基于大规模问答对数据的查询扩展技术研究

The Study on the Query Expansion Technique Based on Q＆A Data

查询扩展是信息检索领域中的一个热门话题,其目标是将与初始查询词相关的其他单词添加到初始查询请求中,以更详细地描述用户的信息需求.本文将查询过程视为特殊的问答过程,并基于此思想提出一种新的查询扩展方法.本文的贡献主要有以下几点：① 借助统计语言模型从大规模问答对数据中挖掘单词之间的扩展关系,并根据单词间的扩展关系对候选扩展词进行评级;② 提出一个新的查询扩展词选取策略,以克服已有查询扩展方法仅依赖评级的扩展词选取策略的不足.通过在真实数据集合上的实验,证明本文提出的查询扩展方法可以取得优于传统方法的性能,具有一定的实用性.

Query expansion technique is a hot topic in information retrieval. The aim of this technique is add terms in the original query to form a suitable query. In this article, we view the question part and the answer part of Q＆A pairs in a huge Q＆A archive as query and the webpages that satisfied the information need in the query. The contribution of this paper is： （1） using a statistical language model to mine the expansion probability between words, and （2） propose a candidate expansion words selection approach to form the new query to avoid the shortcomings of the prior query expansion methods. The experimental results on a real data set show that our approach performs better than the traditional query expansion techniques.