Community Hiding by Link Perturbation in Social Networks

Abstract

Complex social network is a kind of relationship system composed of many nodes according to social relations. Community detection helps scholars to understand this network topology and find out meaningful communities. Many scholars are therefore actively exploring new community detection algorithms. However, it brings privacy issues such as the disclosure of personal or group information of community members and goes against an individual or group desire to be hidden. Hence, how to hide a target community in a network to resist the community detection algorithms becomes critical. Given this, this article studies the community hiding problem, which has not been extensively focused so far and aims to properly hide a target community into other communities by perturbing a budget limited number of social links. First, we formalize the community hiding problem and design a gain function. Second, we prove the feasibility of link perturbation operations and propose an efficient algorithm to solve the community hiding problem. Finally, we conduct extensive experiments on varieties of real small- and large-scale social networks and compare our method with other community deception algorithms. The experimental results demonstrate that the proposed algorithm is more efficient than previously proposed algorithms.