Hits:

Indexed by:期刊论文

Date of Publication:2012-11-01

Journal:Advances in Information Sciences and Service Sciences

Included Journals:EI、Scopus

Volume:4

Issue:20

Page Number:649-657

ISSN No.:19763700

Abstract:In this paper, we study the problem of identifying communities given a seed set. At first, we modify the Approximate Personalized PageRank (APPR) algorithm to find a community with small conductance while examining only a small number of vertices. Secondly, we apply the modified APPR algorithm repeatedly with different approximation ratios to compute multiple communities as candidate communities, and select the one with smallest conductance as a result community. We also utilize network community profile (NCP) for adjusting the parameters of our modified APPR algorithm and quantifying statistical properties of result communities in a global view. Both theoretical and experimental results show the efficiency of the proposed algorithm and the quality of the output.