Research On Multi-feature Anonymous Algorithm Based On Topology And Attributes

With the rapid development of Internet technology,a large amount of user data is generated in various social networks.The way of publicly publishing social network data often accompanies the risk of the leakage of personal privacy information of the user and infringes on the privacy of the user.Thus,data needs to be protected before it is released.In the existing data privacy protection methods,anonymization protection is a relatively extensive protection method,which can effectively ensure the secure release of data without leaking private information.However,most of these methods are aimed at the anonymous protection of the social network graph structure,and don't take into account the protection of node attributes.With the diversity of data and the complex background knowledge possessed by attackers,the existing methods of anonymous protection cannot resist some new attacks,such as community attacks,neighbor attribute attacks,etc.Based on the privacy protection requirements in social network data release,this paper analyzes some privacy protection anonymous models,technical implementation of anonymous methods,and anonymous information metrics,based on the analysis of domestic and foreign data privacy protection technology research and analysis.And research.Based on these researches,this paper conducts in-depth research on the issues of data anonymization technology combining graph structure and attributes in social network data publishing.The research work and main contributions of the paper include:1.Introduce the topic selection background and research significance of this article,and then review the research status at home and abroad.It discusses the attack on the structure and attributes of social networks,introduces and summarizes the related concepts and protection methods of social network protection,and specifies the privacy measurement methods in anonymous protection of social networks.2.For the existing k-degree anonymous algorithm,only the degree of the node is protected,and the community structure of the graph is not protected,resulting in the inability to resist the problem of privacy leakage caused by community attribute information attacks.This article introduces the concept of community division.A k-degree anonymous protection method based on node classification is proposed.This method is based on the traditional kdegree anonymous method,which not only protects nodes or edges,but more importantly protects the community attributes of nodes.Therefore,the concept of node classification is proposed,and nodes are classified according to their different importance in the graph.Experimental results show that,compared with the existing k-degree anonymous algorithm,the algorithm of node classification has a smaller information loss,and maintains good data practicability in terms of intermediary centrality and shortest path distance.When the community is divided,the change rate of nodes in the community is higher than 10%,which effectively disrupts the community structure and can resist community attribute attacks.3.To further analyze and study the KDLD protection method of attribute graph,the edge modification is too large,the added noise data is too high,etc.,and a(k +,l)anonymous protection method based on structure and attribute is proposed,combining The classification of nodes in two aspects of the structure.This method adds less noise data and has a smaller degree of information loss than the KDLD attribute protection method.