A Methodology for Generative Spelling Correction via Natural Spelling Errors Emulation across Multiple Domains and Languages.
Nikita MartynovMark BaushenkoAnastasia KozlovaKaterina KolomeytsevaAleksandr AbramovAlena FenogenovaPublished in: CoRR (2023)
Keyphrases
- spelling correction
- english text
- multiple domains
- discriminative training
- context sensitive
- cross domain
- word sense disambiguation
- language identification
- hidden markov models
- natural language generation
- maximum likelihood
- probabilistic interpretation
- generative model
- log linear models
- search queries
- posterior probability
- language independent
- word level
- support vector machine
- learning algorithm
- cross lingual
- databases
- keywords
- machine learning