PEM4PPM: A Cognitive Perspective on the Process of Process Mining.

During the last decades, process mining (PM) has matured and rapidly increased in its adoption. Making sense of data is a main part of the work of PM analysts, which involves cognitive processes. Recent work has leveraged behavioral data to explain these processes. Still, the process of process mining (PPM) is yet to be well understood and a theoretical foundation for explaining how these processes unfold is missing. This paper attempts to fill this gap by understanding how PPM data can be analyzed in a theory-guided manner and what insights can be gained from this analysis. To investigate these aspects, we analyzed verbal data and interaction traces obtained from analysis sessions with 29 participants performing a PM task. The analysis was based on the Predictive Processing (PP) theory and the derived Prediction Error Minimization (PEM) process, anchored in cognitive science. The results include (1) a theoretical adaptation of the PEM theory to the PPM context, (2) four strategies utilized by PM analysts, identified, and validated based on the adapted theory, and (3) an understanding of the differences in performance between analysts using different strategies and independence of the expertise level and the strategy choice.