分面搜索的分面推荐方法研究

ON METHOD OF FACET RECOMMENDATION IN XML FACETED SEARCH

分面搜索是一种探索式的信息查询技术,是对关键词搜索的补充。现有的分面搜索大都显示所有分面,当文档异构且存在大量分面时,不可能同时显示所有分面。研究将最有价值的分面推荐显示给用户是很有意义的,已有的分面推荐方法主要根据领域专家的经验选取,或者仅根据统计方法进行推荐,没有考虑XML分面之间的相关性,影响了推荐效果。研究XML文档分面搜索中的分面推荐方法,首先根据XML文档结构的语义特点,提出XML分面的相关性定义,然后将面的覆盖率方法和XML分面之间的相关性相结合进行XML分面推荐,并在公用数据集上进行了实验。实验结果表明,结合XML分面之间的相关性可以有效地推荐最有导航能力的分面。

Faceted search is a kind of exploratory information query technique, which is the complement to keyword search. Most of the existing faeeted searches display all facets, while the document is heterogeneous and in a large number, it is impossible to display them all simultaneously, so it is really make sense to study the recommendation of most valuable facets to users. Existing facet recommendation method mainly selects based on the experience of domain experts or recommends just according to statistics but without considering the correlation between XML facets, which affects the recommendation effect. In this article we study the method of facet recommendation in XML document faceted search. First, according to semantic features of XML document structure, we put forward the definition of XML facet correlation, and then we combine the coverage rate method of facet with the correlation between XML facets to carry out XML facet recommendation, and this has been experimented on public dataset. Experimental results show that, to combine the correlation between XML facets can effectively recommend the facet with best navigation ability.