Abstract. Within organizations, workflow systems can automate business processes by centrally coordinating activity sequences. But outside their borders, organizations are autonomous entities that cannot be subject to centralized process control. Their internal processes are autonomously defined and controlled, and what they need is to synchronize those concurrent processes. Just like Petri nets are a valuable tool to model activity sequencing in local business processes, π-calculus becomes a useful tool to model concurrency in inter-organizational processes. After a review of the main developments in cross-organizational workflow management, this paper illustrates the use of π-calculus to model the interactions between business processes running concurrently in different organizations. These interactions range from invoking external services to more complex patterns such as contract negotiation and partner search and selection. The paper concludes with a case study that illustrates the application of the proposed approach in a more realistic business scenario.
BackgroundResearch on the application of workflow management in inter-organizational environments has produced a wealth of interesting approaches on how to deal with the problem of coordinating processes that span multiple organizations. The available solutions for cross-organizational workflow management have been developed in recent years and they have followed a path towards increasing flexibility, from focusing on "low-level" issues of workflow systems interoperability to more flexible, "higher-level" architectures based on contracts and views. Putting these developments in perspective, several trends can be identified, including:Workflow interoperability mechanisms -after the publication of the Workflow Reference Model [1], which described four models of interoperability between workflow systems, supporting cross-organizational workflows seemed to be a matter of selecting the most appropriate interoperability model for a given scenario. In [2], Anzböck and Dustdar describe the application of these models to medical imaging workflows that resemble cross-organizational workflows.