Search Using Semantic Inference in Unstructured P2P Networks

Abstract

Peer-to-Peer networks develop rapidly in the last few years. The search algorithm lies at the centre of these networks. Many search methods have been proposed for unstructured peer-to-peer networks, but complicated organization, high search cost and maintenance overhead make them less practicable. To avoid these weaknesses, in this paper, we propose an adaptive and efficient method for search in unstructured P2P networks, the Semantic Inference Search method (SIS). This approach is based on a simple and powerful principle similar to interest-based locality. It utilizes feedback of not only the requested objects but also semantically related objects from previous searches. It applies Bayesian network to establish an inference model, using semantic inference to direct future searches. Experimental results show that the SIS method achieves high success rate, more discovered objects, low bandwidth consumption, less maintenance and adaptation to changing network topologies.