{"id":"W3208625967","doi":"10.1162/tacl_a_00427","title":"Lexically Aware Semi-Supervised Learning for OCR Post-Correction","year":2021,"lang":"en","type":"article","venue":"Transactions of the Association for Computational Linguistics","topic":"Handwritten Text Recognition Techniques","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"Government of Canada; National Endowment for the Humanities; National Science Foundation","keywords":"Computer science; Decoding methods; Optical character recognition; Consistency (knowledge bases); Artificial intelligence; Natural language processing; Language model; Raw data; Vocabulary; Error detection and correction; Machine learning; Speech recognition; Image (mathematics); Algorithm","routes":{"ca_aff":true,"ca_fund":true,"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":[0.0003921285,0.0001181406,0.0001850995,0.00009061831,0.000496484,0.00009705746,0.0003117512,0.0001229827,0.00001142456],"category_scores_gemma":[0.003247408,0.0001194742,0.0003161676,0.0003858344,0.00002075547,0.00009304576,0.00001996825,0.000185113,0.000003666884],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002164537,"about_ca_system_score_gemma":0.0003438531,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001090879,"about_ca_topic_score_gemma":0.00001573439,"domain_scores_codex":[0.9986709,0.0001021045,0.0004125609,0.0002509015,0.0003804115,0.0001830935],"domain_scores_gemma":[0.9929219,0.001556298,0.0003297672,0.0001768146,0.004972375,0.00004276812],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001513071,0.001304283,0.003183178,0.0004848385,0.000916824,0.000001820339,0.002003108,0.7578551,0.003335119,0.1497678,0.009348501,0.07164813],"study_design_scores_gemma":[0.001000422,0.0002057548,0.001808251,0.00008038197,0.0001307388,0.000007247158,0.0001004478,0.9144182,0.02336099,0.04334361,0.01526807,0.0002759109],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0004082115,0.00001378683,0.9957113,0.0009731967,0.001661366,0.0003858138,0.0001251783,0.000207206,0.0005139672],"genre_scores_gemma":[0.8372139,0.000006473968,0.159342,0.0002586471,0.0002013475,0.0000968961,0.0001706499,0.00001939862,0.002690698],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8368057,"threshold_uncertainty_score":0.4872017,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01389359865853384,"score_gpt":0.2640636733909112,"score_spread":0.2501700747323774,"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."}}