Persona Mining Using Process Mining and Unsupervised Machine-Learning

In the era of digitalization, users of information systems and digital platforms often face challenges such as non-intuitive design, incomplete or redundant data, and difficulties in accessing relevant information, which can result in a suboptimal User eXperience (UX). Our research aims to enhance the experience of users of digital technologies through personalized recommendations based on the users’ data, their intentions, and the context. The current step of our research focuses on Persona Mining as a means to define profiles of users’ types (called personas) based on traces left by users in various information systems. The existing research works do not allow identifying personas based on activity logs. Our proposal combines Process Mining and Unsupervised Machine Learning techniques to analyze event logs, uncovering user behavior patterns and defining distinct Personas based on real user activity. We evaluate the proposed approach through three case studies. By identifying diverse Personas within user traces across applications and platforms, the research contributes to improving recommendation systems to provide better User eXperience.