Complete Trace Detection in Event-Stream Based Process Discovery



Sebastiaan J. van Zelst

Scientific Assistant



MSc Thesis project

Title: Complete Trace Detection in Event-Stream Based Process Discovery

Author: Yi-Chan Tsai

1st examiner: Wil M.P. van der Aalst

2nd examiner: Sebastiaan J. van Zelst


Process discovery is a main field in the area of process mining that aims to discover a process model based on business process execution data. The majority of process mining algorithms are designed to use event logs, which are historical offline data. In this thesis, we use online streams of events as a primary source of input. One method to handle such a data type is to store events in a temporary event log, and then apply normal process discovery algorithms. However, the temporary event log inherently contains incomplete cases. Direct application of process discovery algorithms on the temporary event log containing many incomplete cases often returns an underfitted model, which allows more process executions than actually described in the event streams. In this thesis, we propose a method to filter out incomplete cases in a temporally stored event log so that we can discover a more precise process model. The method is evaluated on synthetic and real-life event data sets. The evaluation shows that the method increases the quality of process models discovered on the basis of event streams.