Repairing Service Compositions in a Changing World

One of the most important benefits of ServiceOriented Computing is to foster the satisfaction of end-user needs through the automatic generation of composite services out of simpler services existing in the user environment. Different approaches have been proposed in the last years to address this issue, e.g., based on model-checking or AI planning. Still, these approaches do not cope with the inherent dynamicity of the service pervasive world, where not only available services, but also user needs, may evolve over time. Setting up service composition in an AI planning framework, we propose in this paper repair techniques enabling service compositions to adapt at run-time, both to service and requirement changes, paving the way for on-demand and sustainable end-user service composition.