{"id":"W2608615602","doi":"10.18653/v1/d17-1263","title":"A Challenge Set Approach to Evaluating Machine Translation","year":2017,"lang":"en","type":"preprint","venue":"","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"ca_institutions":"National Research Council Canada","funders":"","keywords":"Machine translation; Computer science; Phrase; Set (abstract data type); Translation (biology); Natural language processing; Artificial intelligence; Divergence (linguistics); Bridge (graph theory); Strengths and weaknesses; Quality (philosophy); Neural system; Machine learning; Linguistics; Programming language; Psychology","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":[0.01503814,0.002278325,0.001847673,0.003482246,0.001511428,0.002531271,0.002560269,0.002552769,0.003145761],"category_scores_gemma":[0.06550848,0.000349942,0.001495259,0.002884523,0.001901875,0.00290919,0.00425156,0.003141013,0.0009648523],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00164215,"about_ca_system_score_gemma":0.001267905,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002794052,"about_ca_topic_score_gemma":0.003575531,"domain_scores_codex":[0.9813119,0.01290096,0.0007929314,0.00122837,0.003404977,0.0003609813],"domain_scores_gemma":[0.9604372,0.02838654,0.001458722,0.003312756,0.005538866,0.0008659713],"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":[0.003513107,0.002036868,0.03265211,0.002553101,0.002552401,0.0007652043,0.001734058,0.3599727,0.01668765,0.08654638,0.0752302,0.4157562],"study_design_scores_gemma":[0.0002350286,0.001400867,0.01278126,0.0001363248,0.0001853095,0.0004656286,0.0007318489,0.8419599,0.01260366,0.1103174,0.01903094,0.0001518408],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.2586008,0.00383736,0.695985,0.003993123,0.0009037321,0.001366165,0.008298992,0.001425547,0.02558919],"genre_scores_gemma":[0.711769,0.0004872828,0.2671731,0.0009690988,0.0004489539,0.002165797,0.01205814,0.0003979001,0.004530902],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01503814,"threshold_uncertainty_score":0.0795303,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1379582053225443,"score_gpt":0.3916650449870761,"score_spread":0.2537068396645318,"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."}}