Folding Marked Generalized Stochastic Petri Nets for Time Prediction in Business Processes

Abstract

Generalized Stochastic Petri Nets (GSPNs) can be used for performance analysis of business processes. Recently, it was shown that foldings of a GSPN, i.e., a set of model reduction rules, help to avoid over-fitting of the model with respect to the performance characteristics of a process. Yet, these foldings ignore the marking of a GSPN and, thus, are applicable solely for steady-state analysis. In this paper, we discuss how foldings may be lifted to marked nets and provide an assessment of stateful foldings for sequential GSPNs.