基于余弦相似度分组的P2P搜索机制

P2P Search Mechanism Based on Cosine Similarity Grouping

针对现有P2P分组搜索机制存在的不足,提出一种改进的搜索机制,以节点间标引词权重向量的余弦相似度为基础,综合考虑节点间距离因素来建立合理的逻辑分组,采用一种任务分担策略,由不同组员充当外联节点实现组间搜索。实验结果表明,该机制网络扩展性强,具有较高的搜索效率和查全率。

Aiming at the deficit of the existing grouping search mechanism, an improved search mechanism is proposed, which establishes a logical group reasonably based on the cosine similarity of term weight vector, integrating with considering the distance between the nodes. A task-sharing strategy is presented in the inter-group search, letting different members act as the managers of the links. Experimental result shows that the mechanism has strong expansibility, efficient search capability and higher recall rate.