A Generalization Based Approach for Anonymizing Weighted Social Network Graphs

Abstract

The increasing popularity of social networks, such as online communities and telecommunication systems, has generated interesting knowledge discovery and data mining problems. Since social networks usually contain personal information of individuals, preserving privacy in the release of social network data becomes an important concern. An adversary can use many types of background knowledge to conduct an attack, such as topological structure and/or basic graph properties. Unfortunately, most of the previous studies on privacy preservation can deal with simple graphs only, and cannot be applied to weighted graphs. Since there exists numerous unique weight-based information in weighted graphs that can be used to attack a victim’s privacy, to resist such weight-based re-identification attacks becomes a great challenge. In this paper, we investigate the identity disclosure problem in weighted graphs. We propose k-possible anonymity to protect against weight-based attacks and develop a generalization based anonymization approach(named GA) to achieve k-possible anonymity for a weighted graph. Extensive experiments on real datasets show that the algorithm performs well in terms of protection it provides, and properties of the original weighed network can be recovered with relatively little bias through aggregation on a small number of sampled graphs.

The work is partially supported by the National Natural Science Foundation of China (Nos. 60973018, 60973020) and the Fundamental Research Funds for the Central Universities(Nos. N090504004, N100604013, N100704001, N090104001).