摘要
在对P2P用户行为进行分析的基础上,提出了一种自动机制,能够区分出用户对不同主题领域的关注度,计算出邻居节点查询各关注主题领域相关文档的能力,通过选择对特定领域查询能力最强的k个邻居节点转发查询消息提高效率,该机制能够区分出用户的典型行为和即兴行为,通过采用不同策略进一步提高即兴查询的效率。
Efficient select-neighbor-peer strategy is necessary to improve efficiency of full-text retrieval in P2P networks. A dynamic cluster algorithm is proposed by analyzing query history. Our approach can explicitly distinguish different topic of interest, especially persistent interests and transient queries, and allows different search strategies to be applied to different kinds of queries.
出处
《情报科学》
CSSCI
北大核心
2008年第8期1244-1247,1268,共5页
Information Science
关键词
全文检索
用户行为
兴趣主题
动态聚类
full - text retrieval
user behavior
interest topic
dynamic clustering