Computing Relaxed Answers on RDF Databases

Database users may be frustrated by no answers returned when they pose a query on the database. In this paper, we study the problem of relaxing queries on RDF databases in order to acquire approximate answers. We address two problems for efficient query relaxation. First, to ensure the quality of answers, we compute the similarities of relaxed queries with regard to the original query and use them to score the potential relevant answers. We also propose the algorithm to get most relevant answers as soon as possible. Second, to optimise query relaxation process, we characterize a type of unnecessary relaxed queries which do not contribute to the final results and propose the method to prune them from the query relaxation graph. At last, we implement and experimentally evaluate our approach.