Article ID Journal Published Year Pages File Type
974724 Physica A: Statistical Mechanics and its Applications 2009 10 Pages PDF
Abstract

We propose a general spectral method to find communities of a network based on network complement and anti-community concepts. Analytical and numerical results show that the eigenspace of matrices corresponding to a network complement reveals the community structure of a network more accurately than the eigenspace of matrices corresponding to the network itself. It is shown that the Laplacian eigenspace is the best candidate for spectral community detection especially in networks with a heterogeneous community structure. The method is applied to some computer-generated and real-world networks with known community structures.

Related Topics
Physical Sciences and Engineering Mathematics Mathematical Physics
Authors
, ,