动态社会网络隐私保护方法研究

Research on privacy preserving method for dynamic social network

针对动态社会网络数据多重发布中用户的隐私信息泄露问题,结合攻击者基于背景知识的结构化攻击,提出了一种动态社会网络隐私保护方法。该方法首先在每次发布时采用k-同构算法把原始图有效划分为k个同构子图,并最小化匿名成本;然后对节点ID泛化,阻止节点增加或删除时攻击者结合多重发布间的关联识别用户的隐私信息。通过数据集实验证实,提出的方法有较高的匿名质量和较低的信息损失,能有效保护动态社会网络中用户的隐私。

For user＇s privacy-information disclosure issues in multiple-release of dynamic social network data,combined with attacker based on structure attack of background knowledge,this paper proposed a privacy preserving method of dynamic social network,which divided original graph into k isomorpgic graphs through k-isomorphism algorithm in each release,also minimized generalization cost and generalized the node ID to stop the attacker identifying the user＇s private information with the association between multiple-release when nodes were added or removed.Confirmed by data sets experiment,proposes method has higher quality and lower anonymous information loss and can effectively protect user＇s privacy in the dynamic social network.