| Social media sites contain public and confidential information about their users. The quantity and the quality of this confidential information raise a privacy concern. In order to protect the users' privacy, social network graphs have to go through an anonymization process before being publicly published or released to a third party for data mining or statistical analysis. Many anonymization models have been proposed, each with different assumptions and settings regarding the information that needs protection and possible privacy attack scenarios. The ultimate goal of all the anonymization models is to preserve the privacy of the social network's users and, in the same time, preserve enough information to enable a good analysis of the social network. In this work, we study how well we can preserve the important features in a social graph, structure-wise and/or data content-wise while ensuring privacy with different anonymization models. |
