{"id":"W4416034841","doi":"10.18653/v1/2025.findings-emnlp.662","title":"Topic-Guided Reinforcement Learning with LLMs for Enhancing Multi-Document Summarization","year":2025,"lang":"","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Automatic summarization; Reinforcement learning; Reinforcement; Action (physics)","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":[0.001764976,0.0009571466,0.001387332,0.0008773245,0.0003991434,0.0008776685,0.001227656,0.0009892267,0.001462111],"category_scores_gemma":[0.00612138,0.000348933,0.0005131324,0.0009361432,0.0004580122,0.001886239,0.001109199,0.001415859,0.0007206282],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007987462,"about_ca_system_score_gemma":0.00101064,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002570144,"about_ca_topic_score_gemma":0.004338666,"domain_scores_codex":[0.9991637,0.0003247983,0.00005273936,0.0002326082,0.0001587643,0.00006731162],"domain_scores_gemma":[0.997621,0.001531241,0.0002462321,0.0001781296,0.0003206453,0.0001027721],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0003842869,0.0002899063,0.001900423,0.0003506521,0.0001395991,0.0001676353,0.0005355115,0.5215814,0.02087148,0.007090996,0.006993714,0.4396944],"study_design_scores_gemma":[0.00002320143,0.00006406048,0.0001396205,0.000008281077,0.00001792888,0.00001520224,0.00001928161,0.9926274,0.002340163,0.003784734,0.0009516053,0.000008573821],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02706659,0.001015652,0.9678717,0.0003096025,0.00007393491,0.0000674978,0.0001288795,0.002611805,0.0008542703],"genre_scores_gemma":[0.700952,0.0005571776,0.2921865,0.0004058511,0.0003077134,0.0002580548,0.0008762053,0.0004519899,0.004004411],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.002570144,"threshold_uncertainty_score":0.009334207,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02492005559603046,"score_gpt":0.2884353445884622,"score_spread":0.2635152889924317,"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."}}