基于语义关联图的词义消歧方法  

Word Sense Disambiguation Method Based on Semantic Relationship Graph

作　　者：车超[1] 郑晓军[1] 

机构地区：[1]大连大学先进设计与智能计算省部共建教育部重点实验室,辽宁大连116622

出　　处：《科研信息化技术与应用》2014年第4期70-74,共5页E-science Technology & Application

基　　金：国家自然科学基金(61402068;61304206)

摘　　要：为了充分利用和整合词义消歧不同的知识库和语义资源,本文提出了一种基于语义关联图的方法。该方法以语义关系作为边,以词语的概念作为节点建立语义关联图,然后通过计算词义与上下文词语在语义关联图中的关联强度来确定歧义词的词义。语义关联图能够将多种消歧知识源比如词典、标注语料和生语料中的知识整合在一起,扩大了消歧知识的来源。本文方法在Senseval-3汉语词汇样本消歧任务中选择两个词,与一种基于知识的方法和有指导的方法进行了对比实验,实验结果验证了本文方法的有效性。In order to fully utilize and integrate different knowledge and semantic resources of Word Sense Disambiguation（WSD）, this paper proposes a WSD method based on semantic relationship graph.The method constructs the semantic relationship graph using semantic relationship as side and taking word sense as a node,and then determines the sense of ambiguous by calculating the strength of the relationship in semantic relationship graph between the word sense and its context. Semantic relationship graph can integrate a variety of disambiguation knowledge resources such as dictionaries, tagged corpus and raw corpus to expand the sources of disambiguation knowledge.We perform WSD experiment on two words of Chinese lexicon sample task ofSenseval-3 comparing with the method based on knowledge and the supervised method. The experiment result confirms the effectiveness of our method.

关 键 词：语义关联图 词义消歧 知网 

分 类 号：H146.3[语言文字—汉语]