Process Discovery of Collaboration Business Process Containing Invisible Task in Non-Free Choice by using Modified Alpha ++

Abstract

Process mining is a set of techniques that combine the scientific point of view of the data with the point of view of running processes correctly. The process mining can be performed with several algorithms, such as Alpha miner, Inductive Miner, and Fodina. Those existing algorithms only form a process model. Nevertheless, a system can be built not only a process model but also collaboration process models. Collaboration process models are several interconnected process models that are carried out by sending messages between those activities. Besides shortcoming of depicting collaboration process models, another issue is called an Invisible Task in a Non-Free Choice relationship. This issue occurs when particular conditions, i.e. skip, redo and switch, are happened in activities which are interdependent with other activities. Alpha $ is a pioneering algorithm that describes the Invisible Task in Non-Free Choice (IT-in-NFC); however, this algorithm cannot handle collaboration process models. This study proposes Modified Alpha ++ that contains rules of storing messages of activities and modifies Alpha ++ algorithm with additional Invisible Task rules to present Alpha $. The evaluation results show a process model obtained by Modified Alpha ++ is better than other algorithms, i.e. Alpha ++, Fodina and Inductive Miner. That statement is proven with the Fitness, Precision, and F -Measure values of the process model by Modified Alpha ++ are 1. On the other hand, process models of Alpha ++, Fodina and Alpha ++ obtain less than one F -Measure value.