Privacy Preserving Estimation of Social Influence

Exploiting word-of-mouth effect to create viral cascades in social networks is a very appealing possibility from the marketing standpoint. However, in order to set up an effective viral marketing campaign, one has first to accurately estimate social influence. This is usually done by analyzing user activity data. As we point out in this paper, the data analysis and sharing that is needed to estimate social influence raises important privacy issues that may jeopardize the legal, ethical and societal acceptability of such practice, and in turn, the concrete applicability of viral marketing in the real world. In this paper we devise secure multiparty protocols that allow a group of service providers and a social networking platform to jointly compute social influence, in a privacy preserving manner.