{"id":"W4399298003","doi":"10.1017/nlp.2024.5","title":"Calibration and context in human evaluation of machine translation","year":2024,"lang":"en","type":"article","venue":"Natural language processing.","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"National Research Council Canada","funders":"","keywords":"Calibration; Context (archaeology); Translation (biology); Machine translation; Computer science; Artificial intelligence; Natural language processing; Machine learning; Chemistry; Biology; Mathematics; Statistics; Biochemistry","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":[0.0008057994,0.0001293479,0.0001484569,0.0003046615,0.00005214424,0.0002191095,0.0002502378,0.0000939287,0.000009662463],"category_scores_gemma":[0.0000740608,0.0001055554,0.00002853208,0.0006007727,0.00004534278,0.001247403,0.0000494023,0.0002703717,4.160684e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006198776,"about_ca_system_score_gemma":0.00009510027,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001378503,"about_ca_topic_score_gemma":0.000156289,"domain_scores_codex":[0.9987186,0.00009431426,0.0002704202,0.0003285148,0.000447442,0.0001407156],"domain_scores_gemma":[0.9995787,0.00004992955,0.0000750336,0.0001495884,0.0001198197,0.00002697587],"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":[0.000003981127,0.00001247077,0.00005375931,0.0001979219,0.000003229616,0.000008534567,0.004062252,0.000003407231,0.01697714,0.001929277,0.00001244957,0.9767356],"study_design_scores_gemma":[0.000265834,0.00002622875,0.000104929,0.0003320733,0.00001594038,0.00001913333,0.00009079301,0.9762081,0.0190783,0.003697028,0.00002513235,0.0001364826],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"review","genre_gemma":"empirical","genre_scores_codex":[0.02864775,0.908371,0.06125342,0.0006087626,0.0001153465,0.0003176901,0.000003835963,0.0004560583,0.0002261606],"genre_scores_gemma":[0.968545,0.00001555315,0.03125207,0.00007371813,0.00003287249,0.00001757247,0.00002274291,0.00001096719,0.00002952541],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9765991,"threshold_uncertainty_score":0.4304424,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01995254238234069,"score_gpt":0.3295239246854237,"score_spread":0.3095713823030831,"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."}}