Privacy-Preserving Network Aggregation

Abstract. Consider the scenario where information about a large network is distributed across several different parties or commercial entities. Intuitively, we would expect that the aggregate network formed by combining the individual private networks would be a more faithful representation of the network phenomenon as a whole. However, privacy preservation of the individual networks becomes a mandate. Thus, it would be useful, given several portions of an underlying network p1 . . . pn, to securely compute the aggregate of all the networks pi in a manner such that no party learns information about any other party’s network. In this work, we propose a novel privacy preservation protocol for the non-trivial case of weighted networks. The protocol is secure against malicious adversaries.

• Type: Conference
• Year: 2010
• Where: PAKDD
• Authors: Troy Raeder, Marina Blanton, Nitesh V. Chawla, Keith B. Frikken
• Keywords: Aggregate Network, Data Mining, Individual Private Networks, PAKDD 2010, Privacy Preservation