Swevels, A., Zerbato, F., & Fahland, D. (2026). Process Mining in Context: Extending Domain Data Models for Iterative Analysis. In L. Fuentes, P. Plebani, C. Combi, & H. Reijers (Eds.), Advanced Information Systems Engineering: 38th International Conference, CAiSE 2026, Verona, Italy, June 8–12, 2026, Proceedings (Vol. I, pp. 195-212). (Lecture Notes in Computer Science (LNCS); Vol. 16558). Springer. https://doi.org/10.1007/978-3-032-28110-4_11
Abstract
Process mining (PM) seeks to enhance processes in various domains—such as in healthcare or logistics—through iterative, data-driven analysis. However, the data in these domains is usually not modeled in terms of events, cases, and objects, which are essential generic concepts for applying PM techniques. Therefore, existing approaches require mapping domain-specific concepts (e.g., patient) into generic ones (e.g., case) during data preparation. This abstraction disrupts the iterative nature of any PM analysis: analysts, who need to work with cases and events, must repeatedly mentally map PM-specific results to their original domain context. The issue remains even when using Object-Centric Event Data (OCED). To address this problem, we propose an approach to perform iterative analysis within the original domain context. The approach extends a domain-specific data model with types that allow the generalization of domain-specific concepts into abstract PM concepts. With this approach, analysts can apply generic PM techniques directly over domain-specific data and enrich the data with PM results, without the need to map back and forth. We evaluate our approach through a PM analysis on a benchmark dataset, demonstrating how it leads to insights that are relevant from a domain perspective.