Abstract

Automatic web-service composition aims at automating the design of an appropriate combination of existing web services to achieve a global goal. Most proposed AWSC approaches only consider input/output parameters and quality features of services. However, most real-world web services have applicable conditions and require constraints to be considered according to the execution context of composite services. Constraint verification has a significant impact on the composition and execution of composite services. In particular, runtime verification of service constraints can result in the failure of the execution of composite services and eventually waste computational resources and may incur monetary costs. In addition, traditional adaptation approaches for web service composition consider recovery in case of failure when a service becomes unavailable. They do not take into account changes and limitations in service execution environment which potentially can affect the execution of a wide range of services. Externally-defined constraints are likely to be defined and become or cease to be applicable after the composite service has been deployed.
In this thesis, we propose a novel approach to model and verify different types of constraints inside composite services. We not only consider input/output parameters but also the values that can be assigned to parameters during design and execution of composite services.
In addition, we provide novel failure recovery and adaptation approaches for different types of constraints according to the execution context of composite services. In our solution, we develop a new structure including alternative composite services to recover broken composite services and adapt to external constraints. We finally propose a brokerage architecture including all proposed approaches for constraint-aware service composition and adaptation.