Data Flow Correctness in Adaptive Workflow Systems

Enterprises must be able to quickly adapt their business processes to react to changes in their environment. Needed business agility is often hindered by the lacking flexibility of contemporary workflow systems. In response to this inflexibility, adaptive workflow systems have emerged, which enable the dynamic adaptation of running workflows. One of the most important challenges in this context is to avoid inconsistencies and errors. So far, approaches providing respective correctness criteria for dynamic workflow change have mainly focused on control flow correctness (e.g., avoidance of deadlocks). However, little attention has been paid to data flow correctness even though this is crucial for any application of dynamic workflow change in practice. Specifically, missing or inconsistent input data of workflow activities, for example, can lead to blocking or breakdown of the underlying workflow system. This paper deals with fundamental challenges related to data flow correctness. We revisit and discuss data flow correctness at different phases of the workflow life cycle (i.e., buildtime and runtime), and show how data flow correctness can be ensured in an efficient way when dynamically changing a workflow.