|
Peer is a peer to peer model using a network , with the traditional client / server comparison, there are low-cost, high scalability , fault tolerance , autonomy and self-organization , etc. , with the peer to peer network wide range of applications , network sharing text, video , audio, images showing a surprising amount of growth . Therefore, for the retrieval of these contents are increasingly in demand increasingly high requirements . The object of this study is unstructured P2P networks. Unstructured P2P network are defined by Gnutella is a typical representative of a class of networks. In such a network , the node is not specified logical address , using a random method, or inspired policy to a network , network topology changes and as nodes for network communications and evolve . In this paper, the power law and the complex network of small-world characteristics were analyzed , and the now -peer network topology and search download are analyzed and compared , for the Gnutella network message forwarding flooding problem , we propose a small- world model based unstructured P2P network Chinese the retrieval method . In this method , each node in the network maintains a number of neighbor nodes connected to a short-range , and each node also maintains a number of long-range connections to improve efficiency and reduce text retrieval communication overhead between nodes . Neighbors update strategy is based on feedback information inquiry neighbor adjustment: which is a measure of the importance neighbor nodes and sort , and select as high importance neighbor nodes , and was replaced by a low degree of importance of a node a chance to become a long-range connection node . Thus, the node can quickly understand the other nodes in the document content and makes tight connection nodes with similar interests . Experimental results show that , in contrast to the Gnutella network search algorithm , the proposed small- world theory based unstructured P2P network text retrieval algorithm has better performance, but the search continues with this effect will be more obvious.
|