Automatic RDF Query Generation from Person Related Heterogeneous Data

With the advance of the Semantic Web, the amount of data based on RDF is increasing rapidly on the Web. RDF data can easily merge one another because of its simple graphbased data model, but it is difficult to extract useful knowledge from data once merged because the query specifications require the knowledge of the whole graph structure. Therefore we propose Context Structure Matching (CSM) based on a method for extracting complex but characteristic frequent occurrence pattern called common query pattern by analysis of graph structure. CSM enables users who input a simple keyword to extract not only related information as subgraphs from pattern matching on the merged RDF data, but also similar results and comparison points by reusing the common query pattern. To evaluate the feasibility of the proposed method to merge real data of the Web, we perform an experiment with the data of W3C related people. Around 25,000 RDF triples are aggregated from multiple Web sites. As the results, after the merge process with minimal modification effort, CSM can extract more than 20 common query patterns which are useful to query the relationship between W3C related people and their interests.