A middleware service for pervasive social networking

Abstract

Today's online social networking applications (e.g., Face-book, Twitter, Last.Fm) allow users that are socially close to each other (i.e., users with shared interests) to participate in the collective production and sharing of information (i.e., virtual interactions). Pervasive Social Networking (PSN) is a new vision that aims to complement virtual interactions with physical ones, by enabling users who are both socially and physically related to find each other and perform activities of common interest. To enable this vision, both users' social networks and mobility patterns must be reasoned upon. In this paper, we present a social networking middleware service that dynamically combines both social and physical proximity relations between mobile users to accurately recommend them people with whom to perform activities of common interest. At the heart of this service is a social network propagation component that infers users' relations both within the same (intra) and across (inter) users' activities. We evaluate the impact of various middleware deployment strategies on the ability of the social network propagation component to find related users, and analyse the advantages and shortcomings of each of them.