Privacy-Preserving Process Data Generation Based on Dual-Discriminator Conditional Generative Adversarial Networks

Yi Guo¹Zhong Li^2*

• ¹Tongji University, Shanghai, China
• ²Donghua University, Shanghai, China

The expanding adoption of data-centric business analytics demands effective safeguarding techniques for process data that contains procedural details. Although Petri net-driven process mining successfully derives operational knowledge from activity sequences, current protection approaches frequently reduce analytical value. Therefore, maintaining process-related information while ensuring privacy protection remains a critical challenge. This work presents a Privacy-Preserving Process Data Generation method based on Dual-Discriminator Conditional Generative Adversarial Networks (P3DGAN) to generate privacy-preserving process data. To avoid mode collapse during model training, P3DGAN employs two discriminators that separately model the dataflow and workflow characteristics of process data. Furthermore, we propose a game optimization strategy based on Petri net theory to capture the global distribution characteristics of process data. Additionally, we introduce a workflow-level privacy metric based on the Euclidean distance of trace variants (ED-TV) for comprehensive risk assessment. Experimental results on four real process datasets demonstrate that our method can generate high-quality process data with excellent privacy protection compared to its competitive peers.