一种提高文本检索准确性的关联方法  

A CORRELATION METHOD FOR ENHANCING ACCURACY IN TEXT RETRIEVAL

作　　者：施侃晟[1] 刘海涛[1] 舒平达[2] 

机构地区：[1]上海交通大学,上海200240 [2]宁波大学,浙江宁波315211

出　　处：《计算机应用与软件》2010年第5期1-2,23,共3页Computer Applications and Software

基　　金：国家自然科学基金(60373000)

摘　　要：文本检索就是从文本集中发现与用户查询相关的文本的过程。传统的基于查询似然检索模型没有考虑词项之间内在关系和外在共现关系。针对这些缺点,分别提出新关联方法分别予以解决。最后,通过实验对比证明新方法在查询精度上有了明显的提高,验证了新方法的有效性。Text retrieval is the process to find the texts relevant to a query of users in whole text sets.The inherent relation and exterior co-occurrence relation between word items are not considered in likelihood retrieval model of traditional query method.To tackle these drawbacks,in this paper we propose new correlation methods to deal with them respectively.Our new methods have been proven to have significant improvement in query accuracy at last by the contrast made through the experiment,its effectiveness is validated as well.

关 键 词：文本检索 统计语言模型 查询似然 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]