{"id":"W3102645206","doi":"10.18653/v1/2020.emnlp-main.506","title":"Factual Error Correction for Abstractive Summarization Models","year":2020,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":124,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Automatic summarization; Computer science; Consistency (knowledge bases); Artificial intelligence; Heuristic; Machine learning; Natural language processing","routes":{"ca_aff":true,"ca_fund":true,"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":[0.002988867,0.001393147,0.000864869,0.001352143,0.0004678881,0.001593461,0.002167905,0.001340033,0.002704746],"category_scores_gemma":[0.01449995,0.0003543906,0.0007731086,0.0007526793,0.0005709507,0.002338433,0.001026193,0.001840769,0.001646418],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001185519,"about_ca_system_score_gemma":0.001225343,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004294541,"about_ca_topic_score_gemma":0.007291587,"domain_scores_codex":[0.998415,0.0004530158,0.0001936742,0.0005274768,0.0003249489,0.00008591939],"domain_scores_gemma":[0.9934874,0.00284489,0.000883603,0.001213144,0.001430716,0.000140342],"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":[0.0004304883,0.000193539,0.003452475,0.0006726644,0.0002677908,0.0002594552,0.0005743944,0.2674629,0.02318385,0.007608916,0.01981366,0.6760799],"study_design_scores_gemma":[0.00002305431,0.000113641,0.0007105112,0.00004477564,0.00006326482,0.00006835562,0.00005831069,0.971826,0.01629001,0.005274415,0.005504992,0.00002268096],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03817318,0.001680577,0.9372132,0.0008150996,0.0002797342,0.0002657457,0.001039018,0.01786499,0.002668468],"genre_scores_gemma":[0.5521455,0.0008639775,0.4320479,0.0004337544,0.0003347012,0.0003671607,0.004788333,0.0009385832,0.008080024],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004294541,"threshold_uncertainty_score":0.01580679,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08399768128692162,"score_gpt":0.2738264388879877,"score_spread":0.1898287576010661,"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."}}