{"id":"W7132162559","doi":"","title":"Beyond correlation: making sense of the score differences of new MT evaluation metrics","year":2023,"lang":"en","type":"article","venue":"NPARC","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Intuition; Metric (unit); Heuristics; Rule of thumb; BLEU; Meaning (existential)","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"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":[0.08892427,0.001992856,0.00270057,0.009793767,0.002002687,0.01095997,0.003557029,0.002671261,0.002013237],"category_scores_gemma":[0.4619493,0.001012311,0.001168768,0.00982909,0.007062799,0.01837118,0.007140138,0.007603433,0.001396996],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002469252,"about_ca_system_score_gemma":0.002061568,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002028797,"about_ca_topic_score_gemma":0.002533436,"domain_scores_codex":[0.8935108,0.05957771,0.007294958,0.01404956,0.02424514,0.001321901],"domain_scores_gemma":[0.5204142,0.3396871,0.0317024,0.06944866,0.03604688,0.002700777],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001502484,0.0002938015,0.128171,0.001699588,0.002505202,0.0004613353,0.005340059,0.02958356,0.008956239,0.2396438,0.01871451,0.5631284],"study_design_scores_gemma":[0.0001751617,0.001228564,0.06016386,0.0008394959,0.0005279792,0.001219947,0.001578644,0.1573162,0.01427163,0.7370454,0.02516017,0.0004728857],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1076224,0.008260908,0.8490663,0.006664225,0.001898588,0.0003132054,0.001354518,0.002231454,0.02258844],"genre_scores_gemma":[0.761888,0.001188519,0.2283128,0.001948434,0.00149061,0.0005766263,0.001324607,0.001242112,0.002028392],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9110757,"threshold_uncertainty_score":0.4702821,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05685526509738319,"score_gpt":0.3181859076560996,"score_spread":0.2613306425587164,"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."}}