Efficient, failure-resilient semantic web service planning

Abstract

Over the past years service-oriented architectures have been widely adopted by stakeholders from research and industry. Since the number of services increases rapidly, effective methods are required to automatically discover and compose services according to user requirements. For this purpose, machine-understandable semantic annotations have to be applied in order to enable logical reasoning on the functional aspects of services. However, current approaches are not capable of composing workflows in reasonable time, except for planning tools that require domain-dependent heuristics or constrain the expressiveness of the description language. In addition to that, these tools neglect alternative plans, concealing the danger of creating a workflow having insufficient reliability. Therefore, we propose an approach to efficiently pre-cluster similar services according to their parameters. This way the search space is limited and vulnerable intermediate steps in the workflow can be effectively avoided. © 2010 Springer-Verlag.