An Embedding-Based Approach to Detect Semantic Anomalies in Event Logs

Anomaly detection in process mining aims to detect deviant or unexpected behavior in event logs. Existing techniques for this task are primarily frequency-based and argue that behavior is anomalous because it is unusual. In this work, we overcome this caveat and focus on anomaly detection from a semantic perspective by identifying semantically inconsistent execution patterns using state of the art NLP techniques.

Kiran Rouven Busch joined the Kühne Logistics University as a PhD candidate in the field of Data Science under the supervision of Prof. Dr. Henrik Leopold in October 2021. In his research, Kiran is mainly concerned with the interplay between information systems and business processes. He is particularly interested in how to leverage technology from the field of artificial intelligence (such as machine learning and natural language processing) to analyze and support the execution of business processes. Before joining the PhD program at the KLU, Kiran received a Master of Science in Data Science from the Darmstadt University of Applied Sciences. He conducted his master thesis in the principality of Liechtenstein, optimizing the output of flow production lines in an automotive production facility using data science approaches. Furthermore, Kiran gained multiple practical experiences in the field of data science during domestic and international internships in the automotive, pharmaceutical and energy industries.