A method to improve workflow net decomposition for process model repair

Abstract

Creating a process model (PM) is a convenient means to depict the behavior of a particular information system. However, user behavior is not static and tends to change over time. In order for them to sustain relevant, PMs have to be adjusted to ever-changing behavior. Sometimes the existing PM may be of high value (e.g. it is well-structured, or has been developed continuously by experts to later work with), which makes the approach to create a brand-new model using discovery algorithms less preferable. In this case, a different and better suitable approach to adjust PM to new behavior is to work with an existing model through repairing only such PM fragments that do not fit the actual behavior stated in sub-log. This article is to present a method for efficient decomposition of PMs for their future repair. It aims to improve the accuracy of model repair. Unlike the ones introduced earlier, this algorithm suggests finding the minimum spanning tree of undirected graph’s vertices subset. It helps to reduce the size of a fragment to be repaired in a model and enhances the quality of a repaired model according to various conformance metrics.