Discovering High-level BPMN Process Models from Event Data

Process mining is a well established research discipline comprising approaches for process analysis based on the history of process executions. One of the main directions in the process mining field is process discovery. Process discovery aims to develop methods for constructing process models from the event logs. The ultimate goal of process discovery is to obtain readable process models, which represent process behavior in the best possible way. Process discovery techniques rarely use higher-level process modeling notations like BPMN and tend to focus on the control-flow perspective. However, end-users are more familiar with notations like BPMN and are interested in the data and resource perspectives. This paper gives an overview of existing process discovery techniques and presents a BPMN meta-model, which describes models that can be obtained from the event logs using existing discovery techniques. Besides that the paper presents an integrated discovery approach for the construction of high-level BPMN models that comply with the BPMN meta-model. The proposed integrated approach was applied to real-life event logs and it was shown that it allows for obtaining readable process models, which reflect the process behavior