On-The-Fly computing: automatic service discovery and composition in heterogeneous domains

In software markets of the future, customer-specific software will be developed on demand based on distributed software and hardware services. Based on a customer-specific request, available service offers have to be discovered and composed into sophisticated IT services that fulfill the customer's request. A prerequisite of this vision are rich service descriptions, which comprise structural as well as behavioral aspects of the services, otherwise an accurate service discovery and composition is not possible. However, automatic matching of service requests and offers specified in rich service descriptions for the purpose of service discovery is a complex task, due to the multifaceted heterogeneity of the service partners. This heterogeneity includes the use of different specification languages, different underlying ontologies, or different levels of granularity in the specification itself. In this article, we present a comprehensive approach for service discovery and composition, which overcomes the underlying heterogeneity of the service partners. Based on a realistic case study of our industrial partner from the e-tourism domain, we first introduce an automatic matching mechanism for service requests and offers specified in a rich service description language. In addition, we propose an automatic service composition approach, which determines possible service compositions by composing the service protocols through a composition strategy based on labeled transition systems.