The WHY in Business Processes: Unification of Causal Process Models
Yuval David, Fabiana Fournier, Lior Limonad, Inna Skarbovsky
TL;DR
This work tackles unifying causal process variants in causal business process (CBP) models while preserving variant-specific causal dependencies and exposing alternating causal flows. It introduces a gating-based framework that yields unified extended CX graphs (U-CX) by partitioning logs, performing LiNGAM-based causal discovery on partitions, and then unifying results through a matrix-driven gateway identification process, augmented with a noise-thresholding mechanism. The authors formalize soundness and completeness, provide proofs in appendices, and demonstrate scalability across five real-world datasets, with an open-source implementation. The approach enables robust prescriptive process analytics by explicitly representing when different variants trigger different downstream activities, while mitigating missing-data biases through partitioning.
Abstract
Causal reasoning is essential for business process interventions and improvement, requiring a clear understanding of causal relationships among activity execution times in an event log. Recent work introduced a method for discovering causal process models but lacked the ability to capture alternating causal conditions across multiple variants. This raises the challenges of handling missing values and expressing the alternating conditions among log splits when blending traces with varying activities. We propose a novel method to unify multiple causal process variants into a consistent model that preserves the correctness of the original causal models, while explicitly representing their causal-flow alternations. The method is formally defined, proved, evaluated on three open and two proprietary datasets, and released as an open-source implementation.