R3F: RDF triple filtering method for efficient SPARQL query processing

Abstract

With the rapid growth in the amount of graph-structured Resource Description Framework (RDF) data, SPARQL query processing has received significant attention. The most important part of SPARQL query processing is its method of subgraph pattern matching. For this, most RDF stores use relation-based approaches, which can produce a vast number of redundant intermediate results during query evaluation. In order to address this problem, we propose an RDF Triple Filtering (R3F) method that exploits the graph-structural information of RDF data. We design a path-based index called the RDF Path index (RP-index) to efficiently provide filter data for the triple filtering. We also propose a relational operator called the RDF Filter (RFLT) that can conduct the triple filtering with little overhead compared to the original query processing. Through comprehensive experiments on large-scale RDF datasets, we demonstrate that R3F can effectively and efficiently reduce the number of redundant intermediate results and improve the query performance.