A probabilistic approach to analyze and adjust time constraints in workflow management system

Dealing with time and time constraints is crucial in designing and managing business processes, so time management should be an important part of workflow management systems. One key issue in time management is to analyze the feasibility of time constraints. Besides, when time constraints are violated, they should be adjusted to regain a satisfaction state. Traditional work on time constraint analysis considers discrete activity (execution) durations and presents qualitative analysis result such as time constraints are satisfied or violated. However, the dynamic nature of workflows causes high uncertainties in workflow process, thus activities of workflow process probabilistically satisfy their time constraints, e.g., 95% of activities satisfy their time constraints. Therefore, traditional deterministic analysis is too rigorous. For such an issue, probabilistic time constraint workflow nets (PTCWF-nets) are introduced to describe uncertainties in time-constrained workflow process. Based on PTCWF-nets, a probabilistic approach is proposed at worklow build-time. This approach analyzes every activity's probability of satisfying time constraints in a stochastic way. This quantitative analysis result allows process designer to flexibly check the feasibility of time constraints. The result also provides a precise estimation to guide time constraint adjustment. Moreover, the proposed approach is implemented in a real-world workflow management system, and its effectiveness is evaluated by means of a concrete example.