Research On User Privacy Preservation In Social Networks Based On Perturbation

Social networks have attracted great interest from researchers and developers in various fields of applications such as marketing, psychology,epidemiology, and homeland security. As social media become more and more popular, the popularity of online social networking sites and a variety of available social network analysis and mining technology development,privacy in the social network has become particularly concerned about the object, especially the published social network data. Attackers use published social network data and some background knowledge to gain victim privacy information. At the same time, a variety of user data forms and massive data size have become the obstacle of privacy preservation.With the research and exploration of privacy protection technology, the development and progress of related technology has attracted worldwide attention, which will become one of the new technologies in the field. of data mining.In this paper, we study the advantages and disadvantages of various privacy protection technologies in weighted social networks and the corresponding effective attack methods. Also give the definition of privacy protection problem in weighted social network. Based on the signal transfer model, a new establishment method of social networks is proposed.On the foundation of the social networks model, a privacy protection method based on rotation perturbation is brought out. The approach is better at the balance between privacy preservation and data availability.The innovation point of this paper is:(1) Study the existing social network model, establish the vector set model, and compare the different influence of node clustering methods on it. The global topology structure and signal transfer model are proposed to cluster the nodes and improve the clustering accuracy, which lays a good foundation for the establishment of the social network vector set model.(2) We design and improve the privacy preservation method based on rotation perturbation. We combine randomization and rotation perturbation,which not only makes the rotation matrix random, but also chooses the number of vector attributes randomly in the process of perturbation. And determines whether or not to end the disturbance by calculating the privacy protection degree and the initial threshold value. In this paper, the most effective privacy protection can protect the privacy of user data while ensuring data availability can still provide researchers with the relevant analysis.In this paper, the model improvement and privacy protection algorithm can do well in protecting user privacy, because of the normal distribution characteristics of Gauss random function, it can reduce the loss of data availability as much as possible.