma":[0.0003599781,0.0001781023,0.0001278912,0.00009544968,0.00005920735,0.00004977688,0.009419564,0.000002509773,0.00006218038,0.01024414,0.9792648,0.000136433],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6639833,0.0002986573,0.00001562512,0.005528182,0.0007525655,0.00003614987,0.00003228635,0.00001903695,0.3293342],"genre_scores_gemma":[0.8962232,0.00008611551,0.0000217076,0.0004615591,0.001087682,1.902518e-7,8.404369e-7,0.000008511455,0.1021102],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8203909,"threshold_uncertainty_score":0.997296,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2979474259","doi":"10.22148/001c.13147","title":"NovelTM Datasets for English-Language Fiction, 1700-2009","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":37,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"English language; Computer science; Non-fiction; Resilience (materials science); Fragility; History; Linguistics; Literature; Art; Philosophy","retraction":null,"screen_n_in":null,"score":{"opus":0.07261985779052788,"gpt":0.2707456352499021,"spread":0.1981257774593742,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001407966,0.0001322877,0.0002479926,0.00004422431,0.0001903499,0.0007363522,0.0002566087,0.00003149056,0.0007397504],"category_scores_gemma":[0.0003013466,0.00008750244,0.0002374618,0.00003861452,0.00008070723,0.001175287,0.00002877543,0.0002096326,0.00001332607],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002667174,"about_ca_system_score_gemma":0.00002738194,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000006058723,"about_ca_topic_score_gemma":0.00003945186,"domain_scores_codex":[0.999015,0.00001314211,0.0004520359,0.00009495837,0.0002518944,0.0001729216],"domain_scores_gemma":[0.9988045,0.0000550131,0.0003005854,0.00007950664,0.0006070635,0.0001533053],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001726288,0.0001097899,0.0000347509,0.00008681264,0.0002468129,0.00004216115,0.1139312,0.0001209023,0.000116434,0.06238067,0.8211472,0.00161064],"study_design_scores_gemma":[0.0005261612,0.0003828309,0.00001570277,0.00002939787,0.0001180291,0.00001193488,0.04807862,0.0002390538,0.00009164299,0.0003934756,0.9499643,0.0001488402],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7058294,0.00361364,0.0007659327,0.01116564,0.005655103,0.0007503128,0.007693315,0.0002152811,0.2643114],"genre_scores_gemma":[0.9878774,0.00003071578,0.0001609492,0.001757169,0.005481763,9.614287e-7,0.000169152,0.0000143099,0.004507555],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.282048,"threshold_uncertainty_score":0.8099748,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2968217872","doi":"10.22148/16.044","title":"Difficult Heritage and the Complexities of Indigenous Data","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Cultural Heritage Management and Preservation","field":"Arts and Humanities","cited_by":31,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Indigenous; Geography; History; Data science; Computer science; Biology","retraction":null,"screen_n_in":null,"score":{"opus":0.1595948908674861,"gpt":0.2662263728683407,"spread":0.1066314820008546,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003113819,0.00008567885,0.0002546264,0.00003133334,0.0001256622,0.0001853724,0.0003833132,0.00001762418,0.0003896398],"category_scores_gemma":[0.0000400188,0.00003844487,0.000067689,0.00004246695,0.000230537,0.0006244159,0.0001385762,0.0001205958,0.000005962293],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000008801249,"about_ca_system_score_gemma":0.000008808009,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001243796,"about_ca_topic_score_gemma":0.0003654854,"domain_scores_codex":[0.9991918,0.00004374606,0.0003695282,0.00006534459,0.0002399571,0.00008966485],"domain_scores_gemma":[0.9989893,0.00008053652,0.0003841203,0.0001749368,0.0003475,0.00002354681],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0007522794,0.0001619731,0.005215339,0.0005380962,0.001237207,0.00001607669,0.2736377,0.0002329299,0.0005868173,0.6704766,0.04280115,0.004343797],"study_design_scores_gemma":[0.004944583,0.0005448488,0.008885056,0.0002691737,0.000767166,0.00005611449,0.2415433,0.006052597,0.0000754667,0.0119268,0.72448,0.0004548344],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9782867,0.001010022,0.00000573741,0.001301025,0.0002711361,0.0001546715,0.00003291737,0.000006078248,0.01893172],"genre_scores_gemma":[0.9885315,0.0001980778,0.0000766016,0.0001314208,0.0003739805,2.360529e-7,0.00002754268,0.000004106537,0.01065658],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6816789,"threshold_uncertainty_score":0.4266282,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2587380683","doi":"10.22148/16.010","title":"Understanding Gender and Character Agency in the 19th Century Novel","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Folklore, Mythology, and Literature Studies","field":"Arts and Humanities","cited_by":28,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Character (mathematics); Agency (philosophy); Late 19th century; Art; Literature; History; Aesthetics; Sociology; Social science; Period (music); Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.1580922718877203,"gpt":0.2791555401300548,"spread":0.1210632682423346,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002892657,0.0001036077,0.0001812943,0.00007696476,0.0001612455,0.0001374496,0.0001240934,0.00003103705,0.0001058695],"category_scores_gemma":[0.00004734064,0.00003794088,0.00008808819,0.00003534726,0.0002201263,0.0003276658,0.00002135095,0.0001690336,0.000002909922],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004033956,"about_ca_system_score_gemma":0.000009650196,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000002678692,"about_ca_topic_score_gemma":0.00004051939,"domain_scores_codex":[0.9992943,0.00003403384,0.0002818049,0.00007531548,0.0001633203,0.0001511691],"domain_scores_gemma":[0.999535,0.0000790983,0.0001753053,0.00006101422,0.0001184989,0.00003113386],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00008124495,0.0001104349,0.002238284,0.00003990128,0.000272806,0.00008841674,0.2700097,0.000001567508,0.001034269,0.7072186,0.01818811,0.0007166696],"study_design_scores_gemma":[0.00225291,0.0003560841,0.04369714,0.0003156576,0.0003211619,0.0004997908,0.1651381,0.00002711378,0.00002126198,0.01937018,0.7675559,0.0004446492],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8554792,0.007884832,0.0007133628,0.0237251,0.003287487,0.0002601486,0.00007732506,0.00002342793,0.1085491],"genre_scores_gemma":[0.9964519,0.0008546417,0.0000448312,0.000662625,0.0009843694,3.81368e-7,7.414299e-7,0.000004355186,0.000996227],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7493678,"threshold_uncertainty_score":0.1547184,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2500585446","doi":"10.22148/16.005","title":"The Life Cycles of Genres","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":27,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"University of Illinois at Urbana-Champaign","keywords":"Literature; Haven; Literary genre; Safe haven; Art; Aesthetics; History; Sociology; Linguistics; Philosophy; Mathematics; Economics","retraction":null,"screen_n_in":null,"score":{"opus":0.07790262750403741,"gpt":0.2603709562125727,"spread":0.1824683287085352,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002263413,0.00007676261,0.0001665838,0.00003367617,0.0001939569,0.0002879827,0.0002626073,0.00001730618,0.0003290946],"category_scores_gemma":[0.0002079199,0.00002690247,0.0002044005,0.00001667196,0.0003302664,0.0005722441,0.00002672882,0.00008296302,0.00001303052],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001496796,"about_ca_system_score_gemma":0.00003090607,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000006631425,"about_ca_topic_score_gemma":0.0001147733,"domain_scores_codex":[0.9991118,0.00002268568,0.0004319525,0.00003849696,0.0002711379,0.0001239214],"domain_scores_gemma":[0.9987503,0.0001642575,0.000383065,0.00008329774,0.0005498151,0.00006928417],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001167545,0.00006832477,0.0006379664,0.0000206173,0.0004000152,0.00001309866,0.008380999,0.00001028023,0.0003589914,0.9154825,0.05501331,0.01949721],"study_design_scores_gemma":[0.0004739097,0.0003432482,0.00122018,0.000154968,0.0001363043,0.00002544415,0.02746615,0.000009756922,0.0004230798,0.01973919,0.9498503,0.0001574515],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.934434,0.001412738,0.000008448503,0.002983621,0.0005438721,0.00003194161,0.00001963369,0.000006928275,0.06055879],"genre_scores_gemma":[0.9771803,0.0002970724,0.00001102532,0.0001143476,0.0007890319,1.749081e-7,2.461275e-7,0.00000494516,0.02160284],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8957433,"threshold_uncertainty_score":0.3603355,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3117881903","doi":"10.22148/001c.18509","title":"Representing Race and Ethnicity in American Fiction, 1789-1920","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Culinary Culture and Tourism","field":"Agricultural and Biological Sciences","cited_by":23,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Racialization; Ethnic group; Writ; Race (biology); Gender studies; Representation (politics); Sociology; The Imaginary; Historicity (philosophy); Psychology; Political science; Anthropology","retraction":null,"screen_n_in":null,"score":{"opus":0.04070182394424551,"gpt":0.2745896654440442,"spread":0.2338878414997987,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001842518,0.0001062511,0.0002917878,0.0000108785,0.0000828995,0.00006565958,0.0001474747,0.00004627462,0.00005514146],"category_scores_gemma":[0.0002028802,0.00003634899,0.0001197592,0.0005151688,0.00007024634,0.0002549664,0.00005792639,0.0003046737,0.000003291136],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001760407,"about_ca_system_score_gemma":0.000005721737,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001410021,"about_ca_topic_score_gemma":0.0001586041,"domain_scores_codex":[0.9990359,0.00006272483,0.0003907497,0.0001402691,0.0002141797,0.0001561589],"domain_scores_gemma":[0.9992278,0.00007380922,0.0003512904,0.00002239127,0.000162552,0.000162197],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0004958146,0.0003822781,0.4639484,0.00005295463,0.0002705437,0.000708763,0.009468979,0.001438952,0.3152303,0.0002823206,0.1030691,0.1046516],"study_design_scores_gemma":[0.0006813226,0.00128433,0.8074052,0.00007794282,0.000122158,0.0003373332,0.01209675,0.002694948,0.002176039,0.0003477321,0.172356,0.0004202088],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9764214,0.0003358263,0.000008225958,0.02258822,0.00004756748,0.00004124233,0.000002846379,0.00001021766,0.0005445046],"genre_scores_gemma":[0.9977159,0.0004314337,0.0001959779,0.0006826632,0.0007408099,2.58511e-7,0.000003330349,6.287959e-7,0.0002289913],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3434569,"threshold_uncertainty_score":0.1482269,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3183176260","doi":"10.22148/001c.25943","title":"The Measure of the Archive: The Robustness of Network Analysis in Early Modern Correspondence","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social Capital and Networks","field":"Social Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"Arts and Humanities Research Council","keywords":"Robustness (evolution); Computer science; Network analysis; Missing data; Transitive relation; Centrality; Data mining; Data loss; Data science; Statistics; Machine