Vulnerability in socially-informed peer-to-peer systems
Jeremy Blackburn, Nicolas Kourtellis, Adriana Iamnitchi · 2011
The recent increase in the volume of recorded social interactions has the potential to enable a large class of innovative social applications and services. The decentralized management of such social information as a social graph distributed on a user-contributed peer-to-peer network is appealing due to privacy concerns. This paper studies the vulnerability of such a peer-to-peer system to attacks staged by malicious users who try to manipulate the graph or by malicious peers who try to manipulate the mining of the social graph. We discuss the effects and limitations of such attacks and we show experimentally how the distribution of the social data onto peers affects the system's resilience.