Effective Pruning Techniques for Mining Quasi-Cliques

  • Authors:
  • Guimei Liu;Limsoon Wong

  • Affiliations:
  • School of Computing, National University of Singapore, Singapore,;School of Computing, National University of Singapore, Singapore,

  • Venue:
  • ECML PKDD '08 Proceedings of the European conference on Machine Learning and Knowledge Discovery in Databases - Part II
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Many real-world datasets, such as biological networks and social networks, can be modeled as graphs. It is interesting to discover densely connected subgraphs from these graphs, as such subgraphs represent groups of objects sharing some common properties. Several algorithms have been proposed to mine quasi-cliques from undirected graphs, but they have not fully utilized the minimum degree constraint for pruning. In this paper, we propose an efficient algorithm called Quickto find maximal quasi-cliques from undirected graphs. The Quickalgorithm uses several effective pruning techniques based on the degree of the vertices to prune unqualified vertices as early as possible, and these pruning techniques can be integrated into existing algorithms to improve their performance as well. Our experiment results show that Quickis orders of magnitude faster than previous work on mining quasi-cliques.