Didden, L., Fahland, D., & Zerbato, F. (2026). Visually Retracing and Comparing Filter Steps in Exploratory Process Mining. In H. van der Aa, R. Guizzardi, S. Hacks, & L. Pufahl (Eds.), Enterprise, Business-Process and Information Systems Modeling: 27th International Conference, BPMDS 2026, and 31st International Conference, EMMSAD 2026, Verona, Italy, June 8–9, 2026, Proceedings (pp. 73-89). (Lecture Notes in Business Information Processing (LNBIP); Vol. 594). Springer. https://doi.org/10.1007/978-3-032-28274-3_5
Abstract
Exploratory process mining commonly involves applying multiple filters in sequence to uncover local patterns in event logs. As analysts partition the log into increasingly specific subsets, maintaining awareness of how subsets are constructed and how insights transfer across these subsets becomes challenging. The challenge comprises several interrelated aspects: difficulties in retracing filtering steps, relating included and excluded subsets across successive filtering steps, and interpreting subsets consistently with respect to different analytical perspectives. To address this challenge, we design and implement a visual artifact called FilterBranchMap, that makes the provenance of filtering steps explicit, preserves structured comparison across subsets, and integrates multiple analytical perspectives into a single representation. The FilterBranchMap is demonstrated on the Sepsis event log and evaluated qualitatively in a user study with process mining analysts. The results indicate that externalizing the provenance of filtering steps and subset relationships supports coherent sensemaking across successive refinements.