Peter Fettke

Predicting process behaviour using deep learning

Abstract Predicting business process behaviour is an important aspect of business process management. Motivated by research in natural language processing, this paper describes an application of deep learning with recurrent neural networks to the problem of predicting the next event in a business process. This is both a novel method in process prediction, which has largely relied on explicit process models, and also a novel application of deep learning methods. The approach is evaluated on two real datasets and our results surpass the state-of-the-art in prediction precision.

Der Referenzmodellkatalog als Instrument des Wissensmanagements: Methodik und Anwendung

Referenzmodelle konnen als Speicher fur explizites Domanen wissen interpretiert werden. Obgleich in der Literatur inzwischen zahlreiche Referenzmodelle dokumentiert sind, ist der Zugang zu diesen Wissensspeichern uneinheitlich und nicht systematisch moglich. Referenzmodellkataloge ermoglichen Referenzmodelle nach einheitlichen Gesichtspunkten zu ordnen und nach spezifischen Gesichtspunkten zugreifbar zu machen. Damit bilden sie ein zentrales Instrument zur Erschliesung der Wissensquelle Referenzmodell und bilden die Schnittstelle zwischen den Prozessen der Entwicklung von Referenzmodellen einerseits und der Anwendung von Referenzmodellen andererseits. Es werden zwei Referenzmodellkataloge vorgestellt: Der erste Katalog beschreibt Moglichkeiten der Arbeitsplanmodellierung. Der zweite Katalog ist ein nach Wirtschaftszweigen systematisierender Uberblicks-Katalog uber vorhandene Referenzmodelle. Kunftige Arbeiten werden die Entwicklung weiterer Referenzmodellkataloge thematisieren.

Organizational and Technological Options for Business Process Management from the Perspective of Web 2.0

Corporate operative systems are often highly dynamic, a fact which is only insufficiently taken into account by recent process management approaches. In contrast, the perspective of Web 2.0 opens up new options for action in process management. In this contribution, we figure out new organizational and technological options of process management using a design-oriented research approach. The analysis especially considers the aspects of self-organization and collective intelligence in process management. We conceptually develop options for action and illustrate them based on a prototype platform for process management. The paper is complemented by a presentation of real-world application scenarios in the construction industry and results of an evaluation of the design-oriented research approach.

Report : The Process Model Matching Contest 2013

Process model matching refers to the creation of correspondences between activities of process models. Applications of process model matching are manifold, reaching from model validation over harmonization of process variants to effective management of process model collections. Recently, this demand led to the development of different techniques for process model matching. Yet, these techniques are heuristics and, thus, their results are inherently uncertain and need to be evaluated on a common basis. Currently, however, the BPM community lacks established data sets and frameworks for evaluation. The Process Model Matching Contest 2013 aimed at addressing the need for effective evaluation by defining process model matching problems over published data sets.

A Deep Learning Approach for Predicting Process Behaviour at Runtime

Predicting the final state of a running process, the remaining time to completion or the next activity of a running process are important aspects of runtime process management. Runtime management requires the ability to identify processes that are at risk of not meeting certain criteria in order to offer case managers decision information for timely intervention. This in turn requires accurate prediction models for process outcomes and for the next process event, based on runtime information available at the prediction and decision point. In this paper, we describe an initial application of deep learning with recurrent neural networks to the problem of predicting the next process event. This is both a novel method in process prediction, which has previously relied on explicit process models in the form of Hidden Markov Models (HMM) or annotated transition systems, and also a novel application for deep learning methods.

Towards the Reconstruction and Evaluation of Conceptual Model Quality Discourses – Methodical Framework and Application in the Context of Model Understandability

Within the information systems (IS) discipline conceptual models have gained tremendous importance in the past years. Different approaches for systematic model quality evaluation have emerged. However, these approaches are based on different understandings, definitions as well as operationalizations of the term “model quality”. In this article we refrain from conceptualizing and operationalizing model quality a priori. In contrast, assuming that the determination of model quality and appropriate criteria are negotiated in a discourse between modelers and model users based on their different perspectives, we develop a methodical framework for the critical reconstruction and evaluation of conceptual model quality discourses in order to identify relevant model quality criteria and understandings. Our method is exemplarily applied for the reconstruction of the discourse on the quality criterion model understandability based on relevant laboratory experiments. This application shows that many research results on model understandability are hardly comparable due to their different basic assumptions and should preferably be interpreted based on a methodical reconstruction of underlying understandings.

Collaborative Business Process Modeling with CoMoMod - A Toolkit for Model Integration in Distributed Cooperation Environments

Business Process Management (BPM) has gained remarkable importance in research and practice in the last years. Innovative information and communication technology (ICT) is a driver for the effectiveness and efficiency of enterprise collaboration and the management of inter-organizational business processes. In this context, cooperating enterprises need a shared understanding of their own processes, those of their partners as well as of emerging inter-organizational process structures. In order to create such a shared understanding, the concept of collaborative modeling can serve as a useful approach. This article introduces a concept for collaborative business process modeling and its implementation in the CoMoMod tool. The tool supports several aspects of collaborative process modeling, such as simultaneous work of spatially distributed modelers on one process model diagram. Furthermore the tool permits synchronously creating a collaborative model with different modeling languages by automatically mapping and converting the different model representations. The proof of concept utilizes Event-driven Process Chains (EPC) and Petri Nets. The usage of an integrated chat messaging service supports communication during the process of collaboratively modeling business process.

