基于用户相关反馈的带结构语义的XML查询词扩展

Query Expansion with Structural Semantics Based on Users＇ Feedback for XML Documents

在XML文档的信息检索中,检索质量不高的一个主要原因是用户难以提出准确描述其查询意图的查询表达式,而查询扩展技术被认为是可以帮助用户构建符合其查询意图的查询表达式。本文在XML信息检索中提出了基于用户相关反馈的查询扩展技术,在查询扩展中除了考虑词频因素外还充分考虑了XML文档的结构特点对于扩展查询词选取的影响,包括文档中元素的语义权重、元素所在层次和词项与初始查询词间的距离因素对于扩展查询词选取的影响。实验证明本方法是可行的,且能较好地提高检索结果的准确率。

In XML information retrieval, the main reason of lower precision is that our users can＇t submit a precise query expression for their query intensions. Query expansion can help user construct query expressions which can satisfy users＇ intentions, and improve precision effectively. This paper puts forward a new query expansion method based on Relevance Feedback. While expanding on keywords, we should consider the effects that structural characteristics affect the weight of keyword as well as the frequency of keyword. This paper has analyzed three factors which can affect the weight of keyword. One is the semantics of element in XML document, the second is the level of element, and the third is distance between keyword in initial query and keyword in XML document. Based on these factors, we put forward a formula to compute the weight of keyword, which is the biggest weight keyword as the result of query expression. Experiment results show that the above methods can obtain better retrieval result.