The paper titled "LLMs that Understand Processes: Instruction-Tuning for Semantics-Aware Process Mining", by Vira Pyrih, Adrian Rebmann, and Han van der Aa reports on experiments that assess the potential of using instruction-tuning as a means to obtain LLMs that are better at performing process-analytical tasks than out-of-the-box models. The findings suggest that instruction-tuning is a promising path toward more scalable applications of LLMs for semantics-aware process mining, particularly for discovery and predictive tasks.
Read the full paper here.