Relaxing Queries Based on XML Structure and Content Preferences

In order to resolve the problem of the empty or many answers returned from XML, this paper proposes a contextual preference query method of XML structural relaxation and content scoring. To provide users with most relevant and ranked query results, firstly, we propose a XML contextual preference (XCP) model, where all the possible relaxing queries are determined by the user's preferences. The XCP model allows users to express their interest on XML tree nodes, and then users assign interest scores to their interesting nodes for quickly providing best answers. Furthermore, based on the proposed XCP model, we propose a preference queries results ranking method, which includes: a Clusters_Merging algorithm to merge clusters based on the similarity of the context states, a Finding_Orders algorithm to find representative orders of the clusters, and a Top-k ranking algorithm to deal with the many answers problem. Results of preliminary user study demonstrate that our method can provide users with most relevant and ranked query results. The efficiency and effectiveness of the approach are also demonstrated by experimental results.