Goal-driven service mediation solution

This thesis proposes a solution to the interoperability problem of enterprise systems in the context of service mediation. Service mediation is ideal when two or more systems want to interoperate but their functionalities, exposed as services, are fixed and are difficult, if not impossible to change, for the purpose of integration. A Mediator is a software that sits between two or more systems to resolve process and data mismatches. Process mismatches occur when collaborating systems use services that follow different ordering of message exchanges. Data mismatches occur when collaborating systems use different information models (or vocabularies) to describe the messages that are exchanged by their services. However, we find that designing integration solutions has always been traditionally technologydriven where Information Technology (IT) specialists do most of the job of building the technical solutions. We argue that business domain experts should be given the opportunity to participate more actively in the design of the mediation solution. We believe that their involvement is crucial in ensuring that the mediation solution delivers its intended business purpose. This thesis investigates the use of goal-driven approaches to enable business domain experts in specifying and elaborating the requirements of the mediation solution. Goals capture the reasons why the integration is needed. Goals also provide business domain experts a sufficient level of abstraction in specifying and validating integration design choices at the business level and for communicating such choices among different stakeholders. As such, we argue that business domain experts are at the best position to describe the requirements of the service mediation solution through goals. Finally, we investigate the use of model-driven techniques to transform these abstract goals into technology-specific implementations, and investigate how business rules can be used in the process. Model-driven development allows us to raise the problem and solution analyses spaces to a level of abstraction that is technology independent more suited for business-level analysis. We believe that model-driven techniques should give added flexibility to our solution as business requirements, their design specifications, and technical implementations are treated as separate concerns with the resulting artifacts maintainable and reusable.