S-Peer: A Semantics-Based Peer to Peer System

A fundamental problem in P2P systems is content-based full-text search. Traditional approaches have either been centralized or use flooding to ensure accuracy of the results returned. In this paper, we present the design of a distributed P2P system (S-Peer) that supports semantics-based content searches of relevant documents. S-Peer nodes with semantically similar content are grouped into one semantic cluster, and self-organize a semantic network. Queries are routed to the appropriate clusters, increasing the chances that matching files will be found quickly. Experiments show that the S-Peer is more flexible to support semantic queries and can significantly improve query performance.