Research On Technologies Of Search Engine Based On Peer-to-peer Networks

With the development of internet rapidly, more and more information are published on the World Wide Web. The information are highly distributed residing on millions of sites. This trend challenges the traditional information retrieval techniques like SE (search engine), for SE are of the centralized architectures. However, Centralized server can not met the demands of users on the quantity and speed of information processing because the restriction of the capacity of storage bandwidth and the capacity of computation. Meanwhile, the development of P2P (Peer-to-Peer) offers a new research area-establishing a SE based on P2P.The main advantages of P2P search engine as follows:First, information is high-distributed in network and it is in accord with the Architecture of P2P. Second, peers in P2P network are able to provide self resource including the capacity of storage bandwidth and the capacity of computation. Accordingly, P2P network are capable of more hardware resource. However, as a new the application of network, SE based on P2P network has many problems. For instance, low efficiency algorithm is difficult to support complex queries.This paper analysis the technology problems of SE based on P2P and propose History based Multi-keywords Search (HMS) in unstructured peer-to-peer systems, which only requires each peer to maintain partial query information of its neighbors, and exploit these information to autonomously find how many neighbors and which neighbors are likely to answer. The peers will only forward query messages to neighbors which are more likely to reply.At last, we design and implement an information retrieval simulation system based on unstructured P2P and using it to test the efficiency of HMS. The result shows HMS is able to reduce network communication cost and ensure high quality searching results.