SWRank: An Approach for Ranking Semantic Web Reversely and Consistently

The semantic Web is a typical link-based graph consisting of resources (labeled nodes) and relations (labeled edges). Ranking facility is required in order to assist human and intelligent agent in finding appropriate resources from massive knowledge organized in the semantic Web. The complexity lies in that relations are labeled and directed, and a huge number of fine-grained resources can describe either schema or instance data. We propose SWRank as a semantic Web ranking approach which applies an object-level link analysis ranking algorithm reversely to the direction of relations and consistently across the schema graph and the data graph. Our preliminary experiment shows that the approach has a similar converge property to and more reasonable ranking result than PageRank in evaluating the importance of individual resources in the semantic Web.