Abstract:
Workflow technology has met with success in a variety of industries, although several limitations have emerged. One such drawback is the inflexibility of specification languages, including a lack of support for inter-task dependencies. Expressiveness of the specification language is believed to be a determining factor of workflows applicability and its industrial value as solution for process support. This paper attempts to address this limited language expressiveness by suggesting an alternative approach to modelling that more accurately captures behavioural information about tasks and enables greater precision when modelling inter-task dependencies. Current workflow technology associates one generic, predefined finite state machine with each activity in a process, and inter-task dependencies of the type 'completion of one activity triggers scheduling of the next activity' are also enforced. The potential improvement relaxes these constraints to enable the specification of user-defined finite state machines to represent each activity and support the modelling of inter-task constraints at the activity state level. In this paper, we present an introduction to this modelling extension and demonstrate the applicability of existing workflow verification algorithms to these more descriptive process models.
