Recommendation Retrieval in Reputation Assessment for Peer-to-Peer Systems

Recommendation-based reputation assessment in peer-to-peer systems relies on recommendations in predicting the reputation of peers. In this paper, we discuss the effectiveness and cost metrics in the recommendation retrieval. We evaluate the following retrieval methods: flooding, recommendation tree, and the storage peer. The simulation results show that overlay network construction significantly contributes to the performance of recommendation retrieval in terms of effectiveness and cost. Storage peer approach in structured network outperforms the other two approaches as long as the network is stable.