{"id":"W2970600560","doi":"10.18653/v1/d19-1620","title":"Countering the Effects of Lead Bias in News Summarization via Multi-Stage Training and Auxiliary Losses","year":2019,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":34,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"Institut de Valorisation des Données; Compute Canada; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Automatic summarization; Computer science; Natural language processing; Artificial intelligence; Natural (archaeology); Joint (building); Natural language; Speech recognition; Engineering; History","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":[0.01074129,0.001606262,0.001768504,0.001241005,0.0009631761,0.001612727,0.00227161,0.002443939,0.00347586],"category_scores_gemma":[0.0400935,0.000997188,0.0007396336,0.001005616,0.001032015,0.00549692,0.00326907,0.004125142,0.001902682],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007924628,"about_ca_system_score_gemma":0.001336041,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00264887,"about_ca_topic_score_gemma":0.005840819,"domain_scores_codex":[0.9968724,0.001941927,0.0001640268,0.000522501,0.0002868087,0.000212425],"domain_scores_gemma":[0.9669429,0.02678842,0.001052738,0.002331899,0.002339419,0.0005445756],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.004509738,0.0007376542,0.006635208,0.0007169776,0.0004862806,0.0002808685,0.001036786,0.2463307,0.02110061,0.02175254,0.02735788,0.6690547],"study_design_scores_gemma":[0.0001937404,0.000341628,0.001212852,0.00004869543,0.0001242507,0.00005436939,0.00007105972,0.973328,0.006347025,0.01680147,0.001450497,0.00002635091],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1301457,0.003935071,0.8518272,0.002894205,0.0005394184,0.0001766407,0.0005452733,0.0048198,0.005116646],"genre_scores_gemma":[0.7979117,0.0007066944,0.1868211,0.001057141,0.0009189403,0.000263017,0.001722458,0.0009962282,0.009602643],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01074129,"threshold_uncertainty_score":0.05680609,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06762744025516274,"score_gpt":0.2830354277208046,"score_spread":0.2154079874656419,"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."}}