{"id":"W3121217868","doi":"10.1016/j.is.2021.101718","title":"Multi-label legal document classification: A deep learning-based approach with label-attention and domain-specific pre-training","year":2021,"lang":"en","type":"article","venue":"Information Systems","topic":"Text and Document Classification Technologies","field":"Computer Science","cited_by":73,"is_retracted":false,"has_abstract":false,"ca_institutions":"Thomson Reuters (Canada)","funders":"","keywords":"Computer science; Artificial intelligence; Multi-label classification; Domain (mathematical analysis); Training (meteorology); Machine learning; Training set","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":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0004799287,0.000196509,0.0002139572,0.0002374666,0.0003335836,0.00156437,0.0003775302,0.0001274306,0.000004134928],"category_scores_gemma":[0.00002346437,0.0001694723,0.0000292433,0.0006576794,0.00008871951,0.003115146,0.00009989538,0.0002179944,0.00005584566],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001204263,"about_ca_system_score_gemma":0.0001132541,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000008880105,"about_ca_topic_score_gemma":0.000002981439,"domain_scores_codex":[0.9981939,0.0001204331,0.00057619,0.0003307719,0.0005142342,0.0002645055],"domain_scores_gemma":[0.9985068,0.00005603382,0.0004582877,0.0005418338,0.0003545514,0.00008245977],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00005427674,0.000295074,0.003239909,0.0004453867,0.0001070932,0.0000107401,0.009492532,0.004399923,0.003346255,0.7562264,0.0005176134,0.2218647],"study_design_scores_gemma":[0.003862281,0.00019336,0.01247016,0.0001405036,0.00001300107,0.0001597312,0.01561235,0.7643647,0.0003320122,0.000116964,0.2021708,0.0005641083],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01468044,0.0003703678,0.9808335,0.0008071798,0.0002119854,0.0004559892,0.000002154409,0.0007042507,0.001934143],"genre_scores_gemma":[0.843629,0.00003499851,0.1554356,0.00008078929,0.00002634424,0.0002681687,0.00009810821,0.000008532191,0.0004184518],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8289486,"threshold_uncertainty_score":0.9994721,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03582749930198761,"score_gpt":0.2545681606306928,"score_spread":0.2187406613287052,"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."}}