{"id":"W4404395182","doi":"10.2196/60272","title":"Enhancing Bias Assessment for Complex Term Groups in Language Embedding Models: Quantitative Comparison of Methods","year":2024,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Computer science; Artificial intelligence; Natural language processing; Word embedding; Measure (data warehouse); Word2vec; Machine learning; Embedding; Association test; Term (time); Data mining","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.08726801,0.001847226,0.001142788,0.00410594,0.0007225629,0.002676065,0.001254331,0.0019855,0.002490488],"category_scores_gemma":[0.247938,0.000439378,0.002301418,0.002299892,0.001665903,0.004525192,0.003749337,0.002267077,0.0005515992],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001447077,"about_ca_system_score_gemma":0.001796391,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001905485,"about_ca_topic_score_gemma":0.002099666,"domain_scores_codex":[0.9578152,0.03041616,0.003045253,0.003159193,0.0050806,0.0004834876],"domain_scores_gemma":[0.5627746,0.3954342,0.01001817,0.01457028,0.01595897,0.001243784],"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":[0.007550231,0.001108413,0.1297345,0.00400575,0.006144039,0.0001642412,0.002480388,0.1768474,0.006445348,0.01784906,0.007361219,0.6403095],"study_design_scores_gemma":[0.0007162102,0.0028478,0.04078424,0.001189695,0.001595609,0.0003417591,0.001310693,0.8883845,0.01300752,0.04345946,0.006019703,0.000342864],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2921415,0.008765895,0.6889445,0.0008999521,0.0006379826,0.001332177,0.001692445,0.001272597,0.004312966],"genre_scores_gemma":[0.7033762,0.0009380502,0.290898,0.0002270508,0.0001977125,0.001337774,0.002051606,0.0003575416,0.0006160708],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.08726801,"threshold_uncertainty_score":0.4615229,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1731788120296335,"score_gpt":0.5031536617832251,"score_spread":0.3299748497535915,"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."}}