Dynamic Service Evolution for Open Languages in the Grid and Service Oriented Architecture

Abstract

Dynamic behavior is inherent in virtual organizations. Semantics has to be processed to manage dynamism and other properties like state, life cycle, faults, and others. In our paper we identify a gap between service interfaces and semantic for dynamic services to establish virtual organizations. We propose a novel approach called "dynamic service evolution" (DSE) to overcome this gap. A common language is needed to allow dynamic semantic information processing. The evolution of a language enables bilateral agreements between services and clients in an open process. We have applied our approach in N2Grid and present a prototype in this paper. N2Grid deploys artificial neural networks as virtual resources in a Grid. By the DSE we solve the problem, which no general standards (languages) for neural network representations exist.