Query expansion based on folksonomy tag co-occurrence analysis

Abstract

In traditional query expansion techniques, we choose the expansion terms based on their weights in the relevant documents. However, this kind of approaches does not take into account the semantic relationship between the original query terms and the expansion terms. Folksonomy is a social service in Web 2.0, which provides a large amount of social annotations. As the core of folksonomy, tags are high quality descriptors of the information contents and topics. Moreover, different tags describing the same information resource are semantically related to some extent. In this paper, we propose a query expansion method that utilizes the tag co-occurrence information to select the most appropriate expansion terms. Experimental results show that our tag co-occurrence-based query expansion technique consistently improves retrieval performance, compared with no-expansion method. This means the expansion terms we selected are semantically related to the original query, and tags of folksonomy will be the new resource of expansion terms.