中文文本自动校对的语义级查错研究  

Study of Semantic Errors Checking in Automatic Proofreading for Chi nese Text

作　　者：骆卫华[1] 罗振声[1] 龚小谨[1] 

机构地区：[1]清华大学中文系,北京100084

出　　处：《计算机工程与应用》2003年第12期115-118,共4页Computer Engineering and Applications

摘　　要：目前中文文本自动校对的研究集中在词级和句法查错两方面,语义级查错仍是其中的薄弱环节。文章讨论了中文文本自动校对中的语义错误校对技术,综合使用了基于实例、基于统计和基于规则的搭配关系进行检查,提出统计和规则相结合的校对策略,既能检查局部语义限制,也能检查长距离的语义搭配,收到了较好的效果,也为中文自动校对的发展提供了新的思路。The study of Automatic proofreading for Chinese Text focuses on Word-level and Syntactic checking errors at present.Semantic Errors Checking is still a weak point.This paper discusses technology of semantic errors checking in automatic proofreading for Chinese text,uses collocation relationship based on instances,statistics and rules to check er-rors,and proposes proofreading strategy combining statistical approach and rules.It can check both local semantic re-straints and remote semantic collocations.The strategy achieves satisfactory results and provides new thoughts for future development of automatic proofreading for Chinese Text.

关 键 词：中文文本自动校对 基于实例 语义相邻矩阵 语义共现矩阵 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]