Layouting Object-Centric Directly Follows Graphs

Deoksang Lee, Minseok Song, Wil M. P. van der Aalst

Abstract: Process mining extracts insights from event logs to analyze and optimize business processes. Object-Centric Process Mining (OCPM) extends traditional process mining by utilizing Object-Centric Event Logs (OCEL), explicitly capturing interactions among multiple objects within business processes. Despite its advantages, visualizing object-centric process models remains challenging due to increased complexity, leading to denser graphs with more nodes and edges compared to traditional process mining. To address this challenge, this study proposes an advanced process layout generation method designed to improve the visualization clarity of Object-Centric Directly Follows Graph (OC-DFG), a widely used representation of object-centric process models. Our approach systematically assigns distinct axes to each object type, enhancing readability by clearly separating interactions. Additionally, the method incorporates edge cross-minimization and spatial optimization techniques to further improve layout efficiency. We validate the proposed method quantitatively using both traditional layout evaluation metrics and newly developed metrics specifically designed for OCPM. Experimental results indicate notable improvements in layout metrics associated with readability for OC-DFG. This work contributes to process mining, modeling, and analytics by offering a structured visualization approach for complex object-centric process models, which may support stakeholders in analyzing processes and making informed decisions.

External IDs:dblp:conf/bpm/Lee0A25