基于词语搭配关系的一种中文分词歧义性消除方法  

A CHINESE SEGMENTATION DISAMBIGUATION METHOD BASED ON WORD COLLOCATION RELATIONSHIP

作　　者：郭丙华[1] 俞亚堃[1] 李中华[2] 

机构地区：[1]肇庆学院电子信息工程系,广东肇庆526061 [2]中山大学信息科学与技术学院,广东广州510006

出　　处：《计算机应用与软件》2016年第10期94-97,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(61201087);广东省特色创新项目(2014KTSCX191)

摘　　要：汉语中词与词之间存在固定的搭配关系,基于词语搭配关系提出一种分词歧义性消除方法。该方法先利用正向和逆向最大匹配方法进行句子预切分,并对词的歧义性进行检测和词性标注,再对歧义词与词语搭配词典进行匹配或者动宾搭配判断,实现了较为准确的文档词语歧义性消除。通过词的歧义性检测实验和词语搭配检测对比实验,该方法取得了较好的效果。In Chinese there are the fix collocation relationships between words. This paper presents a disambiguation method for Chinese segmentation based on word collocation. It firstly pre-segments the sentences by using the forward maximum matching method and backward maximum matching method,and carries out the word ambiguity detection and tags the part of speech,and then it matches the ambiguous words with word collocation dictionary or makes distinguishment on verb-object collocations,thus achieves the more accurate results of document words disambiguation. The proposed method reaches good results as shown in contrast experiments of word ambiguity detection and word collocation detection.

关 键 词：词语搭配 最大匹配 中文分词 歧义性 动宾搭配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]