A local-perturbation anonymizing approach to preserving community structure in released social networks

Abstract

Social networks provide a large amount of social network data, which is gathered and released for various purposes. Since social network data usually contains much sensitive information of individuals, the data needs to be anonymized before releasing. To protect privacy of individuals in released social network, many anonymizing methods have been proposed. However, most of them were proposed for general purpose, and suffered the over-information loss problem when they were used for specific purposes. In this paper, we focus on the problem of preserving structure information in anonymized social network data, which is the most important knowledge for community analysis. Furthermore, we propose a novel local-perturbation technique that can reach the same privacy requirement of k-anonymity, while minimizing the impact on community structure. We evaluate the performance of our method on real-world data. Experimental results show that our method has less community structure information loss compared with existing techniques.