{"id":"W2952197960","doi":"","title":"Word Sense Disambiguation by Web Mining for Word Co-occurrence Probabilities","year":2004,"lang":"en","type":"preprint","venue":"NPARC","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"ca_institutions":"National Research Council Canada","funders":"University of Waterloo","keywords":"Computer science; Natural language processing; SemEval; Word (group theory); Artificial intelligence; Feature (linguistics); Task (project management); Novelty; Sample (material); Linguistics","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":true,"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":[0.002901091,0.001166923,0.001750253,0.01571041,0.001457506,0.00257552,0.00130877,0.001031053,0.002819404],"category_scores_gemma":[0.0153723,0.0009895641,0.001772468,0.01054755,0.000871575,0.003911213,0.001932117,0.001167161,0.00409258],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005536339,"about_ca_system_score_gemma":0.001390068,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00241998,"about_ca_topic_score_gemma":0.003277832,"domain_scores_codex":[0.9956345,0.0009840883,0.0005195124,0.001529523,0.001157559,0.0001747829],"domain_scores_gemma":[0.9928543,0.004640451,0.0006780127,0.00102698,0.0006689531,0.0001312475],"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":[0.0005839277,0.0003515233,0.02747904,0.0006824821,0.0005142319,0.001323396,0.0008463218,0.03721087,0.02312776,0.02101336,0.01656697,0.8703001],"study_design_scores_gemma":[0.0001241077,0.0001269928,0.01833693,0.0001275588,0.0002550856,0.002581104,0.0004926088,0.8065167,0.03580593,0.103253,0.03220033,0.0001795571],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04601166,0.0008619847,0.9359502,0.000276797,0.00009678563,0.0002908487,0.003444222,0.01075379,0.002313804],"genre_scores_gemma":[0.2470427,0.000558091,0.7432746,0.00007779057,0.0001156732,0.0003844041,0.006560951,0.0007409043,0.001244766],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01571041,"threshold_uncertainty_score":0.01534265,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02588644249646029,"score_gpt":0.3028768655376259,"score_spread":0.2769904230411656,"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."}}