Dealing with Unexpected Runtime Outcomes Within Process Models

Process models are designed to describe the required tasks to achieve a desired business goal. These models can be verified to be compliant with additional requirements, like regulations and business requirements. This means that process models can be designed and verified to behave according to some desired requirements. However, it is possible that some of the outcomes at runtime deviate from the design predictions of the model, which would render the model and the compliance verification obsolete. In this paper, we propose an approach aiming at detecting such runtime deviations through representing the tasks' outcomes as data ranges. When a deviation is detected, the approach re-evaluates compliance of the model given the unexpected outcomes during the execution, and if necessary and possible it adapts the remainder of the model's execution to preemptively avoid breaching the requirements.