Privacy-conscious information diffusion in social networks

Abstract

We present Riposte, a distributed algorithm for disseminating information (ideas, news, opinions, or trends) in a social network. Riposte ensures that information spreads widely if and only if a large fraction of users find it interesting, and this is done in a “privacyconscious” manner, namely without revealing the opinion of any individual user. Whenever an information item is received by a user, Riposte decides to either forward the item to all the user’s neighbors, or not to forward it to anyone. The decision is randomized and is based on the user’s (private) opinion on the item, as well as on an upper bound s on the number of user’s neighbors that have not received the item yet. In short, if the user likes the item, Riposte forwards it with probability slightly larger than 1/s, and if not, the item is forwarded with probability slightly smaller than 1/s. Using a comparison to branching processes, we show for a general family of random directed graphs with arbitrary outdegree sequences, that if the information item appeals to a sufficiently large (constant) fraction of users, then the item spreads to a constant fraction of the network; while if fewer users like it, the dissemination process dies out quickly. In addition, we provide extensive experimental evaluation of Riposte on topologies taken from online social networks, including Twitter and Facebook.