A Cognitive-Based Ego Network Detection System for Mobile Social Networking

In future generation mobile systems information about social networking structures of users will be fundamental, being a key element for social networking applications, and a crucial contextual information for personalising the behaviour of mobile applications/services. In this paper, we focus specifically on the detection of ego networks. They are networks formed by an individual (ego) and all the other people she has a social relationship with. We propose a completely decentralised algorithm that allows each user's mobile device to identify the structure of its user's ego network. The algorithm monitors social interaction patterns between the ego and its peers. It is completely decentralised and runs at each individual node using local information only, scaling with the network size. It does not disclose social interaction patterns, and it is able to dynamically detect changes in the structure of the ego network, being self-adaptive. The algorithm is based on social cognitive heuristics, i.e. Models about how the human brain groups social relationships, described in the cognitive psychology literature. Therefore, our approach reproduces - in users' personal mobile devices - the cognitive processes used by their human users to understand their ego networks' structure. We test it on real datasets of interactions corresponding to (i) physical contacts and (ii) exchange of information in online social networks. We show that in both cases the detected social structures are remarkably consistent with those described in the social sciences literature. In addition, we study the dynamic behaviour of the algorithm, highlighting how such structures evolve dynamically over time.