{"id":"W3105424285","doi":"10.18653/v1/2020.emnlp-main.393","title":"Utility is in the Eye of the User: A Critique of NLP Leaderboards","year":2020,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":137,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Framing (construction); Inference; Artificial intelligence; Natural language processing; Machine learning","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0834595,0.001756976,0.002188122,0.003480125,0.004928741,0.01338816,0.007542246,0.01150257,0.00704162],"category_scores_gemma":[0.2634202,0.001377788,0.002209244,0.003775482,0.03945647,0.02472943,0.009041471,0.02168746,0.002598386],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.009534555,"about_ca_system_score_gemma":0.007303912,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008319877,"about_ca_topic_score_gemma":0.004932274,"domain_scores_codex":[0.9256954,0.05097388,0.002536615,0.007610002,0.01204424,0.001139917],"domain_scores_gemma":[0.6930494,0.255708,0.006505097,0.02652504,0.01545972,0.002752671],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00002981499,0.00001858944,0.0002894,0.0001220815,0.00002203218,0.00005922073,0.002131911,0.002785755,0.00005866815,0.9714697,0.01132719,0.01168567],"study_design_scores_gemma":[0.00003996157,0.00001664365,0.000105568,0.00016662,0.00001040708,0.00003782277,0.000411294,0.01218977,0.0001826845,0.9467349,0.04007539,0.00002901594],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"commentary","genre_scores_codex":[0.01104706,0.005921006,0.4692697,0.424247,0.00142126,0.0001693832,0.0004132931,0.001478198,0.0860332],"genre_scores_gemma":[0.6969238,0.004092801,0.2151457,0.05762985,0.004354284,0.001393213,0.0003125782,0.002839105,0.01730869],"genre_candidate":"commentary","genre_consensus":null,"teacher_disagreement_score":0.9165405,"threshold_uncertainty_score":0.4413813,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05896975066937212,"score_gpt":0.2934283674829232,"score_spread":0.2344586168135511,"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."}}