Presents an automatic Chinese text proof-reading system based on word matching and syntax analysis, which uses the regulation-based and count-based methods to avoid the use of large scale corpus, and improves the RMM and the part corpus according to the input means of text, and then treats disperse strings found through word matching and syntax analysis , finds out error strings and corrects them through interaction between man and computer, and concludes from experimental results that 80% of errors can be found and about 5% are false-find errors, and the system can fulfill the general requirement for proof-reading.%提出了一种词匹配和语法分析相结合的中文文本自动校对法,采用规则与统计相结合的方法,不使用大规模语料库;而且根据原文的输入方式,用逆向最大匹配和局部语料统计的算法,找出散串,通过词匹配和语法分析处理散串,得到错误串的候选串,通过人机交互的方法对错误串进行自动校正. 实验表明,系统的查错率达80%以上,误报率在5%左右,基本满足了应用要求.
展开▼