Measurement study and application of social network in the Maze P2P file-sharing system

Search in P2P networks is an open problem, and most of P2P networks reflect social relationships. By studying and leveraging social networks of P2P systems, search effect can be significantly improved. In this paper, we report results of analyzing and measuring of social networks in Maze, a large-scale peer-to-peer file sharing system. Given the definition and measurement of social relationship, friend network and download network are analyzed based on Maze's daily log. On both networks, the construction, structure, and 'small-world' property are illustrated. And it is proven that both networks reflect social relationships. Leveraging the social networks and EigenTrust algorithm, a search approach EigenForwarding(k,t) is presented. Comparing with flooding and random forwarding, the approach is shown that can gain a good hit rate with small messaging cost and search path length due to sufficient exploiting of sociality. ? 2006 ACM.EI