PIST: A P2P IR Model for Social-Tagging Network

This paper proposes a P2P information retrieval model for social-tagging network (PIST), which is layered on top of a DHT-based overlay network. The tags of document, user´s friends, interests and feedback are exploited to improve the search performance. With the help of friends, the user can get his desired documents with one-hop access. A document is associated with a document profile consisting of tags which describes the document. A user is associated with user profiles which represent his interests. During query processing, the relevant results can be returned according to user´s interest, and user´s feedback is exploited to refine the profiles and friends list. Experiments show that PIST has good performance and it is suitable for social-tagging network.