Text Retrieval In P2P Networks Based On Small World Model

The Peer-to-Peer network is a network topology that applies equality policy. Compared with traditional client/server comparison, there are low cost, high scalability, good fault tolerance,autonomy and self-organizability advantages. With extensive use of the network, the sharing of video, audio and text, the number of image shows a surprisingly increasing speed. There for, the demands for retrieving the contents are more and more and the requirements are higher and higher.The object of this paper is the unstructured P2P networks. Unstructured P2P network is a kind of typical representative for Gnutella network. In such networks, the nodes are not specified logical address, using random methods or inspire strategy to join the network. The network topology changes with the node change and the network communication.This paper analyse the power law and small world model of the complex network, analyzing and comparing the P2P network topology and search download model. For the flooding problem in Gnutella, a text retrieval model is proposed in unstructured P2P network based on small world models. Each node in the P2P networks maintains several neighbours as short-link and several long-link to improve text retrieval efficiency and reduce communication overhead between nodes. Neighbor update strategy is based on feedback information after the search. Calculate the importance degree of neighbors and sort, then select the high importace node as an neighbor. The low importance node which is replaced have a chance to become long-link. Thus, the node can quickly learn about other nodes in the document content and make a similar interest in tight junction nodes. The experiment results show that compared to the Gnutella network search algorithms, the proposed based on small world theory unstructured P2P networks text retrieval algorithm has better performance and the results will be more clear as the search continues.