{"id":"W3186932802","doi":"10.1145/3462757.3466144","title":"Labels distribution matters in performance achieved in legal judgment prediction tasks","year":2021,"lang":"en","type":"article","venue":"","topic":"Artificial Intelligence in Law","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal","funders":"","keywords":"Lawsuit; Computer science; Task (project management); Artificial intelligence; Transformer; Representation (politics); Natural language processing; Intersection (aeronautics); Machine learning; Law; Political science; Engineering","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":[0.0005989802,0.00006485906,0.00009107271,0.00003518608,0.0001517437,0.00006493333,0.0001082168,0.0000807133,0.0005665315],"category_scores_gemma":[0.00009480316,0.00006890029,0.00002318919,0.00048392,0.0001280603,0.0004530504,0.00003454484,0.0001463655,0.00009600369],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004180927,"about_ca_system_score_gemma":0.0001395625,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006053583,"about_ca_topic_score_gemma":0.03255369,"domain_scores_codex":[0.9988448,0.0001435608,0.0002609185,0.0001942774,0.0002690774,0.0002873389],"domain_scores_gemma":[0.999712,0.00004848499,0.00003404728,0.0001050432,0.00005079079,0.00004963486],"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":[0.0001008026,0.0007814519,0.6525466,0.00004128577,0.00001701417,0.00008532718,0.01836568,0.003362443,0.006490843,0.2572939,0.009618175,0.05129652],"study_design_scores_gemma":[0.0007136079,0.0001808914,0.419594,0.0003345967,0.00001944857,0.000006229142,0.03123741,0.01485681,0.09896374,0.004781604,0.4284766,0.0008350685],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.974328,0.00001790908,0.001156133,0.009045939,0.0004379094,0.000168999,0.00001338432,0.00004979002,0.01478197],"genre_scores_gemma":[0.9974544,0.00007444216,0.0001148364,0.0004688744,0.00007872393,0.00002178124,0.00003850953,0.000004008449,0.001744435],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4188584,"threshold_uncertainty_score":0.9850997,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03060566189741133,"score_gpt":0.3112505753430578,"score_spread":0.2806449134456465,"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."}}