基于向量空间的信息检索模型的改进  

Improvement of information retrieval model based on vector space

作　　者：原媛[1] 彭建华[1] 张汝云[1] 

机构地区：[1]国家数字交换系统工程技术研究中心,河南郑州450002

出　　处：《计算机工程与设计》2008年第23期6012-6015,共4页Computer Engineering and Design

基　　金：国家863高技术研究发展计划基金项目(2007AA01Z434);2007年度河南省青年科学基金项目(074100510023)。

摘　　要：向量空间模型是有代表性的信息检索模型之一,针对该模型存在的问题进行了研究和探讨。在仅用统计词频表示文档向量方面,引入知网作为语义知识库,提出基于概念的特征选择模型;在因词语的同义和多义,不能满足文档向量相互独立方面,提出潜在语义索引模型。实验验证了改进后的检索模型更能体现文本的内容,降低文本向量的维数,提高检索的准确率。Vector space model is one of the representative information retrieval models, the problems of the model are researched and discussed. On the aspect of the representing of text vector by statistical term frequency, HowNet is taken as semantic repository, and the model of feature selection based on concept is presented. On the aspect of dissatisfying the independence of text vector as the synonymy and polysemy of words, the model of latent semantic indexing is presented. The experiment demonstrates the improvement of retrieval model can expresses the contents of texts better, the dimensions of text vectors is reduced and the precision of retrieval is advanced.

关 键 词：向量空间模型 基于概念的特征选择 潜在语义索引 知网 奇异值分解 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]