{"id":"W4293428470","doi":"10.1109/tai.2022.3201807","title":"Optimizing Multidocument Summarization by Blending Reinforcement Learning Policies","year":2022,"lang":"en","type":"article","venue":"IEEE Transactions on Artificial Intelligence","topic":"Topic Modeling","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"National Science Foundation","keywords":"Automatic summarization; Computer science; Reinforcement learning; Relevance (law); Redundancy (engineering); Multi-document summarization; Sentence; Artificial intelligence; Quality (philosophy); Machine learning; Information retrieval; Natural language processing","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":[0.0003831424,0.0001705066,0.0001371348,0.0002482845,0.001258044,0.0002005471,0.0006045958,0.00003582644,0.0002588854],"category_scores_gemma":[0.000009167278,0.00020095,0.00008581945,0.0005617246,0.00003417863,0.0004156909,0.00002091119,0.0004733849,0.00005297767],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003080885,"about_ca_system_score_gemma":0.00004935711,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002691812,"about_ca_topic_score_gemma":0.00001648986,"domain_scores_codex":[0.9980723,0.0001303958,0.0004545133,0.0004518884,0.000520952,0.0003699743],"domain_scores_gemma":[0.9992681,0.0001108215,0.0001203359,0.0003663939,0.00004871084,0.00008563301],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000009035955,0.00007043397,0.000001890016,0.000003634912,0.00001093837,0.000001524552,0.002681357,0.894944,0.005555036,0.007648324,0.00002286858,0.08905096],"study_design_scores_gemma":[0.00002571709,0.0001343652,3.479748e-7,0.00001002204,0.000006433656,0.000003902286,0.0008547333,0.8086455,0.1885203,0.0005209425,0.001092235,0.0001855066],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003565371,0.00002873912,0.9938204,0.0006671101,0.001056742,0.0002330393,0.000003066217,0.0002759464,0.0003495484],"genre_scores_gemma":[0.9860172,0.00003616251,0.01255216,0.0002438792,0.00003418651,0.0001137701,0.000003760701,0.00001566866,0.0009831996],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9824519,"threshold_uncertainty_score":0.9675986,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04336013915883275,"score_gpt":0.2878445156778042,"score_spread":0.2444843765189715,"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."}}