learning; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.02610744377729758,"gpt":0.2901900606362046,"spread":0.264082616858907,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001650195,0.000083087,0.0003150756,0.00003030322,0.0004418626,0.00007328807,0.0007037318,0.00007088108,0.00001774405],"category_scores_gemma":[0.0006229004,0.00003439725,0.0005260023,0.00214276,0.0005698553,0.0001282346,0.00007830772,0.0004027188,2.75725e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004975976,"about_ca_system_score_gemma":0.0002735317,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002520569,"about_ca_topic_score_gemma":0.01059212,"domain_scores_codex":[0.9975995,0.000710977,0.0005206108,0.00007541871,0.0008711899,0.000222299],"domain_scores_gemma":[0.9975554,0.00070678,0.000698081,0.0001673261,0.000817635,0.00005474663],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","study_design_scores_codex":[0.0003052507,0.0001336384,0.1168703,0.0000133867,0.001632907,0.00002272825,0.09613542,0.757977,0.0002252636,0.01352992,0.005571205,0.007583017],"study_design_scores_gemma":[0.001073261,0.0002023883,0.7190793,0.000604103,0.003928139,0.00002827354,0.1500738,0.07424117,0.0002994403,0.04171497,0.008234491,0.0005205991],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9900026,0.003648619,0.0006892757,0.003952825,0.0005162999,0.00007744514,0.000003696418,0.000002078118,0.001107122],"genre_scores_gemma":[0.9976147,0.0008816377,0.00005482859,0.00004961309,0.0004783775,5.227847e-7,3.332156e-7,0.000003068555,0.0009168881],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6837358,"threshold_uncertainty_score":0.5910653,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2586621921","doi":"10.22148/16.007","title":"Linguistic Markers of Status in Food Culture: Bourdieu’s Distinction in a Menu Corpus","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Culinary Culture and Tourism","field":"Agricultural and Biological Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Realm; TRACE (psycholinguistics); Mythology; Point (geometry); Sociology; The Internet; Social media; Linguistics; Epistemology; Literature; Computer science; History; Art; Philosophy; World Wide Web; Mathematics; Archaeology","retraction":null,"screen_n_in":null,"score":{"opus":0.01991263327032837,"gpt":0.2467437038893575,"spread":0.2268310706190292,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003248312,0.0001602506,0.0003875049,0.00004456798,0.00004077346,0.00002693654,0.0001968183,0.0001127982,0.00008867102],"category_scores_gemma":[0.0004054374,0.00004620126,0.0002050098,0.0004969817,0.00007345817,0.0002071835,0.00003775466,0.000208172,0.000003389313],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001244473,"about_ca_system_score_gemma":0.00002164763,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001146984,"about_ca_topic_score_gemma":0.001903105,"domain_scores_codex":[0.9984204,0.00008667302,0.0007203835,0.0001496831,0.0003308857,0.0002920165],"domain_scores_gemma":[0.9988835,0.0001217965,0.0005322655,0.00004324996,0.0002902282,0.0001289758],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","study_design_scores_codex":[0.003148267,0.001868054,0.3055085,0.0001334201,0.0004969877,0.0007740775,0.007463187,0.0003190212,0.4991169,0.002742522,0.01032265,0.1681065],"study_design_scores_gemma":[0.004073701,0.006990119,0.781557,0.001177261,0.0001967637,0.000363198,0.003943873,0.0001892583,0.005364778,0.008361603,0.1869363,0.000846132],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9968995,0.0006644069,0.000009500491,0.001069233,0.0002368235,0.00008560238,0.0000240323,0.000006212825,0.001004616],"genre_scores_gemma":[0.9981365,0.0006249343,0.0001166179,0.00004088904,0.0003928558,0.000001051207,0.00000712963,0.000001128439,0.0006789144],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4937521,"threshold_uncertainty_score":0.1884033,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3008187175","doi":"10.22148/001c.12049","title":"The Social Lives of Books: Reading Victorian Literature on Goodreads","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Games and Media","field":"Social Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Popularity; Reading (process); Syllabus; Victorian literature; Sociology; Plot (graphics); Order (exchange); Media studies; History; Psychology; Literature; Pedagogy; Art; Social psychology; Political science; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.03402950775756183,"gpt":0.3202783749835676,"spread":0.2862488672260058,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002781386,0.00007438684,0.0001883962,0.00002086529,0.000169248,0.0002215016,0.0002642265,0.00005242918,0.000009965146],"category_scores_gemma":[0.0005870427,0.00003864878,0.0002149204,0.0002535938,0.0001634991,0.0002587594,0.00001843322,0.0002285793,0.000003347099],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004114015,"about_ca_system_score_gemma":0.00008645724,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003015453,"about_ca_topic_score_gemma":0.00002391281,"domain_scores_codex":[0.9989493,0.00006059743,0.0002987695,0.00005577702,0.0004800736,0.0001554198],"domain_scores_gemma":[0.9989662,0.0001149042,0.0003635869,0.00003536735,0.0003669001,0.0001530204],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0003329997,0.000122296,0.0006575515,0.0000536162,0.0004193027,0.00009435,0.7447567,0.00005682576,0.001728689,0.1461019,0.04276745,0.06290832],"study_design_scores_gemma":[0.0001954914,0.00029165,0.0002984812,0.0001134964,0.00005970577,0.000003692729,0.04120884,0.00001831364,0.0003485201,0.0005759493,0.9568025,0.00008336167],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.2934521,0.002076058,0.00003551503,0.06970374,0.002363743,0.0002280695,0.00003183209,0.00003706749,0.6320719],"genre_scores_gemma":[0.9890972,0.0007068014,0.00005661478,0.0004177901,0.002667897,1.490734e-7,9.442134e-7,0.000004446159,0.007048186],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.914035,"threshold_uncertainty_score":0.2135945,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2913116882","doi":"10.22148/16.032","title":"Social Characters: The Hierarchy of Gender in Contemporary English-Language Fiction","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Media, Gender, and Advertising","field":"Social Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Hollywood; Hierarchy; Variety (cybernetics); Decentralization; Sociology; History; Political science; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.08365479870504956,"gpt":0.3525469657274379,"spread":0.2688921670223884,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008613788,0.00007024124,0.0002180825,0.00007792265,0.0001116638,0.00004530649,0.0002036853,0.00007123352,0.00006937956],"category_scores_gemma":[0.0002273154,0.00004174218,0.0001585101,0.0002522287,0.0001197675,0.0004191305,0.00001666615,0.0002735388,0.000003007982],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009126157,"about_ca_system_score_gemma":0.0001785458,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001046575,"about_ca_topic_score_gemma":0.00004374303,"domain_scores_codex":[0.9987032,0.000188524,0.0004047393,0.00006445537,0.0004850006,0.0001541122],"domain_scores_gemma":[0.9991041,0.00008597008,0.0004309884,0.00005536038,0.0002667138,0.00005687454],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.00008540724,0.00009023529,0.02708348,0.00004114864,0.0001228055,0.00002403517,0.9591008,0.0001018805,0.003561582,0.002388408,0.004694709,0.002705526],"study_design_scores_gemma":[0.001758736,0.0001471122,0.06108411,0.0001426707,0.00008467921,0.00001223811,0.9133847,0.0004301406,0.0007762582,0.001279025,0.02060764,0.0002926814],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9822398,0.0004089404,0.00001801341,0.001033115,0.0005531756,0.00008249348,0.000001725739,0.0000046001,0.01565816],"genre_scores_gemma":[0.9976857,0.0001359573,0.00002394846,0.0001528116,0.001096023,2.11267e-7,0.000001810328,0.000003882877,0.0008996949],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04571608,"threshold_uncertainty_score":0.1702197,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2963370929","doi":"10.22148/16.039","title":"The Carceral Archive: Documentary Records, Narrative Construction, and Predictive Risk Assessment","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Law in Society and Culture","field":"Social Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Scholarship; Narrative; Race (biology); Criminology; Sociology; Oral history; Law; Political science; History; Gender studies; Anthropology; Art; Literature","retraction":null,"screen_n_in":null,"score":{"opus":0.007586517233021723,"gpt":0.302421839749186,"spread":0.2948353225161642,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000673921,0.0000914423,0.0001734153,0.000006659189,0.001064971,0.000204364,0.0001832741,0.00006432087,0.0001139624],"category_scores_gemma":[0.00009545254,0.00004703856,0.0001482794,0.0001380149,0.0005984776,0.0005005046,0.00003170262,0.0004729495,0.000002345527],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001498878,"about_ca_system_score_gemma":0.0001570306,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001776701,"about_ca_topic_score_gemma":0.000383209,"domain_scores_codex":[0.9986863,0.0002962125,0.0002852215,0.00009197339,0.0004657517,0.0001745148],"domain_scores_gemma":[0.9987031,0.0001844314,0.0004679149,0.00006455118,0.0004632912,0.0001167682],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.0002922833,0.00009909587,0.3242927,0.00002354867,0.001963024,0.0000159483,0.4292676,0.0009352646,0.0002514978,0.05209095,0.1814439,0.009324186],"study_design_scores_gemma":[0.0006912922,0.0003341433,0.02229381,0.00005372168,0.0001986149,0.00005264805,0.6725146,0.0002606023,0.00001923597,0.03054949,0.2728663,0.0001655221],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9651611,0.0007984515,0.0001519322,0.01086457,0.001408218,0.0002485041,0.00003755993,0.00001167725,0.02131796],"genre_scores_gemma":[0.977868,0.01231107,0.002872365,0.0002345621,0.0009858531,0.000001547662,0.000005035517,0.000005419103,0.005716137],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3019989,"threshold_uncertainty_score":0.8191003,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2905726793","doi":"10.22148/16.030","title":"Critical Search: A Procedure for Guided Reading in Large-Scale Textual Corpora","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Reading (process); Computer science; Scale (ratio); Information retrieval; Natural language processing; Artificial intelligence; Linguistics; Geography; Philosophy; Cartography","retraction":null,"screen_n_in":null,"score":{"opus":0.0522741677462495,"gpt":0.3675912135698836,"spread":0.3153170458236341,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000888284,0.0001369568,0.0002929616,0.0002043003,0.0001176599,0.0002616434,0.0008480266,0.0001045515,0.000008436657],"category_scores_gemma":[0.001126747,0.00009049823,0.0001328266,0.0005755649,0.0001190826,0.001048678,0.0001452344,0.0003534863,0.000003779547],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001152254,"about_ca_system_score_gemma":0.0001554835,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000004273967,"about_ca_topic_score_gemma":0.00002108777,"domain_scores_codex":[0.9984361,0.00004422658,0.0005446535,0.000192178,0.0004179241,0.0003649354],"domain_scores_gemma":[0.9977239,0.0001252469,0.0002279249,0.0001698928,0.001615203,0.0001377816],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000729073,0.001637615,0.01141245,0.001235065,0.000326229,0.001183778,0.04378487,0.0002104859,0.09104089,0.5926408,0.2174502,0.03834849],"study_design_scores_gemma":[0.004123124,0.004033871,0.001876979,0.002367647,0.0002079781,0.004589188,0.005893474,0.5813993,0.1489238,0.2330705,0.01186681,0.001647348],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06720544,0.0008205953,0.921524,0.009285451,0.0003207801,0.0002523368,0.000009424491,0.0001255306,0.000456418],"genre_scores_gemma":[0.5978286,0.000008924825,0.4012893,0.0003540416,0.0003461732,0.000001292194,8.90703e-7,0.000006049298,0.0001647375],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.5811888,"threshold_uncertainty_score":0.3690411,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3123995752","doi":"10.22148/001c.18841","title":"Abolitionist