On The Expressive Power of Service Models for Automated Service Composition
Wang, Yin; Motahari-Nezhad, Hamid; Singhal, Sharad
Keyword(s): No keywords available.
Abstract: Automatic service composition is an important problem in service oriented computing. Existing service models, whether embodied in specification standards or implicit in composition methods, offer tradeoffs between the level of detail at which component services are described and resulting opportunities for automated composition. For example, WSDL describes service interfaces whereas OWL-S describes preconditions and effects, and the two standards support different approaches to automated composition. The precise relationship between service models and the constraints they place on automated composition, however , remains unclear. It is therefore difficult to understand the full implications of commitment to a particular way of describing services. This paper presents a formal framework that relates service models to composition capabilities. Our framework covers the three most commonly used service models: input/output, precondition/effects, and automaton models. It uses logic formulae and formal languages to characterize precisely the expressiveness and composition capabilities inherent in each service model. Our framework facilitates choosing an appropriate service model for a specific composition task. Applied to common workflow patterns, it identifies minimally expressive service models that enable composition of these patterns. As a case study, we analyze 33 workflows used for IT transformation services within HP Enterprise Services. Our analysis identifies the appropriate service model for the automated construction of these workflows.
External Posting Date: March 6, 2010 [Fulltext]. Approved for External Publication
Internal Posting Date: March 6, 2010 [Fulltext]