| Summary: | 碩士 === 國立中正大學 === 資訊工程研究所 === 83 === Spelling Check is a frequently used function in English word processing applications. But for Chinese, this function is not easy to do, because, first, there are no word delimiters in Chinese writing, the sentences are composed by cotinuing Chinese characters. Second, many Chinese characters can be used as single-character words. Third, the characters can compose words with different character number. Besides, some short words can be composed to be the long words. All these features will bring some difficulties in Chinese word processing.
Thus, in spelling checking, we need to process word identification at the same time. We use dictionary lookup and long-word-first policy to identify words. The completeness of the wordbase (dictionary) will influence the correction rate of spelling check. So we use the relationship of characters in the articles to judge whether the character string can compose a word. By using this method to collect these words which are not in general wordbase, such as name of people, name of place and some terminologies, this method can help users create their own wordbase to improve the performance of spelling check.
Another topic of this project is the correction of spelling errors. Because many users use phonetic input-method which may cause many homophonic errors, we developed a method to correct this kind of errors.
|
|---|
