文摘
We propose a method for network anonymization that consists on sampling a subset of vertices and merging its neighborhoods in the network. In such a way, by publishing the merged graph of the network together with the sampled vertices and their locally anonymized neighborhoods, we obtain a complete anonymized picture of the network. We prove that the anonymization of the merged graph incurs in lower information loss, hence, it has more utility than the direct anonymization of the graph. It also yields an improvement on the quality of the anonymization of the local neighbors of a given subset of vertices.