Process Mining is a set of techniques that aim at discovering, monitoring and improving real processes by using logs of events created and stored by corporate information systems. The growing use of information and communication technologies and the imminent wide deployment of the Internet of Things enable the massive collection of events, which are going to be studied so as to improve all kinds of systems efficiency. Despite its enormous benefits, analyzing event logs might endanger individuals privacy, especially when those logs contain personal and confidential information, such as healthcare data. This article contributes to an emerging research direction within the process mining field, known as Privacy-Preserving Process Mining (PPPM), which embraces the privacy-by-design principle when conducting process mining analyses. We show that current solutions based on pseudonyms and encryption are vulnerable to attacks based on the analysis of the distribution of events combined with well-known location-oriented attacks such as the restricted space identification and the object identification attacks. With the aim to counteract these attacks, we present u-PPPM, a novel privacy-preserving process mining technique based on the uniformization of events distributions. This approach protects the privacy of the individuals appearing in event logs while minimizing the information loss during process discovery analyses. Experimental results, conducted using six real-life event logs, demonstrate the feasibility of our approach in real settings.
Authors: Edgar Batista; Agusti Solanas.
Journal: Peer-to-Peer Networking and Applications.
Date of Publication: 16 January, 2021.