Martin Käppel has defended an outstanding Ph.D. thesis that addresses a matter of genuine significance in process mining: data scarcity. His work thereby advances process monitoring applications at their very foundation: by intervening directly at the level of logged data.
In doing so, his research elevates existing methods and techniques, particularly in the realm of next-event prediction, to a new level of performance when the techniques he developed are applied prior to the learning phase. Martin’s approaches are rooted in the broader research field of small sample learning and have been thoughtfully adapted to our (BPM) domain.
A first line of contribution lies in adapting predictive models to ensure that rare process execution variants are weighted more equitably in relation to frequent standard variants, drawing on principles of cost-sensitive learning. In addition, Martin employs two distinct approaches to data augmentation: one transformation-based, and another generative in nature. The latter builds on large language models and is also presented in a current journal article, which I will introduce here shortly. In this context, LLMs are leveraged to synthetically generate previously unobserved behavior.
Martin, your work brings together academic rigor and excellence with tangible problem-solving impact and strong practical relevance. Under the supervision of Stefan Jablonski at Universität Bayreuth, you have produced a thesis that stands firmly in the finest tradition of Die Wirtschaftsinformatik research – quite remarkable for someone trained as a computer scientist ;). As one of the very few works at your faculty, it has been awarded the highest distinction, “summa cum laude.” I am delighted that you will continue your journey at our chair, now as a postdoctoral researcher – marking, perhaps, your final step toward becoming a “fully-fledged” business informatics scholar at FAU WiSo.
The thesis is entitled: “Entwicklung von Small Sample Learning Methoden für die prädiktive Geschäftsprozessüberwachung”
The photo shows the examination committee: Ruben Mayer, two Martins, Stefan Jablonski, Jörg Müller
We congratulate Martin on successfully defending his dissertation!
