{"id":"W4413036724","doi":"10.1016/j.mlwa.2025.100717","title":"Cross-domain fairness audit of sentiment label bias in foundation models: Comparing human and machine annotations on tweets and reviews","year":2025,"lang":"en","type":"article","venue":"Machine Learning with Applications","topic":"Sentiment Analysis and Opinion Mining","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"York University; Vector Institute; Humber Polytechnic; Brock University","funders":"","keywords":"Foundation (evidence); Audit; Domain (mathematical analysis); Computer science; Artificial intelligence; Information retrieval; Natural language processing; Machine learning; Business; Accounting; Mathematics; Political science","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":[0.0005535234,0.0001408934,0.0002788025,0.0003311081,0.0003603966,0.0001653153,0.0002055471,0.00002962422,0.000007359638],"category_scores_gemma":[0.00001774336,0.0001233366,0.00002600061,0.0007563399,0.00006971158,0.0002176213,0.0001408468,0.000187511,0.000003846136],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003150981,"about_ca_system_score_gemma":0.00001887821,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001833438,"about_ca_topic_score_gemma":0.0001803335,"domain_scores_codex":[0.9987968,0.0001150187,0.0004039757,0.0003954561,0.0001514311,0.0001373723],"domain_scores_gemma":[0.999231,0.0001113337,0.0002530867,0.0003050193,0.00005806687,0.00004143177],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001812545,0.0003157429,0.607316,0.0001540584,0.00007547492,8.38437e-7,0.001431464,0.04078025,0.00084877,0.2993144,0.00003915276,0.04970574],"study_design_scores_gemma":[0.001295025,0.00008093031,0.09608889,0.0002478618,0.00003421379,0.000001956167,0.00006374413,0.8914099,0.0001343671,0.005013787,0.005422309,0.0002070598],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2509335,0.000867426,0.7448297,0.0007911478,0.00001657027,0.0005427761,0.000002041245,0.00005539782,0.001961521],"genre_scores_gemma":[0.9784012,0.0001019924,0.02072291,0.00007281242,0.00001008499,0.0001678696,0.00007548209,0.000007320365,0.0004403262],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8506296,"threshold_uncertainty_score":0.5029522,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06347665065719738,"score_gpt":0.341585826618393,"score_spread":0.2781091759611956,"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."}}