Improved HTN planning approach for service composition

AI planning techniques is exploited for automatic service composition by representing service composition as a planning problem. An enhanced HTN planning method is developed in which action decomposition is used as plan refinements in partial-order planning. How abstract planning constructs map to the domain of Web services is also discussed, which is based on semantic type matching. Furthermore, execution for service composition is separated from planning and an ECA-based workflow engine is used to execute the service-dependent plan.