Research On Social Network Privacy Protection Based On K-symmetric Anonymous Algorithm

In recent years, with the continuous development of the Internet technology, social networking prod-ucts have been integrated into our lives. Social network software is becoming an indispensable part of ourlife, such as QQ, renren, weibo and WeChat. However, while we are enjoying the convenience the socialnetworks provides, a new challenge is proposed in the protection of personal privacy and social relationshipprivacy. Although there have been many achievements in traditional relational database in privacy protec-tion technology, social network data is different from relational data, whose model is graph type structure,so the social network privacy protection cannot apply the traditional relational database privacy protectionmethod mechanically. It is unrealistic to deal with the massive social network data relying on traditional ar-tificial method, and using computer technology to analyze and research the data is an inevitable trend. Be-cause the main aim of using the social network is to participate in social activities and share informationresources with other users, a simple personal information privacy can’t meet this demand. The study in pro-tection of personal privacy social relations is a hot spot currently.This thesis focuses on the research of k-anonymous algorithm in the perspective of data mining. First,it summarizes the current research status of social network privacy protection, introduces the concept andcharacteristics of social network and analyzes several kinds of attack against social network, sketches thesocial network anonymous posting method at the present stage. On this basis, we draw lessons from others,and improved the original k-symmetrical anonymity algorithm. The improved algorithm can be applied toreal life and we design an effective reduction algorithm, find a formula to deduce the k value. k-anonym-ous method is a symmetric algorithm privacy, since nodes in social networks are processed symmetrically,each set of the results in equivalence classes includes k nodes, so that the probability for each attacker toidentify target is not higher than1/k.A reduction algorithm restoring the original social network graph isput forward based on the usability analysis of k–symmetrical anonymous methods. Finally, based on thesocial network data of WeChat discussion groups, the thesis implements the k-symmetrical anonymousposting, and evaluates the usability of the anonymous Posting method, and verifies its effectiveness.