Process Discovery Method in Dynamic Manufacturing and Logistics Environments
Process mining is a promising way to extract insight knowledge on business processes in manufacturing and logistics. However, implementing process mining is challenging in dynamic and complex environments as the discovered process models may not reach the
- PDF / 1,013,966 Bytes
- 21 Pages / 439.37 x 666.142 pts Page_size
- 46 Downloads / 187 Views
3
International Graduate School for Dynamics in Logistics, University of Bremen, Bremen, Germany [email protected], [email protected] 2 Faculty of Business Studies, University of Applied Science Emden/Leer, Emden, Germany [email protected] Department of Mathematics/Informatics, University of Bremen, Bremen, Germany
Abstract. Process mining is a promising way to extract insight knowledge on business processes in manufacturing and logistics. However, implementing process mining is challenging in dynamic and complex environments as the discovered process models may not reach the aspired quality. As a result, current process mining solutions do not hold in practical situations effectively in the domain of manufacturing and logistics. In this paper, we propose a sequence clustering methodology based on Markov Chains and Expectation-Maximization. We propose two approaches to improve the existing method of sequence clustering which provide improvement of finding the main behavior and its variants for each process cluster. We evaluate the proposed methodology with real-world data sets by measuring model quality dimensions. The results demonstrate that the proposed methodology is capable to improve process discovery when confronted with dynamic and complex business processes. The resulting models present the main behavior of business processes miming and process variants with a satisfying process model quality.
Keywords: Process discovery Manufacturing · Logistics
1
· Process mining · Sequence clustering ·
Introduction
The business processes in the manufacturing and logistics domain are considered as a dynamic environment. They are frequently changed and expanded to achieve optimal results. For several years a great effort has been devoted to the improvement of business process performance in manufacturing and logistics. To improve processes, insight and comprehensive knowledge of actual and variant behaviors of business processes are crucial for decision making, problem-solving, c Springer Nature Switzerland AG 2020 M. Freitag et al. (Eds.): S-BPM ONE 2020, CCIS 1278, pp. 143–163, 2020. https://doi.org/10.1007/978-3-030-64351-5_10
144
W. Intayoad et al.
and enhancing productivity [10]. The effective way to automatically extract such valuable knowledge from data is process mining. Process mining analyzes the data generated during the real operations, the so-called ‘event log’. An event log is the collection of data that relates to products, processes, machines, planning, and logistics performance, which can be used to explore and discover valuable information and knowledge [7]. Process mining has been successfully deployed in various domains such as health-care, education, software implementation, telecommunication, and logistics s [5,14,18–21]. Process discovery is an essential application of process mining. However, process discovery algorithms encounter problems when they are used to analyze the event logs from high dynamic environments. The business processes are frequently changed an
Data Loading...