Privacy Preserving In Social Networks

With the fast development of the Internet, the functionality and diversity of Web have reached an unprecedented peak. The sharp increase of new Web concepts and applications marks the coming of Web 2.0 era. All of these convenient and brand-new services attract thousands of users to participate. The traditional Internet has gradually changed into a new category of communities– Social Network. Many users actively share their personal information which aggregates more and more extensive social network data. These vast amounts of network data, in turn, make users rely on social networks more heavily. The social network analysis has been widely applied in business, and many new data mining technologies have been proposed. At the same time, privacy preserving publishing of social networks data becomes a more and more important concern. While due to the different application setting, traditional privacy preserving methods in relational data cannot be applied directly. The wide spread of social network data without corresponding privacy protection posit users'personal information under serious safety threat. Therefore, privacy preserving in social networks has become a hot research topic recently.The research on privacy preserving in social networks comprise three sub-problems: (1) How to define privacy; (2) How to model the background knowledge; (3) How to describe data utility. Among these three problems, the third one is the most significant, but the solutions to the initial two problems will make a difference to the solution of the third one. This thesis provides a systematically survey on the current development of privacy preserving in social networks, and focus on the research and application of graph algorithm in social network based on the graph modeling social networks.This thesis summarizes the latest progress on the research of privacy preserving in social networks, classifies different privacy model, background knowledge and data utility, as well as proposed several effective algorithms. In this paper, our work focuses on the research of following aspects:1. Study on two ways to compute user's privacy score, and compared their effects in really applications. The privacy score is useful for personal information managing when privacy level changes.2. Propose an effective maximum clique algorithm for social network analysis. The algorithm is based on the power-law of social networks.3. Design and implement an effective algorithm for community participation in social networks.