A Naive Bayes classifier for automatic correction of preposition and determiner errors in ESL text

摘要

This is the report for the CNGL ILT team entry to the HOO 2012 shared task. A Naive-Bayes-based classifier was used in the task which involved error detection and correction in ESL exam scripts. The features we use include n-grams of words and POS tags together with features based on the external Google N-Grams corpus. Our system placed 11th out of 14 teams for the detection and recognition tasks and 11th out of 13 teams for the correction task based on F-score for both preposition and determiner errors.