{"id":"W4411119799","doi":"10.18653/v1/2025.naacl-long.118","title":"Fine-Tuned LLMs are “Time Capsules” for Tracking Societal Bias Through Books","year":2025,"lang":"en","type":"article","venue":"","topic":"Hate Speech and Cyberbullying Detection","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Brock University","funders":"","keywords":"Tracking (education); Computer science; Political science; Sociology","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002308636,0.0001515341,0.0001849073,0.00005901951,0.0002845587,0.0002588202,0.0004729371,0.0001030061,0.00003399993],"category_scores_gemma":[0.00007346577,0.0001342869,0.0001789613,0.000297327,0.00003174055,0.0004157526,0.0001132311,0.0001151325,0.0001180424],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004674245,"about_ca_system_score_gemma":0.00008108526,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005271233,"about_ca_topic_score_gemma":0.00004760178,"domain_scores_codex":[0.9988353,0.00003024732,0.0002209135,0.0004108615,0.0001627607,0.0003399363],"domain_scores_gemma":[0.9992849,0.0001041831,0.00007798118,0.0003822311,0.0001069131,0.00004384604],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.00007266612,0.000355094,0.0005454341,0.0002574471,0.000355428,0.00004256885,0.002920528,0.00026156,0.0902845,0.1945887,0.3672435,0.3430726],"study_design_scores_gemma":[0.002666306,0.0003499405,0.00188423,0.0002332704,0.00006410403,0.00003204384,0.0003678754,0.1058962,0.6595049,0.04692301,0.1810612,0.001016912],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03024248,0.00014837,0.9372604,0.00310845,0.0008169305,0.0004101829,0.00000526894,0.0009069721,0.0271009],"genre_scores_gemma":[0.7918307,0.00001202715,0.1311901,0.002911136,0.0002800106,0.0001121669,0.00001105733,0.0000274475,0.07362538],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8060703,"threshold_uncertainty_score":0.5476061,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0365675797256866,"score_gpt":0.2754714269060195,"score_spread":0.2389038471803329,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}