An ant colony optimisation approach for optimising SPARQL queries by reordering triple patterns

Kalayci E. G., Kalayci T. E., Birant D.

Information Systems, vol.50, pp.51-68, 2015 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 50
  • Publication Date: 2015
  • Doi Number: 10.1016/
  • Journal Name: Information Systems
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.51-68
  • Keywords: SPARQL, Query optimisation, Reordering triple patterns, Ant colony optimisation, Ant system, MAX-MIN ant system
  • Dokuz Eylül University Affiliated: Yes


© 2015 Elsevier Ltd. All rights reserved.Processing the excessive volumes of information on the Web is an important issue. The Semantic Web paradigm has been proposed as the solution. However, this approach generates several challenges, such as query processing and optimisation. This paper proposes a novel approach for optimising SPARQL queries with different graph shapes. This new method reorders the triple patterns using Ant Colony Optimisation (ACO) algorithms. Reordering the triple patterns is a way of decreasing the execution times of the SPARQL queries. The proposed approach is focused on in-memory models of RDF data, and it optimises the SPARQL queries by means of Ant System, Elitist Ant System and MAX-MIN Ant System algorithms. The approach is implemented in the Apache Jena ARQ query engine, which is used for the experimentation, and the new method is compared with Normal Execution, Jena Reorder Algorithms, and the Stocker et al. Algorithms. All of the experiments are performed using the LUBM dataset for various shapes of queries, such as chain, star, cyclic, and chain-star. The first contribution is the real-time optimisation of SPARQL query triple pattern orders using ACO algorithms, and the second contribution is the concrete implementation for the ARQ query engine, which is a component of the widely used Semantic Web framework Apache Jena. The experiments demonstrate that the proposed method reduces the execution time of the queries significantly.