Evaluating the compatibility of conversational service interactions

Abstract

Service-oriented systems live in an open world, one in which their functionality and quality of service depend on how the services they interact with evolve. System adaptation has been indicated as a way to cope with the evolution these partner services may have. When a partner does not behave as expected, in an adaptable system we can substitute it with an alternative compatible one. Finding a compatible alternative, however, is a difficult task if we consider conversational services that impose a specific interaction protocol and specific data-types. In this paper we introduce Interaction Sequence Charts (ISC) as an effective notation for describing the interactions a service has with its partners, and an algorithm that uses these charts to establish a "degree of compatibility" between interacting services. The algorithm considers both interaction protocol requirements and datatype similarity, for which fuzzy techniques are adopted. The expressive power of ISC is validated by using it to describe the complex behaviour that can be defined using BPEL 2.0, while the algorithm is validated on an example in the field of Tele-Radiology, and shown to be advantageous in practice.