汉语信息处理中单字的构词方式与合成词的识别和理解

Word Formation and the Recognition of Compounds in Chinese Language Understanding

本文提出了汉语信息处理中关于单字构词方式的基本问题 ,考察了目前对于这个问题的研究和应用情况。认为现有的统计性结论在未登录词处理中对于揭示单字构词的规律缺乏有效的作用。究其原因 ,一是这些结论体现的是词素组合成词之后的结构性质 ,而不是组合过程中的规律 ;二是这些调查统计遵循以句法为本的观点 ,而合成词的结构方式主要是意合。按照意合的构词观点 ,词素组合成词的过程要受多种语言要素和非语言因素的制约。目前还只能运用不完备的构词知识识别未登录词。文章最后给出了一组构词规则的工程化应用实例。

The paper discusses the essential problems in the study of word formation in Chinese language processing. It is found that the current statistical conclusions have been far from effective on the recognition of unregistered Chinese compounds combined from single-syllable characters. Following the syntactic-based viewpoint, those statistical investigations present the structural properties of compounds instead of the way by which they are made up of, while actually most of the compounds are composed by meanings of each character, in the light of linguistic and nonlinguistic restrictions. At present, very limited knowledge of word formation have been expected to work on the recognition of unregistered Chinese compounds. As such an illustration, the paper gives a set of applicable rules and shows its performance in a Chinese new word recognition system.