An Assessment of Encryption-based Approaches for Business Process Models Discovery

Abstract

Process Mining as a Service (PMaaS) could enable organizations to delegate the analysis of their business processes to third parties in the cloud and then avoid costs related to storage and processing. However, this involves outsourcing event log data to the service provider, which can be honest but curious and learn from the data in clear. This potentially compromises the confidentiality and privacy of the event log data that can be sensitive, for example, in the healthcare domain. Thus, it is necessary to both safeguard the privacy of the event log data and guarantee their usefulness in the process mining tasks. This paper presents a study and experimental evaluation of two cryptographic-based methods suitable for PMaaS with security and privacy guaranties, tested under the process discovery task and evaluated through a set of experiments in terms of utility loss, performance and computational cost, using available event logs in the healthcare domain.