Approximating service utility from policies and value function patterns

Service-oriented computing provides the right means for building flexible systems that allow dynamic configuration and on-the-fly composition. In order to realize this vision, the system must be able to choose the most suitable service from a large and constantly changing number of providers. We present an approach for selecting services based on a coherent conceptual policy model and a service utility measure. Our framework is capable of capturing technical and application specific aspects of services and incorporates them into the decision making process. We assist the user in establishing the utility measure from existing policies by attaching value function patterns to the individual attributes. Drawing from the areas of utility theory, foundational ontology, and electronic markets, our work is a promising approach for unifying the heterogeneous methodologies in the service selection process.