Abstract: With rapid development of P2P network，the number of users is increasing continuously，and message exchanging is more frequent.Aiming to reduce the complexity and period of searching，this paper provids a searching approach based on message similarity.The approach divides the nodes of random network into different domains according to message similarity.In searching messages，users match the keys with message vector of domain nodes.This approach converts searching of whole network into domain or adjacent domain searching，and returns the results of searching according to user interest value.The experiment indicates that the approach based on message similarity is valid.

Key words: P2P searching, message similarity, MSS algorithm, User Interest Value（UIV）

摘要： 随着P2P网络的迅速发展，其用户数量不断增加，信息交互也越发频繁。为了降低内容搜索的复杂度和搜索时间，提出了一种以信息相似值为依据的计算方法。这种计算方法将无规则P2P网络中的节点按照节点的信息相似值划分为不同的域。在信息搜索时，将搜索的关键字与域头节点信息向量进行匹配，算法将整网搜索转化成域内或相邻域搜索，并根据用户兴趣值返回搜索结果。实验证明，这种信息相似值的计算方法在降低搜索时间的基础上，有较高的搜索命中率和查询准确率。

关键词: P2P搜索, 信息相似值, MSS算法, 用户兴趣值