History-Aware Process Monitoring
MSc Thesis Project
Title: History-Aware Process Monitoring
Student: Daniel Schuster, MSc.
1st Examiner: Prof.Dr.ir Wil M.P. van der Aalst
2nd Examiner: Prof.Dr.-Ing. Ulrik Schroeder
Daily Supervisor: Dr.ir. Sebastiaan J. van Zelst
The execution of business processes in companies generates valuable data in information systems. Process Mining utilizes such data to generate knowledge of the underlying processes. Within process mining, conformance checking is a subdiscipline that allows us to verify whether the process execution is in accordance with a given process model. Many techniques exist that address this question. However, most of these techniques rely on already finished process executions, i.e., they are defined in an online setting. As a consequence, process deviations are detected after the process execution has finished. Only a few techniques allow us to analyze ongoing process executions, i.e., online conformance checking. However, these techniques either abstract from the process model and the data in such a way that the examination of deviations becomes difficult or these techniques use approximations. Hence, optimality is not guaranteed, i.e., false process-deviation detections may occur. In this thesis, a novel approach is presented to monitor ongoing process executions. In contrast to existing techniques, the presented approach is an exact algorithm, yielding optimal solutions, i.e., false process deviation detections cannot occur. Moreover, the presented algorithm is parameter-free. We evaluated the algorithm using publicly available real data sets and compared our algorithm to other approaches. The results show that the proposed algorithm has a comparatively search efficiency compared to other algorithms, but has slightly longer computation times, while guaranteeing optimality at the same time.