{"id":"W3018695736","doi":"10.18653/v1/2020.acl-srw.16","title":"Considering Likelihood in NLP Classification Explanations with Occlusion and Language Modeling","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Banting and Best Diabetes Centre, University of Toronto; Bundesministerium für Bildung und Forschung","keywords":"Computer science; Artificial intelligence; Natural language processing; Language model; Context (archaeology); Occlusion","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"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":[0.01156886,0.00129509,0.001896351,0.003801161,0.001600632,0.004409556,0.003000655,0.003712092,0.003222651],"category_scores_gemma":[0.07256941,0.001273168,0.001821794,0.003385175,0.003447432,0.009158522,0.004894322,0.004560255,0.0006297196],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002539185,"about_ca_system_score_gemma":0.002133274,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006872055,"about_ca_topic_score_gemma":0.005735136,"domain_scores_codex":[0.9903603,0.005591438,0.0005018401,0.001524671,0.001600757,0.000420957],"domain_scores_gemma":[0.9306118,0.06048613,0.002994338,0.003396434,0.001880171,0.0006311151],"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":[0.0006265808,0.0002186461,0.02311463,0.0007615837,0.0003647112,0.00115959,0.004824877,0.307422,0.002773178,0.331995,0.0095703,0.3171689],"study_design_scores_gemma":[0.00004128405,0.0000396562,0.001247888,0.0000783311,0.00005438658,0.0002353682,0.0001858687,0.769981,0.001064021,0.2231367,0.003884563,0.00005091029],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01037383,0.0004210663,0.9868599,0.0008746778,0.00003062898,0.00007019947,0.000156537,0.0005568586,0.0006563275],"genre_scores_gemma":[0.4762663,0.0007465573,0.5177951,0.0005199919,0.0004568482,0.0004661429,0.001367635,0.0004581322,0.001923116],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01156886,"threshold_uncertainty_score":0.06118268,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0539578300993899,"score_gpt":0.2734041729442856,"score_spread":0.2194463428448957,"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."}}