Estimating Structures of Business Process Models from Execution Logs

Business Process Management (BPM) has been recognized to be important for Business Processes (BP) and enterprise information systems to continuously bring their benefits. Understanding actual BP models is a key step in BPM because predefined models are different from actual ones, which analysis and decision-making should stand on. However, it depends on costly interviews and observations. In this paper, we propose a method of estimating actual BP model structures from execution logs in information systems for effortless BPM. This method can utilize practical execution logs and appropriately estimate structures compared with previously proposed methods. It can also reveal how each work order was transacted and routed so that exceptional situations can be investigated in detail. We apply it to a synthetic execution log and a real-world execution log, then demonstrate its validity in terms of algorithms and its usefulness in BPM. Our method makes it possible to attain continuous BPM.