社会网络子集个性化隐私保护策略

Personalized privacy protection strategy of subset of social networks

现有的社会网络隐私保护通常是基于所有用户完全一致的隐私保护,忽略了用户之间对隐私保护的需求存在差别。针对这一问题,提出个性化隐私保护框架,根据用户不同隐私保护需求提取部分子集,共设置三种隐私保护级别:首先,简单移除原始图节点标签,并为每个节点设置相应的ID值;其次,为保护节点度信息,提出基于动态规划思想的k-d_sub(k-degree_subset)算法;最后,为防止敏感属性被识别将l-diversity与k-d_sub算法结合,提出k-d_l_sub(k-degree_l_subset)算法,添加最少数量的边,降低匿名成本,并且最大化数据效用。实验证明,提出的个性化隐私保护框架有较高的匿名质量,能有效保护社会网络中用户的隐私。

The existing social network privacy protection algorithms base on all users have the same privacy protection needs, ignoring that different users have different preferences. In order to solve this problem, this paper put forward a personal privacy protection framework, according to different user privacy protection needed to extract some subset, set up three kinds of priva- cy protection level ： first of all, it removed the node labels of original graph simply, and set the corresponding ID value for each node; second, in order to protect information of the node degree, it put forword k-d_sub （k-degree_subset） algorithm based on dynamic programming ideas; finally, in order to prevent the sensitive attributes which were identified, it proposed k-d_l_sub （k-degree_l_subset） algorithm, which could add minimum number of edges, and decrease the cost and maximize the data availability. Experiments show that the proposed new personalized framework has higher anonymous quality, and can effectively protect the privacy of users in the social network.