Table of Contents

Process Mining and Verification

B.F. van Dongen
PhD Thesis. Technische Universiteit Eindhoven, Eindhoven, The Netherlands, 2007.

Abstract

Organizations continuously try to improve the way they do business. More and more, large and complex information systems are used to support and/or analyze all aspects of operational processes. Regardless whether these information systems are primarily used for support or analysis, they all typically record information about which activities are performed, at which time, by whom and in the context of which case. Therefore, these records, or event logs provide a good starting point for the analysis of the organization through process mining and verification.

The research area of process mining focuses on extracting information about operational processes by examining event logs of information systems. By assuming that events relate to activities executed within the context of a single case, process discovery algorithms are capable of constructing models that accurately describe a process. As these models are based on objectively compiled information, they provide an ideal starting point for an in-depth analysis of an organization. Using the results of such analysis, organizations then have the opportunity to improve their operational processes.

Most of today’s process discovery algorithms provide models in terms of Petri nets. While such Petri nets are a good mathematical formalism, for which a wide-range or analysis techniques is available, they are only indirectly used in practice. Practitioners, using systems such as SAP R/3, or the IDS-Sheer’s Aris Platform, often use a less formal modelling language named Event-driven Process Chains (EPCs). Therefore, discovery algorithms were developed that construct EPCs, accurately describing the operational processes. Furthermore, verification techniques were developed to show that the discovered EPCs are also error-free. To show that constructing error-free EPCs is not a trivial task, the verification techniques were applied to the SAP reference model, showing these models contain an astonishing number of errors.

To support the implementation of process mining (and related) research, the ProM framework has been developed. The ProM framework is an open-source tool that allows for fast and easy implementations of new algorithms. As it currently contains almost 250 plugins implemented by researchers all over the world. The core development of ProM is done at TU/e, where the internationally leading group in process mining and business process management resides.

Download PDF (16 MB)