Benchmarking Graph Databases on the Problem of Community Detection

  • Sotirios Beis
  • Symeon Papadopoulos
  • Yiannis Kompatsiaris
Conference paper

DOI: 10.1007/978-3-319-10518-5_1

Volume 312 of the book series Advances in Intelligent Systems and Computing (AISC)
Cite this paper as:
Beis S., Papadopoulos S., Kompatsiaris Y. (2015) Benchmarking Graph Databases on the Problem of Community Detection. In: Bassiliades N. et al. (eds) New Trends in Database and Information Systems II. Advances in Intelligent Systems and Computing, vol 312. Springer, Cham

Abstract

Thanks to the proliferation of Online Social Networks (OSNs) and Linked Data, graph data have been constantly increasing, reaching massive scales and complexity. Thus, tools to store and manage such data efficiently are absolutely essential. To address this problem, various technologies have been employed, such as relational, object and graph databases. In this paper we present a benchmark that evaluates graph databases with a set of workloads, inspired from OSN mining use case scenarios. In addition to standard network operations, the paper focuses on the problem of community detection and we propose the adaptation of the Louvain method on top of graph databases. The paper reports a comprehensive comparative evaluation between three popular graph databases, Titan, OrientDB and Neo4j. Our experimental results show that in the current development status Neo4j is the most efficient graph database for most of the employed workloads, while Titan handles better single insertion operations.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer International Publishing Switzerland 2015

Authors and Affiliations

  • Sotirios Beis
    • 1
  • Symeon Papadopoulos
    • 1
  • Yiannis Kompatsiaris
    • 1
  1. 1.Information Technologies InstituteCERTHThermiGreece