基于短语结构文法的分词研究  

Study of Segmentation Based on Phrase-Structured Grammar

作　　者：韩世欣 王开铸[1] 

机构地区：[1]哈尔滨工业大学计算机系

出　　处：《中文信息学报》1992年第3期48-54,共7页Journal of Chinese Information Processing

摘　　要：本文在分析现有各种分词算法的基础上,遵循自然语言理解的层次模型,着重对短语结构文法及其形式化描述进行了探讨。提出了汉语自动分词在短语层的理解模型及短语层分词语义相关原则,最后给出了短语结构分词法(PsG法)的机器实现。实验结果表明,PSG法在提高分词精度上具有相当好的效果。In this paper, we mainly discuss phrase-structured grammar (PSG) and its formalized description according to the hierarchic model of natural language understanding on the basis of analyzing five segmentation algorithms,and present the understanding model and semantic relativity rule in phrase-level in Chinese automatic segmentation. At last, the machine implementation of FSC method is given. The result of experiment shows that PSG method has a very good effect on increasing the accuracy of segmentation.

关 键 词：短语结构文法 分词 自然语言 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]