TRACE4PM: Trace Related Analysis and ClustEring for Process Modeling of Users’ Interactions in Information Systems
Résumé
The rapid growth in information systems usage has highlighted the importance of modeling user behavior to improve usability, deliver personalized recommendations, and refine system design. User interactions with these systems generate rich datasets of event data crucial for such analyses. Process mining plays a pivotal role in modeling these behaviors, often complemented by trace clustering techniques that help manage the complexity of user interactions. In this context, we introduce TRACE4PM, an automated pipeline encapsulated within an API explicitly designed to model user behaviors in information systems using process mining techniques focusing on trace clustering techniques. The platform offers services, including parsing, tagging, clustering, and process modeling of the event data, all based on state-of-the-art Process mining techniques. This API-based solution promotes ease of integration and usage, enabling developers and researchers to use the API platform without requiring them to delve into technical aspects or requirements. To demonstrate the real-world applicability of TRACE4PM, we developed a Moodle plugin that collects and analyzes Moodle logs to cluster and model student interaction traces, successfully extracting helpful learning models from the educational data. Furthermore, researchers and educators feedback validated TRACE4PM’s and the Moodle plugin’s efficiency, applicability, and educational utility, highlighting their potential for broader adoption and refinement.