Grass: An efficient method for RDF subgraph matching

Abstract

Resource Description Framework (RDF) is a standard data model of the Semantic Web, and it has been widely adopted in various domains in recent years for data and knowledge representation. Unlike queries on relational databases, most of queries applied on RDF data are known as graph queries, expressed in the SPARQL language. Subgraph matching, a basic SPARQL operation, is known to be NP-complete. Coupled with the rapidly increasing volumes of RDF data, it makes efficient graph query processing a very challenging problem. This paper primarily focuses on providing an index scheme and corresponding algorithms that support the efficient solution of such queries. We present a subgraph matching query engine based on the FFD-index which is an indexing mechanism encoding a star subgraph into a bit string. A SPARQL query graph is decomposed into several star query subgraphs which can be efficiently processed benefiting from succinct FFD-index data structure. Extensive evaluation shows that our approach outperforms RDF-3X and gStore on solving subgraph matching.