Clustering Traces Using Sequence Alignment

Process mining discovers process models from event logs. Logs containing heterogeneous sets of traces can lead to complex process models that try to account for very different behaviour in a single model. Trace clustering identifies homogeneous sets of traces within a heterogeneous log and allows for the discovery of multiple, simpler process models. In this paper, we present a trace clustering method based on local alignment of sequences, subsequent multidimensional scaling, and k-means clustering. We describe its implementation and show that its performance compares favourably to state-of-the-art clustering approaches on two evaluation problems.

Towards Green Business Process Management: Concept and Implementation of an Artifact to Reduce the Energy Consumption of Business Processes

The consideration of ecological objectives has been identified as one of the major topics for IS research and Business Process Management (BPM). An essential precondition for advancing the emerging discipline of Green BPM still is the availability of methods and tools for detecting and matching the resource usage of a business process to its individual steps. In this paper, a conceptual integration model for the energy consumption of IT components and business processes is developed for the specific setting of administrative business processes. The integration concept is implemented as a software prototype, following a design-science approach. Its application is demonstrated in a sample scenario. Findings show, that the integration concept and the software prototype increase the transparency of the energy consumption of administrative business processes, and enable users to save energy. Furthermore, the prototype can be used to develop and validate effective methods for creating energy-efficient business processes.

What in Fact Is the Role of Stylized Facts in Fundamental Research of Business and Information Systems Engineering

In the discussion initiated by Winter on “What in fact is fundamental research in business and information systems engineering?” in issue 2/2009, Zelewski submitted the interesting proposal to use the concept of stylized facts as developed in economics in business and information systems engineering (BISE) in order to promote theory building within our discipline (Winter et al. 2009). Stylized facts constitute elements of knowledge in the form of generalized statements that describe the important characteristics and relationships in the context of a studied phenomenon and are widely supported empirically. For example, the empirically proven relationship between the longterm abstention from changing the employer in Japanese companies and a much larger salary increase compared to U.S. companies is a typical stylized fact in economics. When using stylized facts in the context of BISE, a number of interesting questions arise: What are the typical stylized facts within BISE? How can stylized facts be obtained systematically? What is the importance of stylized facts in design-oriented disciplines? Which role do stylized facts play in practice? What is the difference between technological rules and stylized facts? To what extent can we consider stylized facts to be “technological” knowledge about cause-effect-relationships that can be utilized for the construction of artifacts of information and communication technology? Which role do stylized facts play in empirical research? Which contribution can stylized facts make for building theories in BISE if a multiperspective research approach is assumed? Currently, only scattered experience exists in the application of stylized facts in BISE (Fettke et al. 2010; Houy et al. 2009, 2011). In order to achieve a broad perspective on the issue in the course of this discussion, well-known representatives from the field of business administration, the German-speaking BISE as well as from Information Systems Research were invited to comment on the topic. The following authors accepted our invitation to this discussion (in alphabetical order): Prof. Dr. Ulrich Frank, Chair of Information Systems and Enterprise Modelling, University of Duisburg-Essen, Germany; Prof. Dr. Armin Heinzl, Chair of General Management and Information Systems, Mannheim University, Germany; Prof. Dr. Juhani Iivari, Department of Information Processing Science, Oulu University, Finland; Prof. Dr. Barbara E. Weisenberger, Chair of Controlling and Business Accounting, Justus-Liebig University of Giesen, Germany; Prof. Dr. Stephan Zelewski, Institute of Production and Industrial Information Management, University of Duisburg-Essen, Germany. Within their contributions the authors highlight different facets of the generation and application of stylized facts that appear relevant and important for BISE. Barbara E. Weisenberger outlines the concept and the derivation of stylized facts based on a case study in the field of IT-based controlling systems. Thus, several empirical studies have shown that the “increasing flexibility of IT systems in accounting and control has a positive impact on management accounting and control”. The author argues that the concept contributes to aggregate and structure fragmented knowledge in a research area, with particular emphasis on revealing research gaps. At the same time she points out that the derivation of stylized facts is the result of comprehensive research endeavors. The contribution by Stephan Zelewski considers the concept of stylized facts in the tension between the more designoriented German-speaking BISE on the one hand and the more behavioristic Anglo-Saxon Information Systems Research on the other. He argues that stylized facts are not directly but indirectly fruitful for the design-oriented BISE: Accordingly, stylized facts form a “seed of crystallization” for the development and testing of theories with causal explanatory power, making it possible to “explain and verify prognoses regarding alleged effects of the recommended design measures”. Despite the attractiveness of the concept, the author also points to difficulties in the implementation in current science. Armin Heinzl identifies application areas for stylized facts in BISE, especially