Privacy Preserving Intersection of Neighbor Sets Exploiting Cross Checking Capability in a Peer to Peer Social Network Service

Due to the privacy concerns on the data generated by the users, Peer to Peer Social Network Services are getting popular these days because the data is kept in a distributed manner. In some cases, the list of neighbors of a node should be kept private, too. However, for some applications, we may need to compute the list of common neighbors between two nodes without revealing the whole list of neighbors. In this paper, we propose a Bloom filter based approach to compute the intersection of neighbors between two nodes in SNSes. We exploit the cross-checking property enabled by the neighbor relationships to simplify the computation while getting more accurate results. Our proposed method can get a near perfect intersection with mostly zero or one false common neighbors. Furthermore, the Bloom filter can successfully hide the neighbor information from attackers. We show the performance through numerical analysis and extensive simulations.