Networks: Modeling Language Change in Nineteenth-Century Activist Newspapers","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Social and Cultural Dynamics","field":"Social Sciences","cited_by":19,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"National Science Foundation","keywords":"Newspaper; Scholarship; Vanguard; Credence; Politics; History; Argument (complex analysis); Sociology; Media studies; Political science; Law; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.04176687371627995,"gpt":0.339984818055101,"spread":0.298217944338821,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002883753,0.0001190754,0.0002770141,0.0000444758,0.0002902022,0.0001910242,0.0001879211,0.000115515,0.0001068388],"category_scores_gemma":[0.000192976,0.000089824,0.0002237914,0.0005764548,0.00009498228,0.0005632933,0.00003042661,0.0003452863,0.000003979451],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002571798,"about_ca_system_score_gemma":0.0001224898,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001165678,"about_ca_topic_score_gemma":0.005949384,"domain_scores_codex":[0.9986112,0.000139703,0.0003842045,0.0001135811,0.0004366869,0.000314615],"domain_scores_gemma":[0.9990614,0.00003963373,0.0002323033,0.00006500958,0.0004388655,0.0001627961],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.0003852885,0.001500932,0.0406865,0.0001544023,0.001038255,0.003055595,0.567637,0.1714526,0.003224263,0.09589391,0.005596128,0.1093752],"study_design_scores_gemma":[0.002472128,0.0001509828,0.01458855,0.0006428953,0.0004277427,0.0001451199,0.7403783,0.1955561,0.00005693025,0.002085505,0.04232414,0.001171569],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9838746,0.004087089,0.0001788942,0.005075546,0.0008205444,0.00009215948,0.000005984932,0.00002085409,0.005844378],"genre_scores_gemma":[0.9907226,0.005823078,0.0002551548,0.0004767884,0.001861379,0.000001339513,0.00001404385,0.000007279285,0.0008383678],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1727413,"threshold_uncertainty_score":0.3662917,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4319604386","doi":"10.22148/001c.55592","title":"Should we really ‘hermeneutise’ the Digital Humanities? A plea for the epistemic productivity of a ‘cultural technique of flattening’ in the Humanities.","year":2023,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":17,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Digital humanities; Hermeneutics; Humanities; Epistemology; Sociology; Philosophy","retraction":null,"screen_n_in":null,"score":{"opus":0.1598366603311808,"gpt":0.3027010683654781,"spread":0.1428644080342973,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.001372157,0.0002735601,0.0005184115,0.0001629617,0.0006480043,0.001128627,0.001064887,0.00006761042,0.00005915983],"category_scores_gemma":[0.0002780527,0.000113502,0.0005547445,0.0002069175,0.001234707,0.0011616,0.0001126768,0.0006166645,0.000002649759],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007547786,"about_ca_system_score_gemma":0.00009454324,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000954247,"about_ca_topic_score_gemma":0.001492375,"domain_scores_codex":[0.9976018,0.0001087565,0.001065794,0.0001586643,0.0007081197,0.0003568365],"domain_scores_gemma":[0.9970999,0.000686295,0.0009756521,0.0003222404,0.0008899836,0.00002590238],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.0005345201,0.0004380915,0.0005175551,0.0007336034,0.000918724,0.00005781012,0.3703752,0.0007105683,0.0009420133,0.5725064,0.04683432,0.005431191],"study_design_scores_gemma":[0.001051361,0.001413779,0.001953437,0.0007892037,0.0005144402,0.0002566183,0.4821534,0.0002650976,0.0005223872,0.03738357,0.4731684,0.0005282859],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9475805,0.001225041,0.00001882986,0.009104247,0.0006882446,0.001550553,0.0004288389,0.00005106503,0.03935271],"genre_scores_gemma":[0.9843513,0.0001914602,0.000009955105,0.0002246815,0.0008195007,0.00003687334,0.00001842569,0.00002294146,0.01432489],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5351229,"threshold_uncertainty_score":0.9999083,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2962933257","doi":"10.22148/16.043","title":"Visual Style in Two Network Era Sitcoms","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Media Studies and Communication","field":"Social Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Scholarship; Framing (construction); Style (visual arts); Computer science; Sociology; Political science; Geography; Visual arts; Art","retraction":null,"screen_n_in":null,"score":{"opus":0.03048623578256494,"gpt":0.3704559229220717,"spread":0.3399696871395068,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006888166,0.00005465754,0.0001905606,0.00003061655,0.0001271162,0.00004776641,0.0002417747,0.00002373473,0.0001313873],"category_scores_gemma":[0.0001040903,0.00003706817,0.00009022294,0.000294796,0.00005414003,0.0002058471,0.00004217585,0.000242994,0.00002839648],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009804107,"about_ca_system_score_gemma":0.00006251843,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002590399,"about_ca_topic_score_gemma":0.002168621,"domain_scores_codex":[0.9990332,0.0001243899,0.0002984001,0.00004612535,0.0003221214,0.0001757648],"domain_scores_gemma":[0.9992881,0.0001107066,0.0002538871,0.000072091,0.0002084505,0.00006674958],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002178554,0.0004609124,0.751902,0.00003493784,0.0004191483,0.00005902809,0.07654774,0.01956327,0.0004611683,0.0765556,0.04357783,0.0302005],"study_design_scores_gemma":[0.003756531,0.0006193025,0.1504157,0.0004323342,0.0002104931,0.0000268414,0.1600171,0.01188945,0.00004837812,0.008358775,0.6635222,0.0007027549],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9794731,0.0006534133,0.00001662633,0.005059681,0.0004435304,0.00006577827,2.239021e-7,0.000004381225,0.01428329],"genre_scores_gemma":[0.9964703,0.001173846,0.0003378678,0.000211622,0.0005948467,3.004957e-7,6.661028e-7,0.000002459042,0.001208052],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6199445,"threshold_uncertainty_score":0.1511596,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4247740782","doi":"10.22148/16.011","title":"Fictionality","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Narrative Theory and Analysis","field":"Arts and Humanities","cited_by":15,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Narrative; Literature; Philosophy; Art","retraction":null,"screen_n_in":null,"score":{"opus":0.05804555801263898,"gpt":0.2682156559321219,"spread":0.2101700979194829,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002455699,0.00006443178,0.0001593251,0.00005221728,0.0001218005,0.00005499636,0.0001044826,0.00001450496,0.004440356],"category_scores_gemma":[0.00008171381,0.00002652204,0.000221437,0.00003261611,0.0001572137,0.0003891046,0.000009096759,0.00006387698,0.00005427351],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003037946,"about_ca_system_score_gemma":0.00001585847,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000002788841,"about_ca_topic_score_gemma":0.00002407546,"domain_scores_codex":[0.9993645,0.000045769,0.0002868283,0.00004416892,0.0001792891,0.0000794257],"domain_scores_gemma":[0.9990539,0.00006563443,0.0002526202,0.00005193236,0.0005225108,0.00005340743],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007850607,0.00008883965,0.0006063574,0.000006965201,0.0004639525,0.00001888488,0.006974905,0.00002520632,0.002458696,0.950352,0.03477804,0.004147661],"study_design_scores_gemma":[0.0006411591,0.0002089385,0.0004545925,0.00008956917,0.0002879893,0.00004593879,0.008970368,0.00003131615,0.001208445,0.06893941,0.9189252,0.0001971379],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.912129,0.0002673526,0.001093253,0.01199517,0.000858465,0.00003391397,0.00003361667,0.00002104156,0.07356818],"genre_scores_gemma":[0.9659431,0.00004384422,0.00004491447,0.0001310327,0.001186033,1.147108e-7,5.130941e-7,0.000002559138,0.03264791],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8841471,"threshold_uncertainty_score":0.9964697,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2985141934","doi":"10.22148/16.048","title":"A Shared Task for the Digital Humanities Chapter 1: Introduction to Annotation, Narrative Levels and Shared Tasks","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":14,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Digital humanities; Annotation; Narrative; Task (project management); Field (mathematics); Computer science; Humanities; World Wide Web; Artificial intelligence; Literature; Philosophy; Art; Engineering; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.0752353978742625,"gpt":0.2643746749599136,"spread":0.1891392770856511,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0001736801,0.0001522391,0.0002339998,0.00009006075,0.0003825274,0.002378132,0.0001809907,0.00002633555,0.0007800618],"category_scores_gemma":[0.0001218937,0.00008566079,0.0001465485,0.00003030345,0.0001321864,0.002090747,0.00003764954,0.0001555894,0.0000202762],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003885381,"about_ca_system_score_gemma":0.00001722215,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000057963,"about_ca_topic_score_gemma":0.00008442579,"domain_scores_codex":[0.999051,0.00001031975,0.0003995815,0.0001269796,0.0002546539,0.0001575005],"domain_scores_gemma":[0.9984675,0.0001006662,0.0002716986,0.00009910279,0.001001225,0.00005981602],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0004260858,0.0001039474,0.0002052266,0.00008785954,0.0006535939,0.000006080457,0.3686572,0.0003351384,0.000247246,0.5494363,0.07542585,0.004415459],"study_design_scores_gemma":[0.0006371835,0.00101403,0.00101267,0.00007422518,0.0001103527,0.00004343106,0.1638281,0.000236364,0.00003072344,0.0066282,0.8261178,0.0002669625],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9719214,0.0004381354,0.0001006025,0.01016645,0.001386476,0.0005904126,0.0005711118,0.00002459931,0.01480078],"genre_scores_gemma":[0.9654097,0.000006278679,0.00004116317,0.0005221156,0.0021997,0.000004122509,0.00002406978,0.00001251237,0.03178031],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.750692,"threshold_uncertainty_score":0.9986575,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3004295495","doi":"10.22148/001c.11829","title":"On the perceived complexity of literature. A response to Nan Z. Da","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Reductionism; Variance (accounting); Literary