基于局部类别分析的查询扩展

Query expansion based on local classification analysis

针对查询扩展中局部分析方法查准率不高的缺点,提出一种新算法。该算法通过分析与用户查询密切相关的文档,从而得到与其相关的文档类别,进而根据相关类别中的文档用词与用户查询用词的共现关系对查询进行扩展。通过与传统的局部分析方法、全局分析方法的实验对比,结果表明新算法具有更快的检索速度和更高的查准率。

To solve the problem that local analysis for query expansion gets low precision in information retrieval, a new method was proposed. It analyzed the top-ranked documents retrieved for an original query, and inferred the possible classifications related to the query, then utilized co-occurrence with the query terms in the related classifications to expand the query. Experimental results indicate that the new method costs less retrieval time and gets higher precision compared with traditional methods such as global analysis and local analysis.