Efficient Routing Method in P2P Systems Based upon Training Knowledge

Peer-to-peer systems have recently achieved a remarkable success in the social, academic and commercial communities. In P2P systems, a very large number of autonomous computing peers pool together their resources and rely on each other for any request. A fundamental problem in Peer-to-Peer networks is how to locate efficiently appropriate peers to answer a specific query (Query routing). A lot of research have been carried out to enhance search result quality as well as to reduce network overhead. Recent research focuses on methods based on query-oriented routing indices, which utilize the historical information of past queries and query hits to route future queries. The major problem of these methods is that upon joining the P2P network, a perhaps no prior knowledge. Therefore, it is impossible for the peer to perform good routing decisions. For this reason, it uses flooding method in order to build an initial knowledgebase. Consequently, during the training phase, any method can achieve slow improvement in routing efficiency. In this paper, we introduce a novel approach that aims to predict user profiles based on the shared documents and builds an initial knowledge base beforehand. In the absence of explicit queries, an alternative is to try to infer users interests implicitly from his shared documents. The interests, formulated as an implicit query, can then be used in further searches to construct an initial knowledge base. Our approach improves the efficiency of routing methods based on query historic during the training phase. We implemented the proposed approach, and tested its retrieval effectiveness in terms of recall and precision, also its efficacity in terms of messages traffic and visited peers number.