自动拼写校对的算法设计和系统实现

在英语作文机器评分中一项重要的环节便是拼写检查与纠错。本文依据实验室的作文自动评分项目,介绍了该项目中拼写检查与纠错模块,详述了该模块为错误单词提供有效纠错建议的算法。此算法主要针对中国大学生的英语拼写错误特点,通过统计CLEC(Chinese Learner English Corpus)中的单词信息而建立了一个基于CLEC的语言模型,以该算法对检查出为错误的单词给出纠错建议,此处的纠错建议即是候选的正确单词列表,位于第一位的就是我们要得到的系统认为最为准确的单词。最后为验证该算法,设计了一个简单的系统(命名为Myspell),并通过实验将该系统与GNU Aspell和Hunspell两个系统进行了对比,进一步肯定了该算法。

The Design and System Implementation of Automated Spelling Check Algorithm

Spelling check and error correction are of significance in machine rating for English compositions. Based on the laboratory's project of automated essay-scoring system, we introduce the module of spelling check and error correction of the project and elaborate the algorithm which is applied for correcting the wrong spelling of English words. The algorithm is considering the characteristics of the spelling errors of Chinese students. By collecting the vocabulary information from CLEC (Chinese Learner English Corpus), we establish a language model on the basis of CLEC. With the help of this algorithm, the correcting suggestion, which serves as the list of supplanting words closest to the meaning of the inputting one, can be given. The most suitable alternative word ranks on the top of that list. In order to verify the algorithm, we design a concise testing system named 'Myspell' and compare it with two similar systems, GNU Aspell and Hunspell, by experiment. Eventually, the algorithm obtains further affirmation.