criticism; Epistemology; Sociology; Linguistics; Philosophy; Economics","retraction":null,"screen_n_in":null,"score":{"opus":0.07676143944045041,"gpt":0.3253034171408474,"spread":0.248541977700397,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004797195,0.0001336921,0.0003386385,0.000109642,0.00007782883,0.000156036,0.001213543,0.00003896379,0.00001495128],"category_scores_gemma":[0.0009692652,0.00006810381,0.0002802784,0.001070993,0.00009372162,0.0003665854,0.0001543559,0.0003350454,0.000007529793],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005473877,"about_ca_system_score_gemma":0.00004303337,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000001442107,"about_ca_topic_score_gemma":0.000001824049,"domain_scores_codex":[0.9984972,0.0002043599,0.0004917736,0.0001456715,0.0005191945,0.000141798],"domain_scores_gemma":[0.9981439,0.0002396859,0.0004437659,0.0002959636,0.0007064937,0.0001702384],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.003130936,0.0004626862,0.0003577849,0.00007074836,0.0008633867,0.0004685378,0.05947386,0.01153763,0.5478448,0.2975384,0.07103467,0.007216523],"study_design_scores_gemma":[0.004306785,0.01934083,0.05985022,0.002898386,0.0009137309,0.0009448749,0.006916361,0.3437669,0.1971721,0.2848944,0.07596045,0.003034899],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7229555,0.000109378,0.1939351,0.08248512,0.00006731479,0.0001357048,0.00000661446,0.00004671271,0.0002585413],"genre_scores_gemma":[0.9509771,0.00001865869,0.04611333,0.002729904,0.00007246019,4.696461e-7,4.176599e-7,0.000004318076,0.00008335767],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3506727,"threshold_uncertainty_score":0.2777193,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2931020970","doi":"10.22148/16.028","title":"Historiography’s Two Voices: Data Infrastructure and History at Scale in the Oxford Dictionary of National Biography (ODNB)","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Philippine History and Culture","field":"Social Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Historiography; Biography; Scale (ratio); History; Classics; Literature; Media studies; Art history; Sociology; Art; Geography; Cartography; Archaeology","retraction":null,"screen_n_in":null,"score":{"opus":0.04836581654379901,"gpt":0.3157815727627241,"spread":0.2674157562189251,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007702224,0.0001033714,0.0001971199,0.0002154877,0.0003101542,0.00001866072,0.00062631,0.0000830076,0.0001331781],"category_scores_gemma":[0.0001129145,0.0000624555,0.0001347262,0.0005632973,0.0010098,0.0006039396,0.00006305732,0.0002458185,8.072851e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004422808,"about_ca_system_score_gemma":0.0001741607,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001685212,"about_ca_topic_score_gemma":0.002717445,"domain_scores_codex":[0.9984617,0.0001484794,0.0003830853,0.0001311349,0.0007369816,0.0001386446],"domain_scores_gemma":[0.9986678,0.00005406663,0.0004736537,0.0001484476,0.0005749832,0.00008106625],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002316325,0.0002579672,0.1077901,0.00004885314,0.0002822038,0.00001567689,0.1673011,0.00005789287,0.002289804,0.009190369,0.7113062,0.001228177],"study_design_scores_gemma":[0.0002961552,0.00008531546,0.0303196,0.00003120725,0.00006712016,0.00003510895,0.006334158,0.0000520288,0.000004723593,0.001683969,0.9609998,0.0000908099],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9589329,0.0105165,0.00002722189,0.002890167,0.001631628,0.0001473072,0.00008629313,0.00001373909,0.0257542],"genre_scores_gemma":[0.9967351,0.0008910772,0.0003491082,0.0003837453,0.00114269,5.478345e-7,0.00001609593,0.000003770925,0.0004778419],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2496936,"threshold_uncertainty_score":0.3720649,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2587367367","doi":"10.22148/16.009","title":"Genre, Computation, and the Varieties of Twentieth-Century U.S. Fiction","year":2016,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":13,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Literary fiction; Literature; Literary genre; Fiction theory; Popular fiction; Techno-thriller; History; World War II; Art; Literary criticism; Archaeology","retraction":null,"screen_n_in":null,"score":{"opus":0.03885955791862655,"gpt":0.2399455145696156,"spread":0.2010859566509891,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002754877,0.00007603391,0.0001899084,0.00004463658,0.0001494925,0.000246578,0.00009274875,0.00002005828,0.0001625047],"category_scores_gemma":[0.00006348315,0.000029712,0.0001324025,0.00002105463,0.0005007416,0.0006450417,0.00002382603,0.00008763529,0.000002251943],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001574526,"about_ca_system_score_gemma":0.00001411123,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001142276,"about_ca_topic_score_gemma":0.00001982611,"domain_scores_codex":[0.9992167,0.0000379877,0.0003882308,0.00004479254,0.000230765,0.00008149414],"domain_scores_gemma":[0.9987994,0.000119003,0.0003869236,0.00004647179,0.0006141876,0.00003404528],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002634715,0.00006408877,0.0003774984,0.00004485554,0.0003097225,0.000004483296,0.03024066,0.00004488773,0.00006756649,0.9360625,0.02265482,0.009865467],"study_design_scores_gemma":[0.005678855,0.0007428109,0.003704618,0.0003690356,0.0006384949,0.0001518632,0.07261026,0.0003430451,0.0001659681,0.1287617,0.7864199,0.0004134279],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9283512,0.003417362,0.0004475411,0.003226228,0.001470641,0.0001208727,0.00003169392,0.00001477821,0.06291967],"genre_scores_gemma":[0.9927886,0.0006548262,0.00003248025,0.0001022214,0.0003807494,2.574677e-7,0.000001136824,0.000004170447,0.006035578],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8073007,"threshold_uncertainty_score":0.2377757,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2898247218","doi":"10.22148/16.026","title":"Image Analytics and the Nineteenth-Century Illustrated Newspaper","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Photography and Visual Culture","field":"Arts and Humanities","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Newspaper; Sublime; Flood myth; History; Analytics; Art history; Media studies; Literature; Art; Computer science; Sociology; Data science; Archaeology","retraction":null,"screen_n_in":null,"score":{"opus":0.03466716520478621,"gpt":0.2718072904928382,"spread":0.237140125288052,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003107919,0.0001804834,0.000328145,0.00007564925,0.0004053701,0.000439501,0.0002331043,0.00005669363,0.0006961644],"category_scores_gemma":[0.00008336929,0.00008006691,0.000272499,0.0001328686,0.001498723,0.0004548945,0.00003817806,0.0003422766,0.00001586308],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001609358,"about_ca_system_score_gemma":0.00002549151,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004572286,"about_ca_topic_score_gemma":0.0001614395,"domain_scores_codex":[0.9988393,0.00007147514,0.0004736239,0.0001022338,0.0003039564,0.0002094599],"domain_scores_gemma":[0.9982854,0.000061029,0.0004019308,0.000114534,0.001017786,0.0001193466],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.002407137,0.0005466904,0.000945594,0.0001504146,0.003706296,0.0001472421,0.1976053,0.00003696238,0.01046158,0.3973838,0.3838261,0.002782824],"study_design_scores_gemma":[0.004399883,0.0009289416,0.0009871245,0.0001467009,0.001347467,0.0002660127,0.0540055,0.00318671,0.0007252256,0.003125365,0.9303656,0.0005154518],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9566483,0.002145253,0.0000492651,0.003974696,0.001069113,0.000204971,0.00003677879,0.00003411839,0.03583752],"genre_scores_gemma":[0.9932857,0.0008373363,0.000144731,0.0008417172,0.002482681,4.909822e-7,0.000004327595,0.000009433707,0.002393608],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5465395,"threshold_uncertainty_score":0.7622512,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4313409859","doi":"10.22148/001c.57195","title":"From Concepts to Texts and Back: Operationalization as a Core Activity of Digital Humanities","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Operationalization; Realm; Epistemology; Process (computing); Construct (python library); Computer science; Sociology; Political science; Philosophy","retraction":null,"screen_n_in":null,"score":{"opus":0.203411329081511,"gpt":0.4160787014155996,"spread":0.2126673723340886,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007466531,0.00006275346,0.0001891625,0.0001487424,0.0001798884,0.0005128946,0.0003884746,0.00001039387,0.0008962275],"category_scores_gemma":[0.0009333392,0.00004097034,0.00006868798,0.0004119571,0.00006388861,0.0005694751,0.0003578818,0.00008018909,0.00002190325],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003972857,"about_ca_system_score_gemma":0.0000468719,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003107542,"about_ca_topic_score_gemma":0.00001241762,"domain_scores_codex":[0.9980307,0.00005630786,0.000453379,0.0001576251,0.001229492,0.00007244764],"domain_scores_gemma":[0.99856,0.0002581641,0.0004156863,0.0001580464,0.0005471263,0.00006103314],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0005259031,0.0006153151,0.02983414,0.00001870564,0.0004177375,0.00007687204,0.03405437,0.1509198,0.00689003,0.01165716,0.6797463,0.08524369],"study_design_scores_gemma":[0.0026412,0.002174649,0.06060961,0.0001338933,0.0002627749,0.0002412825,0.1367145,0.1049757,0.001803294,0.06195781,0.627681,0.000804256],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9952267,0.00004888878,0.002299206,0.0008008137,0.0003667583,0.00004991862,0.0001073651,0.000002448328,0.001097842],"genre_scores_gemma":[0.9936478,0.000002698876,0.000359714,0.000144451,0.0001019906,2.67733e-7,0.00001294677,0.000001920886,0.005728169],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1026602,"threshold_uncertainty_score":0.9813063,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2895759845","doi":"10.22148/16.025","title":"Stable Random Projection: Lightweight, General-Purpose Dimensionality Reduction for Digitized Libraries","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Advanced Data Storage Technologies","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"sort; Computer science; Dimensionality reduction; Digital library; Projection (relational algebra); Random projection; Reduction (mathematics); Computer graphics (images); World Wide Web; Multimedia; Information retrieval; Artificial intelligence; Mathematics; Algorithm; Art","retraction":null,"screen_n_in":null,"score":{"opus":0.03697347895973706,"gpt":0.2909724707646574,"spread":0.2539989918049203,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003118404,0.0001589188,0.0003436482,0.0001384656,0.0003059635,0.000310086,0.0006470841,0.0000852128,0.000009039355],"category_scores_gemma":[0.0003479274,0.00009982896,0.0001745555,0.0005141583,0.0002237995,0.003473201,0.0001827055,0.0001784991,0.000005173416],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009995425,"about_ca_system_score_gemma":0.0001221362,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000002991299,"about_ca_topic_score_gemma":0.000001726126,"domain_scores_codex":[0.9986131,0.00003962516,0.0005438703,0.0002170493,0.0003513808,0.0002349804],"domain_scores_gemma":[0.9978524,0.00008139112,0.0005775683,0.0003283449,0.001082137,0.00007810834],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.002230925,0.0005752986,0.0001789018,0.0001302109,0.0008343807,0.00009230767,0.002369675,0.001266608,0.07589925,0.547153,0.3210278,0.04824159],"study_design_scores_gemma":[0.005280604,0.00174197,0.0001618542,0.0001353147,0.000190319,0.001305999,0.0007220225,0.03374019,0.2863511,0.3359133,0.3337642,0.0006932069],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.06972648,0.0005460573,0.9230286,0.004210607,0.001533813,0.0003142208,0.00001801911,0.0002512606,0.0003708884],"genre_scores_gemma":[0.2361082,0.0002241778,0.7582468,0.0001604886,0.001802352,0.00001067628,0.0000148754,0.00001531408,0.003417149],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.2112397,"threshold_uncertainty_score":0.4070907,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3003312070","doi":"10.22148/001c.11830","title":"Is