中文摘要：

查询扩展技术通过向初始查询请求加入相似或相关的词,组成更为准确的扩展查询表达式,来减少查询请求与相关文献在表达上的不匹配现象,改善检索性能。与传统的查询扩展不同,XML查询扩展不仅要对文档内容进行有效扩展,而且还要考虑结构扩展。本文提出了一种基于伪反馈的XML查询扩展方法,将初始检索结果聚类,获得与查询请求最为相关的文档簇,然后在文档簇中抽取词组,找到符合用户查询意图的扩展查询词组,并在扩展查询词组的基础上进行结构扩展,最终形成完整的“内容＋结构”的查询扩展表达式。相关实验结果表明,相对没有扩展的查询,所提方法具有更好的精度。

英文摘要：

Query expansion technology can reduce word mismatch between query and related documents,construct more precise query expression and improve retrieval result through adding similar or related terms to original query.Compared with traditional query expansion,XML query expansion is different from traditional query expansion,which expands not only to document content,but also structural expansion.In this paper,a method of XML query expansion based on pseudo-relevance feedback is proposed.It firstly performs clustering algorithm to original retrieval results for finding the most relate clusters to user＇ s query.Secondly,the technology of information extraction is used to gain expanded key phrases that can satisfy the user＇ s intention.Finally,structural expansion is made based on the expanded key phrases and a full-edged content-structure query is formalized.Experimental results show that the proposed method outperform original query in precision and obtain the better retrieval quality.