基于主题划分的有组织P2P搜索算法

Distributed Information Search Based on Topic Partition in Structured Peer-to-Peer Networks

提出一种基于主题划分的P2P搜索算法———主题覆盖网络搜索算法(TONS).TONS在有组织P2P网络基础上,将结点按主题组织成覆盖网络,使含有相似主题的结点链接在一起,因此可以根据内容将查询限定在P2P网络的局部范围内.通过在覆盖网络中随机添加一些长距离链接,使覆盖网络具有Small-World特性.TONS为有组织的P2P系统提供了一种能依据部分匹配、多关键词等复杂条件搜索结点数据对象的有效方法,与现有的有组织P2P系统相比,它可以将搜索的查全率提高74.7%,并减少P2P网络信息搜索的平均路径距离和平均消息数目.

A topic overlay network search （TONS） algorithm, P2P search mechanism based on topic partition, is presented. On the basis of the structured networks, the nodes are organized as an overlay network according to topics such that the nodes containing similar topic are linked together. Thus, the query contents can be limited in the local range of P2P network and the overlay network has small world traits by randomly adding some long distance links in the overlay network. TONS provides structured P2P networks with effective approach to search for node data objects based on complicated queries with partial match and multiple keywords. Compared with the existing structured systems, TONS increases the search recall by 74. 7%, and reduces the average oath distance and the average number of messages during the searching process.