there a text in my data? (Part 1): on counting words","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Computer science; Word (group theory); Context (archaeology); Simple (philosophy); Normalization (sociology); Natural language processing; Linguistics; Artificial intelligence; Epistemology; Philosophy; History; Sociology","retraction":null,"screen_n_in":null,"score":{"opus":0.1950699218308501,"gpt":0.292525711556765,"spread":0.09745578972591484,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002279573,0.0001320379,0.0002643903,0.00005338736,0.0001150121,0.0008091169,0.0004649202,0.00002998701,0.001538605],"category_scores_gemma":[0.0001086192,0.00008300057,0.000116892,0.00005997915,0.00008657794,0.001228034,0.00007386616,0.0004003607,0.00007494389],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003475662,"about_ca_system_score_gemma":0.0000336841,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001862996,"about_ca_topic_score_gemma":0.000232279,"domain_scores_codex":[0.998819,0.00002297844,0.000488814,0.000121465,0.0003703984,0.0001773593],"domain_scores_gemma":[0.9992285,0.00004944014,0.0002890542,0.0001331818,0.0002031771,0.00009667697],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000446134,0.0004086852,0.004339511,0.000190846,0.0005112353,0.000537821,0.1778997,0.0003458996,0.00003222658,0.1404863,0.6666762,0.008125448],"study_design_scores_gemma":[0.0003806475,0.0002552917,0.0002331983,0.0001785786,0.00004650369,0.00001046407,0.02903624,0.0005623952,0.00001232677,0.0007920451,0.9683377,0.0001546495],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7872122,0.0006551065,0.000002539022,0.008581355,0.000532428,0.00006959635,0.0001127927,0.00001681514,0.2028172],"genre_scores_gemma":[0.9867961,0.00007943761,0.00001646208,0.003920354,0.0023294,1.989743e-7,0.00001165646,0.0000108827,0.006835485],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3016614,"threshold_uncertainty_score":0.9993742,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3091866587","doi":"10.22148/001c.17585","title":"Divergence and the Complexity of Difference in Text and Culture","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language and cultural evolution","field":"Social Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"John Templeton Foundation","keywords":"Divergence (linguistics); Operationalization; Metaphor; Rhetorical question; Darwin (ADL); Function (biology); Epistemology; Computer science; Linguistics; Construal level theory; Sociology; Philosophy; Social science; Evolutionary biology","retraction":null,"screen_n_in":null,"score":{"opus":0.06567579411856664,"gpt":0.3085290246590696,"spread":0.242853230540503,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002245825,0.00005419886,0.0001881505,0.00001062711,0.0001145137,0.00003499262,0.0001332594,0.00003734514,0.0000264422],"category_scores_gemma":[0.0003399594,0.00002418444,0.00005134953,0.0001935758,0.0008041895,0.0001733278,0.00003673037,0.0001529495,4.77645e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000164406,"about_ca_system_score_gemma":0.00002170412,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004091651,"about_ca_topic_score_gemma":0.000663853,"domain_scores_codex":[0.9992983,0.0001069532,0.0002239385,0.00005385972,0.0002323997,0.0000845539],"domain_scores_gemma":[0.9994777,0.0000347364,0.0002092393,0.00002493979,0.0001743248,0.00007901432],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"observational","study_design_scores_codex":[0.0005552588,0.00008494874,0.06792299,0.0001086423,0.0001584318,0.00003985265,0.7755641,0.0001076026,0.007142475,0.1335573,0.005649249,0.00910907],"study_design_scores_gemma":[0.006828574,0.0006781253,0.4780393,0.0004050301,0.0005276842,0.00009525925,0.4243494,0.006300847,0.0008228703,0.06321181,0.01808647,0.0006546603],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9798921,0.002361003,0.00004048786,0.01682832,0.00003919198,0.0000631829,0.000002723921,0.000002690697,0.0007703457],"genre_scores_gemma":[0.9973534,0.001877219,0.0001368797,0.0002492171,0.0001188617,1.457831e-7,3.493898e-7,9.111364e-7,0.0002630179],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4101163,"threshold_uncertainty_score":0.296307,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2909609490","doi":"10.22148/16.031","title":"Race, Writing, and Computation: Racial Difference and the US Novel, 1880-2000","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Race (biology); Computation; Racial differences; Sociology; Computer science; Gender studies; Ethnic group; Anthropology; Programming language","retraction":null,"screen_n_in":null,"score":{"opus":0.03399302375473869,"gpt":0.3561622399781074,"spread":0.3221692162233687,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001140888,0.00008277799,0.0002834364,0.00005488782,0.0003160137,0.0002352485,0.0001280102,0.00004244148,0.00005365256],"category_scores_gemma":[0.0002349886,0.00004512587,0.0001119514,0.0002554646,0.0003376796,0.0002057638,0.00003771484,0.0001810044,0.000003221519],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003531552,"about_ca_system_score_gemma":0.000100503,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001031717,"about_ca_topic_score_gemma":0.0001081566,"domain_scores_codex":[0.998807,0.0001961549,0.0003472099,0.00008991909,0.0004382991,0.000121466],"domain_scores_gemma":[0.9984861,0.000585452,0.0003647677,0.0000364602,0.0004349526,0.00009225441],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.001005658,0.000384696,0.2083748,0.0001235165,0.001743459,0.00004695566,0.1603253,0.02178289,0.001289454,0.3609624,0.002601117,0.2413598],"study_design_scores_gemma":[0.01182404,0.0004492187,0.5540549,0.000313646,0.001570266,0.0003676244,0.03693128,0.2098789,0.00004759302,0.07553462,0.1079978,0.001030123],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9869877,0.000852076,0.004219784,0.004925872,0.0001729128,0.0000746933,0.000001406135,0.00000551264,0.00276003],"genre_scores_gemma":[0.9937555,0.0002945331,0.003116737,0.0002521264,0.0004332402,2.443065e-7,8.677628e-7,0.000002656852,0.002144051],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3456801,"threshold_uncertainty_score":0.2430554,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3210937173","doi":"10.22148/001c.29802","title":"Audiobook Stylistics: Comparing print and audio in the bestselling segment","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Stylistics; Depiction; Focus (optics); Point (geometry); Style (visual arts); Set (abstract data type); Plot (graphics); Computer science; Advertising; Art; Literature; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.06872344227832917,"gpt":0.3059254956830484,"spread":0.2372020534047192,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008486422,0.00008281991,0.0001949349,0.0000460301,0.00009719329,0.0002516318,0.0003378584,0.00003304927,0.000004783069],"category_scores_gemma":[0.0001214764,0.00004880718,0.00007251258,0.0002612458,0.00002984892,0.0002474713,0.0001010121,0.0003422268,0.000002945802],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005975058,"about_ca_system_score_gemma":0.00006896296,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000002349808,"about_ca_topic_score_gemma":0.00001126733,"domain_scores_codex":[0.9989161,0.0001242809,0.0003974395,0.0001038878,0.0003092436,0.0001490757],"domain_scores_gemma":[0.9991978,0.0001086132,0.0002250031,0.0001294996,0.0002706271,0.00006846355],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000085136,0.0005840521,0.1361489,0.0003351773,0.0005927049,0.003425613,0.08498356,0.1216437,0.007663975,0.6102176,0.003791854,0.03052768],"study_design_scores_gemma":[0.002780622,0.0004156319,0.07550018,0.0009517676,0.0002834556,0.004846983,0.01960247,0.8196155,0.01522803,0.02592378,0.03386863,0.0009829415],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.473221,0.001969483,0.5160739,0.007300537,0.0003632874,0.00007595741,7.675535e-7,0.00001402765,0.0009810557],"genre_scores_gemma":[0.9843547,0.0002380139,0.0148079,0.0003461922,0.000110317,2.842258e-7,6.765089e-7,0.000001890647,0.0001400213],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6979718,"threshold_uncertainty_score":0.2426491,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2993640618","doi":"10.22148/001c.11826","title":"Do we know what we are doing?","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Mental Health Research Topics","field":"Psychology","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Open science; Value (mathematics); Norm (philosophy); Psychological science; Replication (statistics); Psychology; Space (punctuation); Social psychology; Epistemology; Sociology; Political science; Computer science; Law; Mathematics; Statistics","retraction":null,"screen_n_in":null,"score":{"opus":0.1495892918248977,"gpt":0.445563802616336,"spread":0.2959745107914384,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0003170427,0.0001532857,0.0003815482,0.00008273369,0.00009289548,0.0001859266,0.0004042257,0.0001126123,0.001993747],"category_scores_gemma":[0.0001147924,0.0001001817,0.0002211722,0.0003922799,0.00007181274,0.0005068785,0.00007168676,0.000772071,0.0004239428],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001220054,"about_ca_system_score_gemma":0.00006197015,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000004814674,"about_ca_topic_score_gemma":0.000008080524,"domain_scores_codex":[0.9980402,0.0001660471,0.0006434752,0.0001623183,0.000606832,0.0003811186],"domain_scores_gemma":[0.9982305,0.00009664144,0.0005433508,0.000158626,0.0003842361,0.0005866952],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0008679875,0.0004517976,0.005079911,0.000626872,0.0007434299,0.002100662,0.0288665,0.0002385889,0.0006941616,0.004383584,0.50116,0.4547865],"study_design_scores_gemma":[0.001000647,0.0006788286,0.001102408,0.0005666651,0.00006939725,0.0002739927,0.03233403,0.0002938181,0.0001650071,0.0004750005,0.9628704,0.0001698323],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.2754123,0.1639962,0.0005782277,0.5443587,0.004440377,0.0005184491,0.00002693544,0.00006618686,0.01060254],"genre_scores_gemma":[0.8947737,0.07882398,0.002510202,0.006321201,0.004937559,0.000003545232,0.000004192535,0.00004776517,0.01257787],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6193614,"threshold_uncertainty_score":0.9989186,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2993740281","doi":"10.22148/16.035","title":"Data Cultures, Culture as Data – Special Issue of Cultural Analytics","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":11,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Research data; Data science; Computer science; Sociology; Data