{"id":"W4385573830","doi":"10.18653/v1/2022.findings-emnlp.76","title":"Faithful to the Document or to the World? Mitigating Hallucinations via Entity-Linked Knowledge in Abstractive Summarization","year":2022,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Automatic summarization; Computer science; Knowledge base; Information retrieval; Source text; World Wide Web; Data science; Artificial intelligence","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006082804,0.0008821069,0.0006543469,0.002293101,0.0008309056,0.003186625,0.001337804,0.001328695,0.00176356],"category_scores_gemma":[0.04328051,0.000438667,0.0006002345,0.001809747,0.001580619,0.01042958,0.003097533,0.002677856,0.0007414911],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007679207,"about_ca_system_score_gemma":0.0008248133,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002607343,"about_ca_topic_score_gemma":0.003874572,"domain_scores_codex":[0.9971041,0.001553922,0.0001946492,0.0004854605,0.0005177119,0.0001441191],"domain_scores_gemma":[0.9742213,0.01687339,0.002649399,0.004283275,0.001670026,0.0003025565],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001302659,0.0003799401,0.01476374,0.001512835,0.0004648025,0.001282458,0.01105547,0.1201436,0.03245336,0.04565757,0.01227632,0.7587072],"study_design_scores_gemma":[0.0001151191,0.0005258038,0.008296876,0.0004241444,0.0006092795,0.0007749973,0.004573898,0.730994,0.04852194,0.1732161,0.03174781,0.0002000589],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1140234,0.001992178,0.8735118,0.002866775,0.0001094255,0.0001905227,0.0008978773,0.002955072,0.003452921],"genre_scores_gemma":[0.7831123,0.00110289,0.2113578,0.0004451294,0.0001928859,0.0001083235,0.001862447,0.0002197671,0.001598532],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006082804,"threshold_uncertainty_score":0.03216934,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02367705475067786,"score_gpt":0.2922954430037961,"score_spread":0.2686183882531182,"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."}}