Increasing the Anonymity of a Social Network Based on Splitting Users into Constant Usefulness and Zero Information Loss

Social networks are a part of today's society. In such networks, privacy-preserving is considered as an important field since the user's identity is usually revealed by inference attacks. Within this context, security system designers use an isomorphic graph to stop attackers by editing nodes and links, so, the attackers cannot access to users' identities when the vertexes and links of the graph are converted into isomorphic parameters. The security system designers employ random links, clustering of the nodes, weight balancing, nodes addition or de-anonymization techniques (nodes labeling) to confuse malicious attackers. But, these techniques have many defects, such as the loss of information and the reduction of usefulness parameters that evaluate the final social network graph. This paper proposes a new method named as a splitting method where three techniques are proposed to improve mentioned parameters and in general, to improve further the network management.