Query expansion based on books tags of user's social relations in XML book search

摘要

Queries provided by users are usually too fuzzy and simple, which cannot provide sufficient evidence for the book search system to find relevant books. Previous query expansion focused on expanding similar words for query, which may not represent different user real-time interested. Since the user friends and interesting libraries may represent the interesting book types for the user for a certain stage, this paper proposed a novel method by finding expanded words from the books tags marked by user social relations in their personal digital library. This paper presents two query expansion methods: Query expansion based on books tags of user friends; Query expansion based on books tags of user interesting libraries. Experiments on the actual book dataset show that the two methods greatly improve the nDCG@10 values of the search results and the user satisfaction.