Leveraging Neighborhood Summaries for Efficient RDF Queries on RDBMS

Abstract: Using structural informations to summarize graph-structured RDF data is helpful in tackling query performance issues. However, leveraging structural indexes needs to revise or even redesign the internal of RDF systems. Given an RDF dataset that have already been bulk loaded into a relational RDF system, we aim at improving the query performance on such systems. We do so by summarizing neighborhood structures and encoding them into triples which can be managed along side the exist instance data. At query time, we optimally select the effective structural patterns, and adding these patterns to the existing queries to gain an improved query performance. Empirical evaluations shown the effectiveness of our method.