基于k-邻域同构的动态社会网络隐私保护方法

社会网络数据分析蕴藏着巨大的经济利益,但是直接研究社会网络数据可能造成用户敏感信息泄漏,对个人隐私构成威胁.目前的隐私保护技术集中于研究单次数据发布,即静态网络中的隐私保护,然而社会网络数据动态发布需要动态的隐私保护方法.文中针对攻击者拥有在不同时刻的节点1-邻域子图作为背景知识的应用场景,提出了一种基于动态社会网络的隐私保护方法,该方法利用相邻时间片网络图之间的关联关系,依据信息变化增量确定邻域同构等价组中的基准节点,并通过对下三角矩阵操作来实现等价组中节点邻域子图匿名化的持久性.实验结果表明该模型能够有效地抵制邻域攻击,保护动态社会网络发布的用户数据隐私.

Privacy Preservation in Dynamic Social Networks Based on k-neighborhood Isomorphism

The analysis of social network data has enormous value. Studying the social network data directly results in personal information disclosure, thus threating the personal privacy. The current privacy protection technology focuses on single data publishing, it only can be used in the static network. However, the publishing of dynamic social network data requires a dynamic approach to deal with. Based on as- sumption the attacker has the background of 1-neighborhood sub-graphs of vertices at different time, a pri- vacy preserving model for dynamic networks is proposed using the association between the adjacent time slices of the network graph to determine the standard vertex for equivalence group in which the vertices are isomorphic according to the incremental changes of information, thus satisfying the lasting anonymity of the neighborhood sub-graph in each equivalence group. Experimental results show that the model can thwart 1-neighborhood attacks and protect the privacy of publishing in the dynamic social networks.