基于边差分隐私的社交网络图结构数据发布方法，首发论文 -凯发网

social network graph structure data publishing method based on edge differential privacy

• 1、beijing key laboratory of security and privacy in intelligent transportation, beijingjiaotong university, beijing 100044
• 2、school of computer science and technology, beijing jiaotong university, beijing 100044

abstract：in social networks, attackers can infer individual privacy through its structural characteristics, so as to obtain and disclose potentially sensitive information such as user relationship networks and transaction records. however, current research lacks data availability, ignores the global importance of nodes, and fails to flexibly add noise to different stages. based on this, we propose a social network graph data publishing method based on edge differential privacy. we iteratively update the noisy katz centrality, group the nodes to implement edge differential privacy perturbation, and use a random response mechanism to adjust the node neighbor relationship. experiments show that we have less information loss and higher data utility in different scale data sets. finally, we achieve the goal of maximally releasing the real social network graph under the constraint of edge differential privacy.

keywords： computer system architecture social network edge differencial privacy katz centrality