Much of the research on workflow management has assumed a centrally-controlled work environment where common definitions for tasks and information are known and respected by all workgroups using the system. Autonomous groups, however, may be free to redefine or eliminate individual activities or sequences, making predefined coordination with others highly problematic. While the general problem of unconstrained workflow coordination is difficult, some classes of inter-operating workflows are constrained enough to be manageable and studyable---for example, repetitive, operations-level workflows that constitute the much of the work performed in many organizations. This project aims to produce a conceptual and formal model of coordinating interoperating workflow management systems (WFMSs) that allows compensations for task redefinitions to be fully or partially automated. The model will be partially validated by constructing an executable version of the model, and exercising the model with a set of test cases. An approach of progressive model refinement will be used to gain the ability to "grow" the model, using the executable model as a simulator and incorporating simulation findings back into the model. The transparent, well-articulated architecture of the model will improve understanding of dynamic task coordination and will enable improved functionality of interoperating WFMS in dynamic environments.
