基于d-邻域子图匿名的社会网络隐私保护*

社会网络分析可能会侵害到个体的隐私信息，需要在发布的同时进行隐私保护。针对社会网络发布中存在的邻域攻击问题，提出了基于超边矩阵表示的d-邻域子图k-匿名模型。该模型采用矩阵表示顶点的d-邻域子图，通过矩阵的匹配来实现子图的k-匿名，使得匿名化网络中的每个节点都拥有不少于k个同构的d-邻域子图。实验结果表明该模型能够有效地抵制邻域攻击，保护隐私信息。

Preserving privacy in social networks based on d-neighborhood subgraph anonymity

Preserving privacy is very necessary for social network information publishing, because analysis of social networks can violate the individual privacy. This paper proposed a k-anonymity model of d-neighborhood subgraph described by matrix of supe-edge. It transformed the anonymization of subgraph into matching the matrix which represented the d-neighborhood subgraph of vertex, and ensured that the numbers of isomorphic d-neighborhood subgraph was no less than k for every vertex. Experimental results show that the proposed model can effectively resist neighborhood attacks and preserve privacy information.