中文摘要：

如何利用具有本体标注的结构化文档中的语义信息组织P2P网络，提供对基于语义的信息共享与查询的P2P网络支持，是当前P2P网络的研究热点之一。本文提出采用Peer所存储文档中的加权本体概念向量作为Peer的特征向量，通过相似度计算将Peer聚成Peer组，从而构造基于语义的半结构化P2P网络。用户的查询请求由各Peer组内的组服务器负责路由转发，组服务器计算查询请求与各路由表项之间的相似度，将查询向最有可能包含查询目标的Peer组转发。文中较全面地阐述了本体概念局部和全局权重的计算方法。由于P2P网络拓扑的建立过程和查询路由过程均基于语义信息，使得网络的各项性能与基于关键字处理的P2P网络相比，得到了较大的提高。

英文摘要：

In this paper, we proposed a semantic based semi-structured P2P network architecture for ontology annotated documents＇ sharing and retrieval. The vector of ontology concepts is used as the feature of a peer. Based on the similarity function, the peers are clustered into peer groups. And the query is routed by peer group servers according to the similarities of the query and the group vectors. The weighting ftmctions of the ontology concepts and similarity functions of concept vectors are described in detail in this paper as also. Benefiting from using the semantic information to construct the P2P network, the performance of peer clustering and query routing is quit enhanced comparing to the pure text based P2P network, which is proved by the experiment results.