HySoN : A Distributed Agent Gossip Protocol for Group Formation in Social Networks

On-line social networks allows people to easily interact wi th each other by means of social computer services. This scenario makes po ssible to search in a social network for affinities or new opportunities that sati fy specific requirements. However, for many users such activities often imply undesir able accesses to personal sensitive data. In this scenario we propose a novel app ro ch, called HySoN, based on an overlay network of software agents, which exploi ts a gossip protocol. HySoN allows users to locally maintain sensitive user’ s data, satisfying the privacy requirements preserving sensitive data. Indeed, t he properties involved in the HySoN user aggregation are inferred by local data not pub lished in the social network. Some experimental results obtained on simulated o n-line social networks data show the searching of suitable nodes is very efficient du e o the topology of the overlay network, which exhibits the small-world propertie s.