Post by Manal L.

Many AI and process mining techniques are evaluated under a hidden assumption: that the underlying process behavior is stable. However, in reality, processes evolve… 📄 Our newest publication addresses this challenge by formalizing how we can generate and adapt ground truth models when process behavior changes 📈 💡This matters because it means that evaluation setups may not fully reflect how AI behaves in real-world settings where change is constant. Although the paper focuses on the mathematical foundations 🔢, it had practical implications for my PhD. The foundations allowed me to simulate audit environments: safe testbeds that I later used throughout my research to develop and evaluate AI techniques for auditing. My goal was to better understand when AI techniques work well, where their limitations lie, and how they can be applied more confidently in practice! 🙏 This is probably the most formal piece of research I worked on during my PhD 📚thanks in large part to Nicola Gigante and Marco Montali. THANK YOU for welcoming me with open arms in Bolzano. Some of the best brainstorming sessions of this research happened there during long discussions, and plenty of GOOD coffee! 🌄☕ Thank you, Mieke and Benoît for supervising me during my PhD, and for the many nice collaborations we had! 💬 Interested in learning more? I'm always happy to chat! Or, if you'd rather go straight to the paper: https://lnkd.in/d3phFBQR #BusinessInformatics #Research @Digital Future Lab @UHasselt @Universiteit Maastricht @Free University of Bozen-Bolzano