Abstract: One of essential problems on P2P network is how to find a node which saves expected data.So exploring lookup algorithms is a key component for P2P technology，which decides the performance of P2P system.Each node keeps a list that records the routing information of its neighbor.This paper focuses on building a Bayesian network corresponding with their routing table of nodes.Depending on this Bayesian network，we analyze the node probability of receiving a lookup request and obtain a formula to compute this probability.For making this formula more closed to reality，we assign a weight to each of nodes and adjust these weights.At last，a scheme of assigning reliability to each node is presented and an improved routing algorithm is given based on Bayesian method.Some simulations show that this improved routing algorithm can enhance the performance of P2P network to some extent.

Key words: Pere to Pere（P2P）, Bayesian network, lookup, routing

摘要： 对等网络中的一个关键问题就是如何找到储存有期望数据的节点，因而目标资源的查询算法研究是P2P网络的关键部分，该算法决定了P2P系统的性能。在对等网络中的每一个节点都存有一张记录与之相邻的节点的路由信息表，着重讨论如何根据各节点所存储的路由表建立相应的贝叶斯网络，并分析某一节点接收到查询请求的概率，进而得出一个计算概率的数学公式。为使所得概率计算公式尽可能地符合现实情况，每一个对等点分别被赋予不同的权，随后对所赋权进行调整。最后，对每个节点的可信度提出一个设置方案，并基于可信度提出一个改进的路由算法，试验表明该算法能够一定程度上改善对等网络的性能。

关键词: P2P, 贝叶斯网络, 查找, 路由