{"id":"W4200325561","doi":"10.1109/istas52410.2021.9629201","title":"Protecting marginalized communities by mitigating discrimination in toxic language detection","year":2021,"lang":"en","type":"article","venue":"","topic":"Hate Speech and Cyberbullying Detection","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Concordia University","funders":"","keywords":"Generalizability theory; Computer science; Language identification; Language model; Classifier (UML); Artificial intelligence; Machine learning; Identification (biology); Natural language processing; Natural language; Psychology","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005816308,0.001252082,0.001140545,0.001354477,0.001710956,0.001705126,0.001528404,0.001632384,0.001249974],"category_scores_gemma":[0.02372373,0.0004338359,0.0006623843,0.0006075375,0.001604241,0.004321051,0.004285765,0.002178139,0.0009264016],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008183374,"about_ca_system_score_gemma":0.001956043,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004681147,"about_ca_topic_score_gemma":0.007254329,"domain_scores_codex":[0.996712,0.001569572,0.0001125982,0.0006643107,0.0005857598,0.0003557064],"domain_scores_gemma":[0.9872953,0.005971564,0.001835966,0.002290696,0.001854215,0.0007523047],"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":[0.001370852,0.00207386,0.1322249,0.0005947073,0.0002385405,0.0007192548,0.004494314,0.1809338,0.04330657,0.02153949,0.01470947,0.5977942],"study_design_scores_gemma":[0.00003883819,0.0003874825,0.008526596,0.00006646767,0.00006936186,0.0002737988,0.001007516,0.9453866,0.01598568,0.02440318,0.003798651,0.00005577902],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5548346,0.000538588,0.4328517,0.002392182,0.0001370164,0.0003211979,0.0003610404,0.001921128,0.00664252],"genre_scores_gemma":[0.9604148,0.00008294117,0.03686421,0.0004526761,0.00006327149,0.00008404043,0.0002821637,0.00006743095,0.001688404],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.005816308,"threshold_uncertainty_score":0.03075993,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01360007845269604,"score_gpt":0.2412180201285734,"score_spread":0.2276179416758773,"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."}}