Research On Social Network Privacy Preserving Scheme

With the wide application of Internet in the world, people are not only restricted to the traditional way of social, but also use online social networks to social with friends in rapid development of science and technology era. Online social networks such as Facebook, Twitter and Google+ are among the most popular applications on the Internet. However, online social network applications severely suffer from various privacy exposures. The common privacy exposures mainly include: users' data privacy leakage and users' social attributes privacy leakage.Data encryption is a common solution to solve users' data privacy leakage that will bring with huge decryption time cost. However, the solution violates positive user experience, provided by social network service providers, because of the decryption time overhead of the encrypted data. To address privacy leak of users' social attributes, it is accustomed to disturb the structure of social network. However, disturbing the structure of social network is useless for privacy leak of users' social attributes in access control. Therefore, in the context of preserving privacy of users' data, this thesis introduces the solution of improving user experience. And a random walk based social attributes privacy preserving scheme is proposed. This scheme solves the problem of privacy leak of users' social attributes in access control effectively. Main work of this thesis is:1. A neighbor caching based privacy preserving scheme has been proposed to tackle privacy exposures with a special emphasis on the privacy of users' data. First of all, a social attribute-based encryption is used to encrypt users' data, and the encrypted data is stored in distribute hash table nodes. To provide positive user experience, which includes reducing the time of generating the updates lists and reducing the storage space, social activity is introduced as an evaluation indicator for every user. The computation of social activity is according to the number of common friends and the average online time. In addition, presence protocol, neighbor caching protocol and triangle relationship discovery algorithm are used to accelerate the process of users generating the updates lists. The experimental results and analysis demonstrate that our scheme can protect users' privacy more intensity and provide positive user experience.2. In this thesis, we also propose a random walker based social attributes privacy preserving scheme, which transforms data access control to a series of computation of social distance using random walk algorithm. The experimental results show that our scheme responds requests of visiting contents of links, not revealing visitors' social attributes.