Quantifying Trust Relationships based on Real-World Social Interactions

Deriving trust relationships from real-world social interactions may contribute significant information towards social behaviour understanding. The level of trust among people constitutes an insightful parameter for describing the social context but also an important measure for security and privacy in pervasive systems. Current works for deriving trust relationships either consider only on-line social networks to create trust networks or focus on users' on-line social interactions. This article presents MobTrust, an opportunistic sensing system that derives and quantifies trust relationships among people through smartphones based on the detected real-world social interactions. A real-world social graph is derived from users' daily social interactions by also considering snapshots of their social relation. A hybrid model was developed to quantify users' trust relationships based on the extracted real-world social graph, the estimated social relations and the contextual information provided by the detected social interactions. As a proof of concept, a real-world evaluation of the system is performed. The evaluation shows that MobTrust can reliably derive and quantify trust relationships, having as such the potential of empowering a variety of real-world scenarios that can leverage such knowledge.