A general search method based on social communities in P2P networks

With the increase of the amount of information stored in P2P networks, how to search the information satisfying users´ needs efficiently becomes very important. Current researches on search algorithms focus on increasing search efficiency(measured by the length of routing path) or decreasing search cost(measured by the number of messages) at the cost of sacrificing the recall rate. However, there is no work which increases search efficiency, decreases search cost and increase the recall rate. In this paper, we propose a general search method which can be applied in both unstructured P2P networks and structured P2P networks. In this method, social communities are formed dynamically on top of P2P overlay networks. Each social community is made up of peers who share similar characteristics, such as interests, search behavior, etc. These characteristics are dynamic, so the social communities will change as any peer´s characteristics have changed. In this method, a search request is forwarded by taking use of the social communities. Simulation results show our proposed method achieves a higher search efficiency, a lower search cost and a higher recall rate compared with traditional search algorithms.