K-anonymity against neighborhood attacks in weighted social networks

Nowadays, with the advance of Internet technology, social network is getting popular, which combines the virtual network and the real world. People employ this network to communicate with all social things of their interest, including shopping, making friends, sharing experiences of life, and so on. In social networks, the social data expands rapidly and the malicious users can easily get the social data. With the social information, they could conjecture the relationship among social network data via the systematical analysis tool. Hence, the personal privacy data in social network may be exposed to some unknown risks, and recently, these issues arising in such a network catch much attention. The protection of personal privacy social data becomes an important and urgent research in social networks. One of personal privacy social information is the relations between the individuals and their social groups, namely human relationships. Neighborhood attacks are incident to the exposure of human relationships. Previous studies try to conceal human relationships information with well-known k-anonymity protection to resist this attack in social networks. However, those researches do not take care of those attacks in weighted social network, which does not make sense due to the fact that people should have different relationships with different persons. In such weighted social network, the edge denotes the human relationship and the weight denotes the degree of this human relationship in social networks. Our study focuses on the k-anonymity protection scheme in weighted social networks, and our scheme can achieve k-anonymity protection under the expected conditions, less virtual edges added and fewer weights changed. Through the analysis with MATLAB tool, we show that our k-anonymity algorithm can achieve high anonymity protection rate under various k-anonymity policies.