Matching observed behavior and modeled behavior: An approach based on Petri nets and integer programming

Abstract

Inspired by the way SAP R/3 and other transactional information systems log events, we focus on the problem to decide whether a process model and a frequency profile “fit” together. The problem is formulated in terms of Petri nets and an approach based on integer programming is proposed to tackle the problem. The integer program provides necessary conditions and, as shown in this paper, for relevant subclasses these conditions are sufficient. Unlike traditional approaches, the approach allows for labeled Petri nets with “hidden transitions”, noise, etc.