A Language-Based Framework for Analyzing Service Representation Models and Service Composition Approaches

摘要

Automatic service composition is an important problem in service computing. Existing works on service composition assume different representation models with various expressiveness for component services as well as for the composition logic. This has made understanding these composition approaches and their applications difficult. This paper presents a novel language-based formal framework that acts as a basis for understanding the expressiveness of service models and composition logics. It covers the three common service representation models, i.e., input/output, precondition/effects, and automaton models. The novelty of this framework is in using the execution language of service composition, characterized by logic formulae and formal languages, as a common ground to analyze service models and composition logics. We have studied the composition capability of each service model in terms of enabling common workflow patterns, and analyzed the computational complexity of composition algorithms of different service models. The framework simplifies the job of understanding service representation models and composition logics for a given service composition task.