{"id":"W2963903950","doi":"10.18653/v1/d16-1230","title":"How NOT To Evaluate Your Dialogue System: An Empirical Study of Unsupervised Evaluation Metrics for Dialogue Response Generation","year":2016,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":916,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal; McGill University","funders":"","keywords":"Computer science; Empirical research; Artificial intelligence; Machine learning; Natural language processing; Statistics; Mathematics","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":[0.006673118,0.0001679202,0.00028308,0.0004269074,0.0001186963,0.0001714551,0.0006975773,0.00008847511,0.000003489981],"category_scores_gemma":[0.00207707,0.000113781,0.000063953,0.0006043391,0.00001082707,0.0006733279,0.0001645678,0.00004046387,0.000008152426],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002584695,"about_ca_system_score_gemma":0.000234544,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007360114,"about_ca_topic_score_gemma":0.0001695687,"domain_scores_codex":[0.996129,0.001395629,0.0004658456,0.0006885901,0.001052992,0.0002679643],"domain_scores_gemma":[0.997132,0.0005350269,0.0001347645,0.001031848,0.001008711,0.0001576348],"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":[0.001712132,0.001895783,0.006831488,0.0001111407,0.0001888749,0.00001051969,0.02493049,0.01975474,0.4562955,0.00915315,0.001363337,0.4777528],"study_design_scores_gemma":[0.002695557,0.00167884,0.006000621,0.0000152464,0.00005970445,0.000002015784,0.0005347801,0.9783614,0.01021887,0.0001686121,0.00004782202,0.0002165055],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4955571,0.000004970046,0.5013325,0.001517159,0.0003406292,0.001170243,0.000006217076,0.00006380085,0.000007371614],"genre_scores_gemma":[0.9523298,5.445966e-7,0.04687208,0.0001559312,0.0002280359,0.0002718078,0.000006446245,0.0000136502,0.0001217226],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9586067,"threshold_uncertainty_score":0.4639854,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2550579332141903,"score_gpt":0.3930564024462317,"score_spread":0.1379984692320415,"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."}}