Chinese Word Spelling Correction Based on Rule Induction

Abstract

The importance of learning Chinese is increasing in the latest decades. However, the learning of Chinese is not easy for foreigners as a second language learning. Sometimes they write some text or document, but there always have many error words. So, how to detect the error word in document is becoming more then more important. This issue is very extensive, not only can help foreigners to learning Chinese but also can detect the error word. This paper had proposed method can divide five sections of structure: First sections are input sentence; second sections are parsing and word segmentation; third sections are fine the wrong word; forth sections are remove duplicate; fifth sections are final output. In this paper we use language model to detect Chinese spelling. It is had four part, E-Hownet, CKIP, similar pronunciation and shape dictionary, use the preset word to compare the word correction which in database. We use the bi-gram to promote our performance.