An Efficient Probabilistic Approach to Network Community Mining

摘要

A network community refers to a group of vertices within which the links are dense but between which they are sparse. A network community mining problem (NCMP) is the problem to find all such communities from a given network. A wide variety of applications can be formalized as NCMPs such as complex network analysis, Web pages clustering as well as image segmentation. How to solve a NCMP efficiently and accurately remains an open challenge. Distinct from other works, the paper addresses the problem from a probabilistic perspective and presents an efficient algorithm that can linearly scale to the size of networks based on a proposed Markov random walk model. The proposed algorithm is strictly tested against several benchmark networks including a semantic social network. The experimental results show its good performance with respect to both speed and accuracy.