Efficient SPARQL Query Processing Based on Adjacent-Predicate Structure Index

Abstract

RDF is the standard data model for the Semantic Web, SPARQL is the standard query language for RDF data. In recent years, a large amount of RDF data has been released, SPARQL query processing has become a research hotspot. However, the existing methods such as RP-filter and R3F still have some problems due to the complexity of RDF query graph, the index rely excessively on one-way predicate paths that resulting in path redundancy. Concerning the problem, RDF Adjacent-Predicate Structure (RAPS) index is proposed. Firstly, the incoming-predicate path tree is built for vertices in data graph, based on this, the out-going predicate path is added into the predicate path tree. Then, vertices in the data graph are classified and stored according to their adjacent predicate structures. The predicate path tree is used to find the candidate vertices as the preliminary search space for the query vertices. Then, the search space are narrowed down by the out-going predicate path to find the matching vertices of the query vertices. Finally, the sub-graph where the matching vertex is located is output as the query result. Compared the query response times of RAPS, RDF-3X and R3F in different data sets. The experimental results show that, RAPS has less query response time and higher query processing efficiency in processing complex query graphs.