Finding maximal k-edge-connected subgraphs from a large graph

  • Authors:

  • Rui Zhou;Chengfei Liu;Jeffrey Xu Yu;Weifa Liang;Baichen Chen;Jianxin Li

  • Affiliations:

  • Swinburne University of Technology, Melbourne, Australia;Swinburne University of Technology, Melbourne, Australia;The Chinese University of Hong Kong, Hong Kong, China;Australian National University, Canberra, Australia;Australian National University, Canberra, Australia;Swinburne University of Technology, Melbourne, Australia

  • Venue:
  • Proceedings of the 15th International Conference on Extending Database Technology
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we study how to find maximal k-edge-connected subgraphs from a large graph. k-edge-connected subgraphs can be used to capture closely related vertices, and finding such vertex clusters is interesting in many applications, e. g., social network analysis, bioinformatics, web link research. Compared with other explicit structures for modeling vertex clusters, such as quasi-clique, k-core, which only set the requirement on vertex degrees, k-edge-connected subgraph further requires high connectivity within a subgraph (a stronger requirement), and hence defines a more closely related vertex cluster. To find maximal k-edge-connected subgraphs from a graph, a basic approach is to repeatedly apply minimum cut algorithm to the connected components of the input graph until all connected components are k-connected. However, the basic approach is very expensive if the input graph is large. To tackle the problem, we propose three major techniques: vertex reduction, edge reduction and cut pruning. These speed-up techniques are applied on top of the basic approach. We conduct extensive experiments and show that the speed-up techniques are very effective.