{"id":"W4392669868","doi":"10.18653/v1/2023.findings-ijcnlp.16","title":"The Glass Ceiling of Automatic Evaluation in Natural Language Generation","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Grand Équipement National De Calcul Intensif","keywords":"Computer science; Readability; Metric (unit); Fidelity; Field (mathematics); Rank (graph theory); Machine learning; Data mining; Natural language; Natural language generation; Artificial intelligence; Data science; Programming language; Engineering","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":[0.05551471,0.00232978,0.004121942,0.003183296,0.002148896,0.01096073,0.003570341,0.005753794,0.02276513],"category_scores_gemma":[0.1253439,0.001789884,0.001569902,0.001996655,0.005842511,0.01735499,0.007372505,0.008901615,0.01175725],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003381253,"about_ca_system_score_gemma":0.003351741,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007055924,"about_ca_topic_score_gemma":0.006322523,"domain_scores_codex":[0.9352949,0.04312854,0.002078935,0.005781226,0.0115205,0.002195799],"domain_scores_gemma":[0.8211507,0.1427703,0.001407567,0.01507924,0.01626101,0.003331204],"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":[0.001546026,0.0004823072,0.004135788,0.0008337683,0.0002403219,0.0001288762,0.0006977642,0.01091615,0.007206712,0.05043031,0.1616316,0.7617503],"study_design_scores_gemma":[0.0003050039,0.0009020455,0.003964518,0.0006536384,0.0001703704,0.0003813916,0.0007670926,0.6092973,0.01912925,0.2812136,0.08299794,0.0002178287],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03587066,0.02687467,0.8200018,0.04925959,0.004264695,0.0005491016,0.001741205,0.02738185,0.03405634],"genre_scores_gemma":[0.5742956,0.005218203,0.3606114,0.009677776,0.005834432,0.0008117749,0.004892592,0.008359581,0.03029857],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.05551471,"threshold_uncertainty_score":0.2935933,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04116435527696538,"score_gpt":0.3149507560117089,"score_spread":0.2737864007347435,"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."}}