RS2D: Fast Adaptive Search for Semantic Web Services in Unstructured P2P Networks

In this paper, we present an approach, called RS2D v1, to adaptive probabilistic search for semantic web services in unstructured P2P networks. Each service agent dynamically learns the averaged queryanswer behavior of its neighbor peers, and forwards service requests to those with minimal mixed Bayesian risk of doing so in terms of estimated semantic gain and commmunication cost. Experimental evaluation shows that the RS2D search mechanism is robust against changes in the network, and fast with reasonably high precision compared to other existing relevant approaches.