In this paper, we describe a system for correcting grammatical errors in texts written by non-native English learners. In our approach, a given sentence is sent to a number of modules, each focuses on a specific error type. The modules apply different approaches tailored to different types of errors, mostly based on probabilities of language models. A main program integrates corrections from these modules and outputs the corrected sen- tence. We evaluated our system on the official test data of the CoNLL-2014 shared task and obtained 0.36 in F-measure.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。