Collaborative Model Optimization in Federated Stochastic Process Discovery

Abstract

Process discovery focuses on constructing process models that accurately represent system behavior from event logs generated by these systems. Traditional methods rely on centralized event logs, which pose significant challenges in distributed environments where organizations are reluctant to share raw data due to privacy concerns. Federated process discovery has emerged as a promising approach to enabling collaborative model generation without compromising data privacy. However, existing federated approaches often produce overly complex process models, limiting their practical utility. This paper introduces an approach that enables organizations to collaboratively construct, via iterative r..