Mining Anomalous Sub-graphs in Graph Data Using Non-negative Matrix Factorization

  • N. N. R. Ranga Suri
  • Musti Narasimha Murty
  • Gopalasamy Athithan
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8251)


Mining graph data has been an important data mining task due to its significance in network analysis and many other contemporary applications. Detecting anomalies in graph data is challenging due to the unsupervised nature of the problem and the size of the data itself to be dealt with. Recent research efforts in this direction have explored graph data for identifying anomalous nodes and anomalous edges of a given graph. However, in many real life applications where the data is inherently networked in nature, the requirement is to detect anomalous sub-graphs with distinguishing characteristics such as near cliques, etc. In this context, we propose a novel method for addressing the anomalous sub-graph mining problem through community detection by employing the non-negative matrix factorization technique. Anomalous sub-graphs are identified by applying some existing techniques on the detected communities for measuring their deviation from the normal characteristics. We demonstrate the effectiveness of the proposed method through experimental evaluation on various benchmark graph data sets.


Data mining Mining graph data Anomalous sub-graphs Community detection 


  1. 1.
    Akoglu, L., McGlohon, M., Faloutsos, C.: Oddball: Spotting anomalies in weighted graphs. In: Zaki, M.J., Yu, J.X., Ravindran, B., Pudi, V. (eds.) PAKDD 2010. LNCS, vol. 6119, pp. 410–421. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  2. 2.
    Albanese, A., Pal, S.K., Petrosino, A.: Rough sets, kernel set and spatio-temporal outlier detection. IEEE Trans. on Knowledge and Data Engineering (2012) (online)Google Scholar
  3. 3.
    Chandola, V., Banerjee, A., Kumar, V.: Anomaly detection: A survey. ACM Computing Surveys 41(3), 15.1–15.58 (2009)Google Scholar
  4. 4.
    Girvan, M., Newman, M.E.J.: Community structure in social and biological networks. PNAS 99(12), 7821–7826 (2002)MathSciNetCrossRefzbMATHGoogle Scholar
  5. 5.
    Lee, D.D., Seung, H.S.: Learning the parts of objects by non-negative matrix factorization. Nature 401, 788–791 (1999)CrossRefGoogle Scholar
  6. 6.
    Leskovec, J.: Stanford network analysis platform, SNAP (2013),
  7. 7.
    McAuley, J., Leskovec, J.: Learning to discover social circles in ego networks. In: NIPS, Nevada, USA, pp. 548–556 (2012)Google Scholar
  8. 8.
    Noble, C.C., Cook, D.J.: Graph-based anomaly detection. In: Proc. SIGKDD, Washington, DC, USA, pp. 631–636 (August 2003)Google Scholar
  9. 9.
    Rattigan, M.J., Jensen, D.: The case for anomalous link discovery. SIGKDD Explorations 7(2), 41–47 (2006)CrossRefGoogle Scholar
  10. 10.
    Suri, N.N.R.R., Murty, M.N., Athithan, G.: Data mining techniques for outlier detection. In: Zhang, Q., Segall, R.S., Cao, M. (eds.) Visual Analytics and Interactive Technologies: Data, Text and Web Mining Applications, ch. 2, pp. 22–38. IGI Global, New York (2011)Google Scholar
  11. 11.
    Wang, F., Li, T., Wang, X., Zhu, S., Ding, C.: Community discovery using nonnegative matrix factorization. DMKD 22(3), 493–521 (2011)MathSciNetCrossRefzbMATHGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  • N. N. R. Ranga Suri
    • 1
  • Musti Narasimha Murty
    • 2
  • Gopalasamy Athithan
    • 1
    • 3
  1. 1.Centre for AI and Robotics (CAIR)BangaloreIndia
  2. 2.Dept of CSAIndian Institute of Science (IISc)BangaloreIndia
  3. 3.Presently working at Scientific Analysis Group (SAG)DelhiIndia

Personalised recommendations