{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000973254,0.00002865784,0.00003781551,0.00005300768,0.00003831062,0.00003802259,0.0002072744,0.00001357281,0.000003389732],"category_scores_gemma":[0.0001111551,0.00001915791,0.00001311947,0.0002859861,0.000005380667,0.0001335844,0.00005802477,0.00003744397,0.00001166558],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002685316,"about_ca_system_score_gemma":0.00003502741,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004928398,"about_ca_topic_score_gemma":0.0002234077,"domain_scores_codex":[0.9993922,0.00006461817,0.0001384439,0.00009387726,0.0002285038,0.00008232599],"domain_scores_gemma":[0.9996396,0.00007971204,0.00003038984,0.0002080946,0.00003516279,0.000007013686],"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":[4.23571e-7,0.000007261388,0.0003851855,0.00001267147,0.000004608385,0.000001814692,0.004690465,0.03488513,0.02769323,0.0249073,0.0001713412,0.9072406],"study_design_scores_gemma":[0.00008019316,0.000002839489,0.00246668,0.000005368434,7.430026e-7,4.066921e-7,0.0001310245,0.9938492,0.002923014,0.0005114274,0.000006126382,0.00002297522],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9222153,0.0001341821,0.07614786,0.0006029478,0.0003255921,0.0001118446,5.263823e-8,0.00008485734,0.0003773138],"genre_scores_gemma":[0.9913125,0.000002749531,0.008457543,0.00002685595,0.00003207245,0.00000884072,0.000001583609,0.000001468383,0.0001563893],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.958964,"threshold_uncertainty_score":0.07812367,"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."}}