Research On User Behavior-based Searching Mechanism In Structured P2P Network

With the rapid development of Internet technology and the applications of network, the information on the Internet is expanding rapiddly, the traditional client/server mode (C/S mode) is no longer adapt to network development, so Peer-to-Peer (P2P) technology appears on the stage of history. P2P technology with the capability of improving the efficiency of the network, good scalability, strong fault tolerance, becomes the focus of scientific research and product development. P2P resource sharing is the most important applications, so in-depth study P2P network search mechanism is very important. Unstructured P2P network model of the search mechanisms generally use the flooding method, the shortcomings of such a mechanism is that it will generate a lot of redundancy in the network. The structured P2P network model is a pure distributed system to provide search by keyword, location-based services. The latest research results are mostly based on DHT (Distributed Hash Table) distributed routing search algorithm, it has good scalability, efficiency, adaptability, and so on. But its searching mechanism is a single words based exact searching method, and doesn't support the semantics. Many cases, people can not accurately describe the goal to search for given key words describing the same goal could not find satisfactory results. Therefore, this paper presents a user behavior-based rearching mechanism in Structured P2P network. User behavior analysis can find the relationship between key words and resources according to the users'searching and downloading behavior.Then the community discovery method is inducted to the results of user behavior analysis to mine the deep relationship between key words and resources. So, the searching mechanism can increase hit rate and accuracy, and make the resources sorted by download times.This research work includes the following:Firstly, several P2P network models and their searching mechanisms are studied and analyzed. The searching mechanisms'advantages and shortcomings have been compared.Secondly, the user behavior method is analyzed. The method is to find the relationship between keywords and resources, and their related degrees. Thirdly, the theory of community is studied, and the community discovery method is inducted to the results of the user behavior analysis inorder to mine the deep relation of key words and resources. Two methods of community discovery have been researched in this artical. One is based on bipartite graphs; the other is based on the resource relation network. The searching method is researched to make the resources sorted by download times.Finally, in order to verify the proposed search mechanism of the feasibility and effectiveness, a simulation experiment is taken in this article by matlab. Experimental results show that the user behavior based searching mechanism is simple and feasible.The system is built on the DHT network, not replaced of the DHT network, so it can be used widely. The mechanism also can adapt its own performance by the users'behavior. This mechanism has a higher hit rate and accuracy.