基于《〈知网〉》词汇语义相关度计算的消歧方法

歧义字段的处理是汉语自动分词系统中重要而困难的问题之一.歧义处理正确率的高低直接影响着分词的质量.为了解决汉语自动分词系统中的歧义问题,本文基于<知网>的义原关系,在分析了影响词语语义相关度的主要因素后,引入相关度计算,利用<知网>中义原之间的纵向和横向关系及实例因素计算出不同词性的相关度,根据相似度的对称性计算实例的影响因素,由此来提高语义相关度的准确率,同时针对所提出的方法给出了实例验证.实验结果表明,该方法计算得到的语义相关度结果更加合理,为消除汉语自动分词中的歧义问题提供了良好的支撑依据,也使得汉语自动分词系统处理歧义字段更有效.

Disambiguating Method for Computing Relevancy Based on HowNet Semantic Knowledge

Disambiguating word is important and difficult in Chinese automatic segmenting system.Its validity directly influences the quality of word segmenting.In order to solve the problems of disambiguation,based on the relevance of semantic knowledge resource of HowNet,this paper analyzes the main influencing factors of semantic relevance.By Introducing relevancy, the paper uses factor of examples and the relationship between horizontal and vertical directions of HowNet semantic knowledge resource to compute the r...