Business rule driven composite service optimisation and selection
Quality of Service (QoS) is often essential when the service consumer looks for a single service from a large pool. However, QoS of a composite service is aggregated in a rough way because the concrete service providers in the composition are seen as independent ones. In reality, business relationships such as dependencies and conflicts often exist among service providers, which unavoidably affect some QoS dimensions when the corresponding providers are selected to realise a composite service. Therefore, effects of business relationships must be analysed in the service selection process. This research proposes a composite service selection approach with the full consideration of business relationships. A formal business rule description language is defined to describe various types of business relationships and their effects on QoS. This research adopts the genetic algorithm to discover the near optimal service composition plan. Business rules are incorporated in computing fitness values and performing crossover and mutation functions. The experimental results demonstrate that the proposed approach is able to handle various business rules properly in selection.