The availability of social network data is indispensable for numerous types of research. Nevertheless, data owners are often reluctant to release social network data, as the release may reveal the private information of the individuals involved in the data. To address this problem, several techniques have been proposed to anonymize social networks for privacy preserving publications. To evaluate the privacy protection of existing techniques, this work presents an algorithm designed for de-anonymizing the anonymized graphs produced by the existing techniques. With experiments on a large set of anonymized graphs generated by the existing techniques, we demonstrate that our algorithm can re-identify a large portion of individuals in many anonymized graphs, which sheds light on their eectiveness and relative superiority
- J. Zhang, Y. Tang, X. Xiao, Y. Yang, Z. Zhang and M. Winslett. An Iterative Algorithm for Graph De-Anonymization. WSDM, 2013, poster.