The Privacy Protection And The Study Of The Usability Of Social Network

With the rapid development of social networks, more and more people attend to the new social party, they communicate and share information with each other by social network websites, so that personal information and social network relations those are left from the networks attract many researcher to study and develop the valuable wealth in it, because it has very important commercial value.Therefore,How to mining more valuable information from the networks, and ensure the privacy of user at the same time is a challenge for every researcher. On the other hand, social network website set the default privacy protection to a lower level, for the sake of increasing the user’s enthusiasm of participation and information sharing, but it leads sensitive information is mostly not well protected.To solve these problems at above, in this paper the social network privacy protection method is divided into interpersonal privacy protection and the privacy protection of personal information.The paper studies interpersonal privacy protection of social networks by spectral constraints and sensitive areas division,and develops the personal privacy protection system which sets the privacy parameter automatically.According to the above two methods, this paper focuses on the research of the following aspects.1. The international relationship privacy protection of social network focus on the degree of privacy protection and the social network data availability after perturbation. So this paper has two improvements on this problem.(1) This paper presents a new "transportation hub" sensitive areas division,It improves the sensitive area division algorithm so that the protection objects are more clearer.(2) In order to ensure the data availability as well as the sensitive relationship not to be leaked, this paper improves the original spectrum perturbation algorithm combined with the sensitive areas division algorithm proposed above.2. The access control of social network problem is setting of access permissions of "friends". This paper proposes an automatically setting system. This paper solves the access permissions through the classifier, mainly make the following two improvements.(1) It improves the overlapping community division method based on node splitting. So that the method can be applied on dynamic network,not only It can reduces the complexity of the algorithm, but also it increases the community divided accuracy.(2) It study on the support vector machine based on unbalanced data by combination of genetic algorithm to optimize the parameters. So that the accuracy can be much better.3. This paper do the experiments and design and verification of prototype system for model and algorithm presented above.The results show that those methods in this paper proposed can achieve the requirements. The method based on spectral constraints and sensitive area division can well balance the degree of privacy protection and the usability. The automatically setting system has good auto setting accuracy and user setting convenience.