概念向量空间模型在智能答疑系统中的应用  

Conceptual Vector Space Model for Intelligent Question Answering

作　　者：乌庆敏[1] 杨思春[1] 

机构地区：[1]安徽工业大学计算机学院,安徽马鞍山243002

出　　处：《安徽工业大学学报（自然科学版）》2008年第2期193-196,共4页Journal of Anhui University of Technology（Natural Science）

基　　金：安徽工业大学校级教改项目(2007JG22)

摘　　要：为解决智能答疑系统中因词的同义或多义现象而导致的"漏答"或"错答"问题,采用概念向量空间模型来计算句子相似度。针对概念向量空间模型生成的向量矩阵空间仍是一个稀疏大空间的缺点,提出了一种基于FAQ的概念向量空间模型来降低向量矩阵空间的维数。通过对基于事实的简单陈述问题的提问,结果显示明显优于向量空间模型。While extracting answers in intelligent question answering system, synonymy and polysemy will lead to losing correct answers or extracting wrong answers. In order to solve these problems, a method based on conceptual vector space model, is used to calculate similarity between sentences. Aiming to its shortcoming that the vector matrix space produced by it is a big sparse space, proposes an improved conceptual vector space model based on FAQ to reduce the dimension of its vector matrix space. With some simple fact statement questions, the results show that this method is better than vector space model obviously.

关 键 词：智能答疑 常问问题库 句子相似度 概念向量空间模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]