curation","retraction":null,"screen_n_in":null,"score":{"opus":0.1263592046849277,"gpt":0.3235582065088711,"spread":0.1971990018239434,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0006010638,0.0004208313,0.0009077869,0.000144907,0.0002505353,0.00134818,0.003354106,0.0001510015,0.007021094],"category_scores_gemma":[0.0003482439,0.0002509706,0.0003377084,0.0001318396,0.0004117999,0.005368162,0.0008185334,0.0007932212,0.0003070674],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008156996,"about_ca_system_score_gemma":0.0001378887,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009359731,"about_ca_topic_score_gemma":0.0006624048,"domain_scores_codex":[0.9965551,0.0000759668,0.001339412,0.0004193672,0.00117207,0.0004380445],"domain_scores_gemma":[0.9955323,0.00007456904,0.001251732,0.001302845,0.001596931,0.0002415953],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001875331,0.0002718547,0.000312419,0.0001427322,0.0008289012,0.00008504861,0.02056751,0.0001374556,0.0001330837,0.04817985,0.927821,0.001332556],"study_design_scores_gemma":[0.0007198695,0.0004243542,0.00007050575,0.0001693668,0.0004812406,0.0001393997,0.04780245,0.0004448002,0.00005871513,0.001368513,0.9479311,0.0003896549],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.2525089,0.003096891,0.00001185724,0.002564549,0.006294766,0.0005178056,0.004866995,0.00006479594,0.7300735],"genre_scores_gemma":[0.7220341,0.0009514285,0.0004421735,0.000978183,0.02466464,4.972925e-7,0.002137095,0.00005475561,0.2487371],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.4813363,"threshold_uncertainty_score":0.9999943,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2802843067","doi":"10.22148/16.022","title":"Self-Repetition and East Asian Literary Modernity, 1900-1930","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Music Technology and Sound Studies","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"BATES; Feeling; Sampling (signal processing); Literature; Aesthetics; Sound (geography); Empirical research; History; Psychology; Art; Social psychology; Computer science; Epistemology; Philosophy; Engineering; Acoustics; Telecommunications","retraction":null,"screen_n_in":null,"score":{"opus":0.02039322726241128,"gpt":0.2531978847216531,"spread":0.2328046574592418,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001850042,0.0000936621,0.0001786493,0.00009024431,0.0001984973,0.0001278552,0.0003294238,0.00007095886,0.000007790939],"category_scores_gemma":[0.00004749621,0.00006076502,0.00006194598,0.000205625,0.0001318356,0.0005844456,0.0001530812,0.0002024027,0.000008484346],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000256145,"about_ca_system_score_gemma":0.00002712347,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000001159218,"about_ca_topic_score_gemma":0.00001121131,"domain_scores_codex":[0.9992736,0.00003016411,0.0002574415,0.0001100315,0.0001938945,0.0001348899],"domain_scores_gemma":[0.9991573,0.00001412001,0.0002033635,0.000137148,0.0004210852,0.00006694418],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001851918,0.00109556,0.0371072,0.0003231841,0.002919061,0.000904231,0.211123,0.00005486522,0.001953385,0.3906214,0.1116579,0.2420551],"study_design_scores_gemma":[0.003453444,0.003762035,0.05750791,0.0005233066,0.0005879861,0.007250868,0.004967227,0.12761,0.001262106,0.4483429,0.3432244,0.001507844],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8290507,0.001988203,0.1259335,0.02650866,0.001285195,0.0001127999,0.000003900203,0.0002081768,0.01490882],"genre_scores_gemma":[0.9785387,0.000142044,0.02030156,0.0003122868,0.0003618153,2.609314e-7,3.057135e-7,0.000002317782,0.0003406926],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2405472,"threshold_uncertainty_score":0.2477926,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3000500424","doi":"10.22148/001c.11652","title":"Fleshing Out Models of Gender in English-Language Novels (1850 – 2000)","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Gender Studies in Language","field":"Social Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Binary opposition; Psychology; Characterization (materials science); Linguistics; Sociology; Gender studies; Philosophy","retraction":null,"screen_n_in":null,"score":{"opus":0.1068315857302793,"gpt":0.3483203648240278,"spread":0.2414887790937486,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006420391,0.0001269861,0.0004089496,0.00007909884,0.00009441883,0.00005336358,0.0004508129,0.00009031422,0.00009419511],"category_scores_gemma":[0.001126515,0.00009447928,0.0002010556,0.0004058779,0.0001506731,0.0005270832,0.00008598823,0.0003647921,0.000003975597],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001759276,"about_ca_system_score_gemma":0.0001774809,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001348604,"about_ca_topic_score_gemma":0.0004292149,"domain_scores_codex":[0.9981249,0.0001167099,0.0006336492,0.0001219244,0.0007175206,0.0002853276],"domain_scores_gemma":[0.9986745,0.00009268059,0.00047106,0.00009079715,0.0004963704,0.000174656],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.00002643564,0.00005038195,0.0009068525,0.0000364336,0.0001137007,0.00007766186,0.9843717,0.008604225,0.0007688556,0.002245647,0.001873874,0.000924255],"study_design_scores_gemma":[0.001148591,0.0001320318,0.0006288612,0.00007931954,0.0001834102,0.000007908106,0.9848407,0.004335371,0.0007191774,0.001775049,0.005800123,0.0003494832],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9042105,0.00395465,0.002589642,0.003578287,0.0009617212,0.0001953317,0.00002164389,0.00003578264,0.08445244],"genre_scores_gemma":[0.9963316,0.0001775628,0.001815584,0.0003750705,0.0009026328,3.484619e-7,8.110698e-7,0.000008480134,0.0003879375],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.09212106,"threshold_uncertainty_score":0.3852754,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2988314392","doi":"10.22148/16.049","title":"A Shared Task for the Digital Humanities Chapter 2: Evaluating Annotation Guidelines","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Guideline; Annotation; Task (project management); Section (typography); Computer science; Data science; Artificial intelligence; Political science; Engineering; Systems engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.1018420387922206,"gpt":0.3715093349101505,"spread":0.2696672961179299,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004400817,0.00009930227,0.0001543452,0.00006475738,0.0001074268,0.0006875288,0.0006813385,0.00003307668,0.00000864346],"category_scores_gemma":[0.0005134908,0.00004937507,0.0001491836,0.0001213658,0.0000252599,0.001411406,0.00008158749,0.0001351314,0.000004578614],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003419332,"about_ca_system_score_gemma":0.00003553005,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000002071346,"about_ca_topic_score_gemma":0.000001214916,"domain_scores_codex":[0.9989566,0.00001101795,0.0004312768,0.00009892352,0.000384575,0.0001176599],"domain_scores_gemma":[0.9974825,0.0001571689,0.0005021181,0.000149214,0.001684718,0.00002428416],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002308572,0.0001212296,0.001131758,0.0002820738,0.0007565171,0.00003825226,0.01477267,0.005629469,0.0456671,0.2856811,0.0304919,0.6151971],"study_design_scores_gemma":[0.001233227,0.001251709,0.0004563506,0.0005597718,0.0001914391,0.0004245721,0.002257594,0.8769921,0.005303404,0.09174637,0.01900338,0.0005801012],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09434935,0.008749652,0.8791858,0.01499478,0.001134822,0.0006680347,0.00002481141,0.0001922852,0.0007004633],"genre_scores_gemma":[0.7944111,0.00002047229,0.2037668,0.00055717,0.0003017345,0.00000187977,0.000002655195,0.0000064561,0.0009317147],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8713626,"threshold_uncertainty_score":0.6629856,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4313363893","doi":"10.22148/001c.46662","title":"Are Computational Literary Studies Structuralist?","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Structuralism (philosophy of science); Narrative; Epistemology; Literary criticism; Presentation (obstetrics); Relation (database); Literary theory; Sociology; Philosophy; Computer science; Linguistics","retraction":null,"screen_n_in":null,"score":{"opus":0.1411680910892733,"gpt":0.3041771103153833,"spread":0.16300901922611,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.000170346,0.000111012,0.0002561613,0.00009397336,0.000594059,0.0004450637,0.0002268878,0.000009148191,0.00156763],"category_scores_gemma":[0.00005509283,0.00007328399,0.0002055946,0.00004486536,0.0001540343,0.0008152916,0.00009615473,0.0003036537,0.000005024236],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001036208,"about_ca_system_score_gemma":0.00002315816,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000001998682,"about_ca_topic_score_gemma":0.00002217931,"domain_scores_codex":[0.9988353,0.00004893769,0.0004329101,0.00007028328,0.0004851342,0.0001274213],"domain_scores_gemma":[0.9984866,0.00006756457,0.0006499947,0.00005457357,0.0006863179,0.00005496082],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000181253,0.0001952618,0.001360937,0.0001048845,0.001265732,0.000570722,0.1749484,0.01308046,0.00000729015,0.5361276,0.2685044,0.003653123],"study_design_scores_gemma":[0.0005033113,0.0003446315,0.001106235,0.00005310513,0.0001247408,0.000322058,0.1388469,0.0002914912,0.000002950743,0.2227996,0.635349,0.0002560047],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9836583,0.004211089,0.000001886345,0.001605368,0.001332957,0.00004543559,0.0001223522,0.00001688675,0.009005705],"genre_scores_gemma":[0.9846882,0.00001896095,0.00006215028,0.001088105,0.0008175079,8.87799e-7,0.00001334133,0.000007323318,0.01330353],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3668446,"threshold_uncertainty_score":0.9993451,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3005998127","doi":"10.22148/16.034","title":"A BLAST-based, Language-agnostic Text Reuse Algorithm with a MARKUS Implementation and Sequence Alignment Optimized for Large Chinese Corpora","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Computer science; Reuse; Natural language processing; Phrase; Sequence (biology); Artificial intelligence; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.01158736095800556,"gpt":0.3090124364441694,"spread":0.2974250754861639,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004421076,0.0001831802,0.0003053477,0.0001225645,0.00007245774,0.0002323124,0.0005808324,0.00004945395,0.00001524381],"category_scores_gemma":[0.0001222139,0.000104539,0.00009059459,0.000302438,0.00002799344,0.0007186912,0.0001198862,0.0001695337,0.000001564879],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001012225,"about_ca_system_score_gemma":0.0001105271,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001620211,"about_ca_topic_score_gemma":0.00001028883,"domain_scores_codex":[0.9987382,0.00004417521,0.0003807585,0.0002004407,0.0004008738,0.0002356053],"domain_scores_gemma":[0.998405,0.0001505976,0.000610895,0.0002643191,0.0004601569,0.0001089997],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.005115679,0.002779701,0.04659068,0.003265742,0.003814781,0.00489786,0.05369989,0.01604754,0.1889545,0.02903188,0.03569426,0.6101075],"study_design_scores_gemma":[0.01769361,0.005526351,0.001163092,0.001093711,0.0005699162,0.002721467,0.004679005,0.9208737,0.03315221,0.009682055,0.001294888,0.001549991],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1864178,0.0009082883,0.8104522,0.001605819,0.0001023621,0.0004088714,0.00003107634,0.00006110004,0.00001244069],"genre_scores_gemma":[0.4133109,0.00002206607,0.5862447,0.0002777987,0.00004669969,0.000004354605,0.000008425473,0.000007299993,0.00007770504],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9048262,"threshold_uncertainty_score":0.4262978,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3129389071","doi":"10.22148/001c.21182","title":"Cultural Capitals: Modeling Minor European