基于多知识源的汉语词义消歧方法  

A Study on the Disambiguation of Chinese Words Based on Multi-Knowledge-Source

作　　者：张仰森[1] 黄改娟[1] 

机构地区：[1]北京信息科技大学

出　　处：《汉语学报》2008年第2期46-52,共7页Chinese Linguistics

基　　金：国家973基金项目(2004CB318102)资助;中国博士后科学基金项目(2005038026)资助;北京信息科技大学科研基金(55061039)资助

摘　　要：针对已有词义消歧研究方法的不足,分析了多种不同结构知识词典的可计算性及其计算复杂度,选择北大计算语言所的《现代汉语语法信息词典》、《现代汉语语义词典》和已经标注了词义的人民日报语料作为词义消歧知识源,研究异构知识源的知识获取与融合方法,并结合统计与规则方法构建词义消歧模型,取得了比较满意的词义消歧实验结果。In view of the shortcomings existed in the present methods of disambiguation, the computability and computing complexity of some dictionaries are analyzed in the paper. Taking Grammatical Knowledge-base of Contemporary Chinese, Semantic Knowledge-base of Modern Chinese and the sense-tagged corpus of People’s Daily as knowledge sources of disambiguation, the method of knowledge extraction and integration from heterogeneous knowledge sources is studied, and a disambiguation model is constructed.

关 键 词：多知识源 词义消歧方法 消歧模型 

分 类 号：H146[语言文字—汉语]