Candidate re-ranking for SMT-based grammatical error correction.
- Submitting institution
-
University of Cambridge
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1876
- Type
- E - Conference contribution
- DOI
-
-
- Title of conference / published proceedings
- BEA@NAACL-HLT
- First page
- 256
- Volume
- -
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- -
- Year of publication
- 2016
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
2
- Research group(s)
-
-
- Citation count
- -
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Statistical machine translation models for grammatical error correction (GEC) output a set of possible translations of a 'bad' English sentence which are ranked using a language model trained on 'good' English. This is the first paper to propose a supervised approach to reranking this output, which considerably improved system performance on GEC at the time. Reranking is now a standard component of GEC systems based on neural machine translation models, as well as for weakly supervised approaches which utilise language models directly.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -