A content-based method for Persian real-word spell checking
Mohammad Hossein Samani, Zeinab Rahimi, Sara Rahimi · 2015
Here, a content-based method for real-word spell checking in Persian language is presented. In this method real-word mistakes are classified in 5 categories and are resolved using a content-based procedure. Each word which may cause any real-word error is listed in a candidate set in a same entry with its similar words (potential mistakes in a single entry). In next step, a content-word list is constructed based on adjacent frequent N-grams for each word in confusion set. Evaluations indicate that proposed method not only provides promising performance and acceptable precision, but also outperforms a similar existing system from precision and recall points of view.