基于关系运算的汉语词汇语义相关度计算  

SEMANTIC CORRELATION COMPUTATION FOR CHINESE VOCABULARY BASED ON RELATIONAL OPERATIONS

作　　者：张瑞霞[1] 杨彬[1] 闫新庆[1] 杨国增[2] 

机构地区：[1]华北水利水电学院信息工程学院,河南郑州450011 [2]郑州师范学院数学系,河南郑州450044

出　　处：《计算机应用与软件》2013年第7期175-179,共5页Computer Applications and Software

基　　金：河南省科技攻关计划项目(102102210224)

摘　　要：提出一种基于关系运算的汉语词汇语义相关度计算方法。该方法首先以知网为语义资源,根据义原特征文件构造知识库概念图;然后从集合论角度对语义关系的运算进行研究,形式化不同语义关系间的运算规律;接着根据语义运算,提出不同情况下义原相关度的计算方法;最后根据知网知识词典,提出不同情况下词汇语义相关度的计算方法。该方法在计算过程中不但能够充分利用知网中的语义信息,而且考虑到了语义间蕴含的规律,实验证明其是有效的。In the paper a new method based on relational operations is put forward to compute the semantic correlation between Chinese words. First, it takes HowNet as the semantic resource, and constructs conceptual graph of knowledge library according to sememe feature files; Secondly, it studies the operations of semantic relations from the perspective of set theory and formalises the semantic operating rules between different semantic relations; After that, according to semantic operations, it puts forward the computation method of sememe correlation under different circumstances; Finally it presents the computation method of vocabulary＇ s semantic correlation under different situations according to the dictionary in HowNet. This approach can make full use of the semantic information in HowNet in computation process, moreover, it also takes into account the rules implicated between the semantics. Experiment proves it to he effective.

关 键 词：知网 概念图 语义相关度 语义关系 关系运算 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]