Paper: A Noisy Channel Model Framework for Grammatical Correction

ACL ID W13-3615
Title A Noisy Channel Model Framework for Grammatical Correction
Venue International Conference on Computational Natural Language Learning
Session shared task
Year 2013
Authors

We report on the TOR system that partic- ipated in the 2013 CoNLL shared task on grammatical correction. The system was a provisional implementation of a beam search correction over a noisy channel model. Although the results on the shared task test set were poor, the approach may still be promising, as there are many as- pects of the current implementation that could be optimised. Grammatical correc- tion is inherently difficult both to perform and to evaluate. As such, possible im- provements to the evaluation are also dis- cussed.