{"id":"W4283790221","doi":"10.1609/aaai.v36i10.21369","title":"Post-OCR Document Correction with Large Ensembles of Character Sequence-to-Sequence Models","year":2022,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Handwritten Text Recognition Techniques","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"ca_institutions":"Dalhousie University","funders":"Dalhousie University; Consejo de Investigación, Universidad Nacional de Salta; Consejo Nacional de Investigaciones Científicas y Técnicas; Compute Canada","keywords":"Sequence (biology); Computer science; Optical character recognition; Character (mathematics); Artificial intelligence; Code (set theory); Natural language processing; Task (project management); Process (computing); Pattern recognition (psychology); Speech recognition; Image (mathematics); Mathematics; Programming language","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001884896,0.001955379,0.001567188,0.001364589,0.0008217366,0.001227082,0.002318101,0.001644528,0.002658661],"category_scores_gemma":[0.006470019,0.0008004244,0.001662917,0.0009841402,0.0005782124,0.002401398,0.001382359,0.002888779,0.003247763],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00091318,"about_ca_system_score_gemma":0.001499816,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007720762,"about_ca_topic_score_gemma":0.01511075,"domain_scores_codex":[0.9982868,0.0002480172,0.0001013544,0.000593428,0.0006175181,0.0001528016],"domain_scores_gemma":[0.9954548,0.001272785,0.0003730941,0.001273137,0.001453462,0.0001727982],"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":[0.0004152894,0.0002078628,0.00336409,0.0001504255,0.0003157587,0.0002615209,0.0001856009,0.1994943,0.05335135,0.001656504,0.005738848,0.7348585],"study_design_scores_gemma":[0.00001021222,0.0001006244,0.0008941218,0.0000176641,0.00005271167,0.0001483604,0.00003517434,0.9615311,0.03128449,0.001948114,0.003947073,0.00003034532],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.039792,0.0007959533,0.9471685,0.0002619516,0.0002920106,0.0001283495,0.0002555643,0.009959968,0.001345699],"genre_scores_gemma":[0.4001292,0.0005120757,0.5840236,0.0005164639,0.000306493,0.0002124684,0.001878513,0.001526659,0.01089456],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007720762,"threshold_uncertainty_score":0.01535165,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07268029614619377,"score_gpt":0.2963970258287851,"score_spread":0.2237167296825914,"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."}}