Abstract: To overcome some disadvantages of the centralized search engines such as not facilitating human user collaboration, ignoring completely the interests and preferences of users, not being scalable and timely, this paper proposed a P2P based Web search system called PeerBridge for personalized Web information searching. PeerBride adopted the distributed hash table (DHT) to organize a large number of network nodes to form a structured P2P overlay networks. Each peer acted as a topic driven search engine, which can seek useful information according user’s interests. Furthermore, those peers that have similar topics were associated together to form peer clusters. Peers cooperated to search the Web and share information. Some technique such as topic driven Web crawling, semantic concept based document classification, personalized linkage analysis, and topic segment based P2P searching were used to improve the PeerBridge’s performance.

Key words: Web search, Peer-to-peer network, Web Mining

摘要： 针对中心化的Web信息搜索系统在覆盖率、及时性、个性化、可扩展性等方面存在的问题,本文提出一种基于Peer-to-Peer(P2P)的可扩展、个性化的Web搜索系统PeerBridge. PeerBridge基于分布式哈希表组织大量的网络结点形成有组织的P2P覆盖网络, 每个对等体作为一个主题搜索引擎,根据用户兴趣从Web中搜索特定主题相关的信息,而具有相似主题的对等体被聚集在一起形成基于主题的对等体簇,协作进行Web搜索与信息共享.并采用主题驱动的Web爬行、基于语义概念的文档分类、个性化的链接分析和基于主题划分的P2P搜索等机制来改善PeerBridge的性能.

关键词: Web搜索, 对等网络, Web挖掘