AI for Peer-to-Peer Systems

Chaokun Wang,Jianzhong Li
2002-01-01
Abstract:Peer-to-Peer systems have become a focus of distributed computing these years. Document discovery is a key problem in this field. In this paper, abstract indices are proposed for content-based information retrieval in Peer-to-Peer systems, especially for multimedia documents. Abstract information is extracted from the shared documents of each peer, then the similar documents on other peers can be retrieved based on content via abstract indices in both centralized P2P systems and distributed P2P systems. Abstract indices (AI) make it possible to search fewer peers but get more similar documents. Experiments show AI works well in simulated P2P systems.
What problem does this paper attempt to address?