{"id":"W7131965811","doi":"","title":"WeBiText: building large heterogeneous translation memories from parallel Web content","year":2008,"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":"National Research Council Canada","funders":"Université du Québec en Outaouais","keywords":"Field (mathematics); Parallel corpora; Machine translation; Government (linguistics); Translation (biology); Cover (algebra); Web page","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":true,"about_ca":true,"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":[0.002414743,0.001445789,0.000899036,0.002906833,0.00131206,0.002270178,0.002069089,0.0008548145,0.007093452],"category_scores_gemma":[0.011874,0.0008158432,0.001238059,0.003678916,0.0009316569,0.005914654,0.002905241,0.001489385,0.005720158],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008539956,"about_ca_system_score_gemma":0.001976253,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007407858,"about_ca_topic_score_gemma":0.01290027,"domain_scores_codex":[0.9986052,0.0004443125,0.0001281408,0.0004634894,0.0002571126,0.0001017569],"domain_scores_gemma":[0.9939284,0.00271376,0.0003593164,0.001864962,0.0009393704,0.0001940611],"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":[0.001306173,0.0004422655,0.005490927,0.001567749,0.0004876254,0.00137619,0.002966144,0.02848582,0.0449126,0.01139346,0.04944469,0.8521264],"study_design_scores_gemma":[0.0004775946,0.001032132,0.008667016,0.0003860662,0.0007598949,0.002080472,0.005775212,0.5148429,0.1847669,0.04257278,0.2383316,0.0003073691],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.09852117,0.001102251,0.825707,0.0008291485,0.0002675762,0.0009032344,0.009919454,0.05354844,0.009201645],"genre_scores_gemma":[0.2323925,0.0006417059,0.7158431,0.0003332689,0.0001391465,0.0008539869,0.03630473,0.004760845,0.008730671],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007407858,"threshold_uncertainty_score":0.02372998,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03832649372604355,"score_gpt":0.2624293158410198,"score_spread":0.2241028221149763,"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."}}