Khalil Mecheraoui
Object-Centric Process Mining (OCPM) often discovers models that are too dense and complex for effective human comprehension. To address this challenge, this work proposes a novel framework for the automated decomposition and aggregation of Object-Centric Petri Nets (OCPNs). The framework is based on the principles of high cohesion and low coupling. First, the Composite Cohesion Metric is introduced to measure the relational strength between object types from three distinct perspectives: behavioral synchronization, structural composition, and empirical log-based co-occurrence. By mapping these measures to a weighted graph and applying modularity-based community detection, the overall model is partitioned into semantically coherent sub-process modules. Finally, an Aggregated Net is constructed to offer a high-level architectural view of the system. The proposed framework is evaluated using the standard Order Management benchmark. The results demonstrate that the approach used successfully structures the process into distinct logical execution layers. The approach correctly assigns resource objects to their functional modules. The proposed framework significantly reduces the complexity of the model without affecting its global validity.