Literature","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Translation Studies and Practices","field":"Arts and Humanities","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Minor (academic); Hierarchy; German; History; Media studies; Literature; Sociology; Political science; Art; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.09238203772860684,"gpt":0.2961933228713019,"spread":0.2038112851426951,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001906922,0.0001283626,0.0002302232,0.00004023284,0.0002647583,0.0007251269,0.000117558,0.00002421454,0.000425168],"category_scores_gemma":[0.00004700436,0.00007479249,0.0002575091,0.00006895447,0.00005265328,0.001009438,0.00002324029,0.0002896914,0.00003018895],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001660492,"about_ca_system_score_gemma":0.00002162302,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000008327123,"about_ca_topic_score_gemma":0.0001027228,"domain_scores_codex":[0.9989668,0.00009333705,0.0004493821,0.00009074353,0.00026079,0.0001389555],"domain_scores_gemma":[0.9982497,0.00004333293,0.0002820485,0.00007465212,0.001276695,0.00007354448],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0003094525,0.0005411318,0.0004786256,0.0003809211,0.003428569,0.002678046,0.653809,0.02366053,0.001848362,0.1312947,0.141675,0.03989569],"study_design_scores_gemma":[0.0004054208,0.00006294937,0.00002269926,0.0001594744,0.0002451343,0.00024124,0.04373733,0.002867893,0.00003799951,0.0004393774,0.9515992,0.00018127],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.4436743,0.08113819,0.0005548189,0.01647146,0.003140877,0.00009785902,0.00008866387,0.00005853914,0.4547753],"genre_scores_gemma":[0.9858758,0.00285229,0.0005621804,0.0004285465,0.002342657,1.470289e-7,0.0000149513,0.00001023216,0.007913144],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8099242,"threshold_uncertainty_score":0.6992415,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4283727265","doi":"10.22148/001c.36446","title":"Literary value in the era of big data. Operationalizing critical distance in professional and non-professional reviews","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Communication and Language","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Operationalization; Literary criticism; Value (mathematics); Literary science; Sociology; Criticism; Newspaper; Phenomenon; Praxis; Literary theory; Epistemology; Reading (process); Computer science; Literature; Media studies; Political science; Law; Art; Philosophy","retraction":null,"screen_n_in":null,"score":{"opus":0.08848718552736484,"gpt":0.371559383609055,"spread":0.2830721980816902,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001544739,0.0000729888,0.0002032613,0.00006787379,0.0001128536,0.00009477229,0.001378606,0.0000175478,0.00001496651],"category_scores_gemma":[0.0002782511,0.00004045657,0.00004605034,0.0004491862,0.00005773793,0.0008745493,0.0006585378,0.0005687787,4.963192e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003439899,"about_ca_system_score_gemma":0.0001219187,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000006315221,"about_ca_topic_score_gemma":0.00001430932,"domain_scores_codex":[0.998322,0.0004108332,0.000604532,0.0001104844,0.0004542166,0.00009796515],"domain_scores_gemma":[0.9990202,0.0002912967,0.0002183153,0.0003375732,0.000094875,0.00003777308],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002258304,0.002865263,0.01884215,0.0004195641,0.00007480842,0.0004468135,0.05839335,0.002383447,0.001692512,0.8015137,0.03767174,0.07547081],"study_design_scores_gemma":[0.003928688,0.000769838,0.1288263,0.003901489,0.00007816889,0.001571829,0.02286324,0.3062906,0.00011242,0.04314419,0.4874635,0.001049751],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7615052,0.06983381,0.01605087,0.1354349,0.003745686,0.001224081,0.0002075428,0.00002305882,0.01197485],"genre_scores_gemma":[0.9943363,0.0002279195,0.003788868,0.00137568,0.00004716133,0.000003417256,0.00001682961,0.000002111746,0.0002017039],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7583695,"threshold_uncertainty_score":0.2561816,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4211124067","doi":"10.22148/001c.32551","title":"Shakespeare and Company Project Data Sets","year":2022,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Shakespeare, Adaptation, and Literary Criticism","field":"Arts and Humanities","cited_by":8,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"University Research Committee, Emory University; Princeton University","keywords":"Bridge (graph theory); Work (physics); Set (abstract data type); Data set; Library science; History; World Wide Web; Computer science; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.1678217392894345,"gpt":0.3192285861083995,"spread":0.151406846818965,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0003037384,0.0001065398,0.0002184241,0.0001055067,0.0004732278,0.000293202,0.0003597702,0.00001272027,0.001198715],"category_scores_gemma":[0.00004753294,0.00007436768,0.00006819075,0.0000616157,0.0001111948,0.0007477751,0.0001376553,0.000267753,0.000003578735],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000244049,"about_ca_system_score_gemma":0.00006721276,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005749779,"about_ca_topic_score_gemma":0.0001033498,"domain_scores_codex":[0.9988867,0.00008050478,0.0003891597,0.0001198815,0.0003834192,0.000140307],"domain_scores_gemma":[0.9992137,0.00005707023,0.000229462,0.0001664326,0.0002710333,0.00006230567],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002511633,0.000393663,0.001455094,0.000148169,0.000620036,0.0004046994,0.3410642,0.0004002009,0.00008663459,0.1310205,0.5163043,0.00785144],"study_design_scores_gemma":[0.0004047518,0.0002086284,0.0004978753,0.0000221909,0.0001235008,0.0002902256,0.06613311,0.003212523,0.00000203117,0.0006610435,0.92831,0.0001341215],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9487361,0.00512875,0.00003222679,0.004614071,0.00176127,0.0002567687,0.00105271,0.00004721202,0.03837095],"genre_scores_gemma":[0.9907667,0.00007886449,0.0002401406,0.000535543,0.0007733814,8.162106e-7,0.0001105179,0.00001031287,0.007483694],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4120058,"threshold_uncertainty_score":0.9997143,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3174730804","doi":"10.22148/001c.25273","title":"Too isolated, too insular: American Literature and the World","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Historical Economic and Social Studies","field":"Economics, Econometrics and Finance","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"American literature; History; Identity (music); Politics; National identity; Principal (computer security); Political science; Literature; Law; Aesthetics; Art","retraction":null,"screen_n_in":null,"score":{"opus":0.02049754593415602,"gpt":0.2272385805091287,"spread":0.2067410345749727,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002899134,0.0001019654,0.0005567074,0.0000605311,0.0001733472,0.0001523417,0.0001245949,0.00003749766,0.00004583506],"category_scores_gemma":[0.0002417271,0.00006595276,0.0002369853,0.0004286868,0.0002186058,0.0001878124,0.0000488977,0.0002955758,0.000018035],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001240326,"about_ca_system_score_gemma":0.00001829896,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004264716,"about_ca_topic_score_gemma":0.00005934442,"domain_scores_codex":[0.9990889,0.00002453767,0.0005952464,0.0001156264,0.00003712703,0.0001385443],"domain_scores_gemma":[0.9989878,0.00008208118,0.0005942482,0.00008821632,0.0001721761,0.0000755039],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002744408,0.0001569135,0.01896667,0.000042472,0.001551938,0.0001870009,0.02131386,0.0002282727,0.00001770088,0.8561037,0.09463605,0.006521008],"study_design_scores_gemma":[0.001413495,0.00008456139,0.005718843,0.00002820725,0.00007215195,0.0000922178,0.001853353,0.0009457595,0.0000067732,0.03245523,0.9570887,0.0002407164],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.70572,0.1261415,0.001556114,0.074287,0.001894618,0.0001668656,0.00009276655,0.00002260856,0.09011853],"genre_scores_gemma":[0.9713144,0.006557294,0.0003887791,0.001858493,0.000421246,6.909615e-7,0.000002066284,0.00000732536,0.01944966],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8624526,"threshold_uncertainty_score":0.2689476,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2985974955","doi":"10.22148/16.050","title":"A Shared Task for the Digital Humanities Chapter 3: Description of Submitted Guidelines and Final Evaluation Results","year":2019,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Task (project management); Annotation; Reflection (computer programming); Narrative; Computer science; Digital humanities; Mathematics education; Information retrieval; Data science; Artificial intelligence; Psychology; World Wide Web; Linguistics; Programming language; Engineering; Philosophy","retraction":null,"screen_n_in":null,"score":{"opus":0.3084406244997505,"gpt":0.3152306524364105,"spread":0.006790027936659948,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005591805,0.0001171559,0.000218046,0.00007588842,0.0001629931,0.0009152458,0.0001527016,0.0000282026,0.0001745422],"category_scores_gemma":[0.0003072856,0.00006321221,0.0001750489,0.0000200627,0.0001529499,0.001405691,0.00002627215,0.0001064538,0.00000366272],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002693962,"about_ca_system_score_gemma":0.0000257079,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001505647,"about_ca_topic_score_gemma":0.0001133783,"domain_scores_codex":[0.9987152,0.00001424007,0.0006721123,0.00008439334,0.0004020453,0.0001120188],"domain_scores_gemma":[0.9969321,0.0001276143,0.0005437598,0.0000921898,0.00227569,0.00002857432],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.003894525,0.0003514416,0.002336359,0.0004247171,0.002025614,0.00001033841,0.1118472,0.001379724,0.0009557154,0.7557247,0.07210046,0.04894926],"study_design_scores_gemma":[0.007272746,0.003223716,0.0103555,0.0007787804,0.001440554,0.0001293613,0.1272807,0.02434484,0.0001185183,0.03711849,0.7871725,0.0007643027],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9794781,0.0008807393,0.00002161664,0.0008212765,0.0004616847,0.000276715,0.0002334207,0.000006862459,0.0178196],"genre_scores_gemma":[0.9863748,0.00004831356,0.00003921171,0.0001348418,0.0005477629,0.000001861205,0.00003299834,0.000007784608,0.01281243],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7186062,"threshold_uncertainty_score":0.8825736,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4242882238","doi":"10.22148/16.014","title":"Topic Modeling, Epistemology, and the English and German Novel","year":2017,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Contextualization; German; Context (archaeology); Historicism; Meaning (existential); Epistemology; Space (punctuation); Linguistics; Sociology; Philosophy; History; Interpretation (philosophy); Archaeology","retraction":null,"screen_n_in":null,"score":{"opus":0.07670709901819155,"gpt":0.4049405778109269,"spread":0.3282334787927353,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001255097,0.0000534561,0.0001940162,0.00002777652,0.0007951657,0.0002816149,0.0002124204,0.00003885748,0.00001060084],"category_scores_gemma":[0.0009835948,0.00002751841,0.00009553017,0.00003749718,0.0005040824,0.0002530473,0.00005421701,0.000135159,3.110495e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001560683,"about_ca_system_score_gemma":0.00003595664,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002925466,"about_ca_topic_score_gemma":0.0002879444,"domain_scores_codex":[0.9993219,0.0001005664,0.0002215691,0.000059342,0.0002080269,0.00008859498],"domain_scores_gemma":[0.998917,0.0001651137,0.0003196622,0.0000722201,0.0004535043,0.00007252951],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001842307,0.0001059372,0.02044865,0.00003519382,0.0009387586,0.00003478682,0.09773772,0.004028935,0.0001001849,0.8134148,0.001467286,0.06150351],"study_design_scores_gemma":[0.007036013,0.0001712483,0.05522994,0.000143134,0.001694889,0.0002070719,0.02800755,0.4657553,0.0000188687,0.3175487,0.1235604,0.0006268806],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.97106,0.0007545579,0.005141563,0.01556273,0.0003261647,0.00004811392,9.287316e-7,0.00000450079,0.007101404],"genre_scores_gemma":[0.9954559,0.0004508633,0.001845305,0.0001227922,0.0006709979,1.788735e-7,1.835942e-7,0.000001515118,0.001452275],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4958661,"threshold_uncertainty_score":0.6115852,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3153523203","doi":"10.22148/001c.22335","title":"Introduction: Contemporary Culture After Data Science","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Affordance; Analytics; Sociology; Cultural analytics; Data science; Epistemology; Social science; Computer science; Human–computer interaction; World Wide Web; Philosophy; The Internet","retraction":null,"screen_n_in":null,"score":{"opus":0.1232672934213013,"gpt":0.2890828548198972,"spread":0.1658155613985959,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0003659207,0.0001111733,0.0002105611,0.00006497731,0.0002688171,0.001602852,0.0005398634,0.00002369994,0.001958567],"category_scores_gemma":[0.0002054723,0.00006856412,0.0001074253,0.0001096481,0.000517439,0.00505863,0.0001409236,0.000284038,0.00003228462],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004536046,"about_ca_system_score_gemma":0.0002353349,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003547608,"about_ca_topic_score_gemma":0.00005423296,"domain_scores_codex":[0.99874,0.00002177713,0.0004029366,0.0001705955,0.0005045944,0.0001601096],"domain_scores_gemma":[0.9976045,0.00001318734,0.0002290468,0.0003144585,0.001725724,0.0001130144],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001210108,0.0002432858,0.0003399188,0.00005631271,0.0002490269,0.0007263847,0.0262366,0.0000192977,0.0004389888,0.2179541,0.7520463,0.00156876],"study_design_scores_gemma":[0.0001673279,0.00007784643,0.0001160616,0.00003675081,0.00004971862,0.0002285052,0.01914587,0.00002460111,0.0001839918,0.001183917,0.9786664,0.0001189706],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.3641061,0.01036901,0.00002179335,0.01608113,0.006522454,0.0001047407,0.0002382855,0.00004525425,0.6025112],"genre_scores_gemma":[0.942575,0.000093661,0.0001098536,0.0005503737,0.006125568,2.240273e-7,0.00002964912,0.000005970356,0.05050968],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.5784689,"threshold_uncertainty_score":0.9994336,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3040858410","doi":"10.22148/001c.13680","title":"Like Two Pis in a Pod: Author Similarity Across Time in the Ancient Greek Corpus","year":2020,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"Alfred P. Sloan Foundation; National Science Foundation","keywords":"Similarity (geometry); Divergence (linguistics); Focus (optics); Imitation; Range (aeronautics); Natural language processing; Reading (process); Feature (linguistics); Linguistics; Computer science; Artificial intelligence; History; Psychology","retraction":null,"screen_n_in":null,"score":{"opus":0.08545780393490782,"gpt":0.3457816242500205,"spread":0.2603238203151127,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001516638,0.000123161,0.0002686445,0.0000543278,0.00009429392,0.0002263586,0.0009412127,0.00006014251,0.00001060151],"category_scores_gemma":[0.0002916838,0.00007159697,0.0001489408,0.001088367,0.00004803925,0.0004876753,0.0001363494,0.0006548582,0.0000244382],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009059838,"about_ca_system_score_gemma":0.00008813373,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001532086,"about_ca_topic_score_gemma":0.00005531218,"domain_scores_codex":[0.9982576,0.000223332,0.0005660471,0.0001564112,0.0004980472,0.0002985725],"domain_scores_gemma":[0.9990408,0.0001097713,0.0003313609,0.0001453995,0.0002196667,0.0001529526],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001038895,0.001996305,0.2022032,0.0003015435,0.0004377284,0.01215318,0.4571145,0.1089597,0.01938217,0.03184259,0.1047204,0.05984978],"study_design_scores_gemma":[0.002575503,0.0006152355,0.05547602,0.0001448505,0.00004454334,0.0008438972,0.00439297,0.8948573,0.0007576405,0.00288914,0.03686895,0.0005338833],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8556786,0.0004988411,0.05519286,0.08769213,0.0004649278,0.0001835502,0.00001410781,0.0000305676,0.0002444127],"genre_scores_gemma":[0.9909091,0.00001759021,0.003909915,0.00485034,0.0001536858,5.439432e-7,0.000001594884,0.000003257294,0.0001539904],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7858976,"threshold_uncertainty_score":0.291964,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4402197367","doi":"10.22148/001c.121866","title":"Digital Humanities and Distributed Cognition: From a Lack of Theory to its Visual Augmentation","year":2024,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Language and cultural evolution","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Cognition; Cognitive science; Digital humanities; Conceptual framework; Visualization; Psychology; Sociology; Computer science; Epistemology; Cognitive psychology; Artificial intelligence; Social science","retraction":null,"screen_n_in":null,"score":{"opus":0.06055083490596529,"gpt":0.3632976391978363,"spread":0.302746804291871,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001631535,0.00005936232,0.0001228861,0.00003962534,0.0001044408,0.000248488,0.00006150744,0.00003778172,0.0001585097],"category_scores_gemma":[0.0001784895,0.00003754871,0.00007492314,0.0001820117,0.00008372335,0.0007764521,0.00001546556,0.00007782328,0.00001082177],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005872631,"about_ca_system_score_gemma":0.00004228329,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000332268,"about_ca_topic_score_gemma":0.00008743265,"domain_scores_codex":[0.9992896,0.00004686364,0.0002352796,0.00005893907,0.0002854064,0.00008389419],"domain_scores_gemma":[0.9993974,0.00008153163,0.0001066725,0.00001760731,0.0003414853,0.00005530995],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.001219386,0.0005843615,0.002454774,0.0004294247,0.003056964,0.0004799797,0.5493865,0.0005490743,0.0520745,0.1615674,0.07003352,0.1581641],"study_design_scores_gemma":[0.001807236,0.001405461,0.01478105,0.001799919,0.001896055,0.00009434714,0.7991282,0.00138909,0.007565167,0.04487829,0.1243713,0.0008838629],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9934707,0.001601177,0.0008896486,0.001074043,0.0002035226,0.00006862668,0.0001353745,0.0000156299,0.002541296],"genre_scores_gemma":[0.9974979,0.0001503179,0.00004725052,0.0000946925,0.0004536689,3.979104e-7,0.00004088126,0.000002547241,0.001712344],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2497417,"threshold_uncertainty_score":0.2396175,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2978801148","doi":"10.22148/16.024","title":"Nation, Ethnicity, and the Geography of British Fiction, 1880-1940","year":2018,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Travel Writing and Literature","field":"Arts and Humanities","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Unconscious mind; Ethnic group; Empire; History; First world war; Literature; Sociology; Geography; Anthropology; Ancient history; Art; Psychology; Psychoanalysis","retraction":null,"screen_n_in":null,"score":{"opus":0.02283525077622644,"gpt":0.2388033116570712,"spread":0.2159680608808448,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002982369,0.00006193849,0.0001752732,0.00005515408,0.000222115,0.0002108228,0.0001060907,0.00003249645,0.000119342],"category_scores_gemma":[0.00006444695,0.00003647285,0.0001381353,0.00006487279,0.0007400205,0.0002054233,0.00001461678,0.0001828361,0.000001053233],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000005469132,"about_ca_system_score_gemma":0.00001186242,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001653163,"about_ca_topic_score_gemma":0.000247792,"domain_scores_codex":[0.9992676,0.00004135771,0.0003498849,0.0000517189,0.0002083649,0.00008108049],"domain_scores_gemma":[0.9982678,0.00005374294,0.0003145694,0.0000518501,0.001276754,0.00003530118],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0004645158,0.0002873723,0.006802016,0.0002091713,0.001408069,0.00003314187,0.1323117,0.00006322149,0.0002109142,0.3570828,0.486578,0.01454916],"study_design_scores_gemma":[0.003067378,0.0005470925,0.02747869,0.0005253283,0.0005310262,0.0005856379,0.01839867,0.0005496045,0.0001244332,0.01288534,0.935012,0.0002948007],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9642075,0.003071881,0.00006113436,0.003048959,0.0007787096,0.00007166249,0.00005729816,0.0000150206,0.02868783],"genre_scores_gemma":[0.9942238,0.0002667779,0.00007843381,0.0002885,0.001730438,2.042912e-7,0.000002186399,0.000003937807,0.003405736],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.448434,"threshold_uncertainty_score":0.2726636,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3151065226","doi":"10.22148/001c.21993","title":"Imaginative Networks: Tracing Connections Among Early Modern Book Dedications","year":2021,"lang":"en","type":"article","venue":"Journal of Cultural Analytics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"","keywords":"Politics; Popularity; Centrality; Literature; History; Classics; Law; Art; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.03109520831080825,"gpt":0.295098267797732,"spread":0.2640030594869237,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003410949,0.00009816188,0.0001900546,0.00007442015,0.0002461563,0.0002398909,0.0003245626,0.00005774838,0.00003188961],"category_scores_gemma":[0.0002336251,0.00007547496,0.0001678169,0.0005752735,0.00007941511,0.001382826,0.00006315782,0.0004203534,0.000006314086],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008223462,"about_ca_system_score_gemma":0.0001298561,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000004166614,"about_ca_topic_score_gemma":0.000008099201,"domain_scores_codex":[0.998877,0.0001120476,0.0004228412,0.0001340128,0.0002706946,0.0001834205],"domain_scores_gemma":[0.9979333,0.0001162579,0.0003856372,0.0001559771,0.001247704,0.0001610942],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00004355334,0.0007106243,0.03394311,0.00006563248,0.001727326,0.001079963,0.03797247,0.3923397,0.003902078,0.4594924,0.03116033,0.03756283],"study_design_scores_gemma":[0.00043566,0.0000810282,0.01430773,0.0001195189,0.0001314516,0.0004561454,0.0008613285,0.9650978,0.002792826,0.01327155,0.002188637,0.0002563329],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03076529,0.001352938,0.9618062,0.005014315,0.0003615518,0.00003185528,0.000001675329,0.00002805296,0.0006381455],"genre_scores_gemma":[0.9888965,0.0001583765,0.009157754,0.000365426,0.0002285676,7.537849e-7,0.000002701349,0.000004087238,0.001185787],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9581313,"threshold_uncertainty_score":0.307778,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}