Behavioral web tracking in e-learning: an educational process mining application

This paper introduces an experiment and the first results of a research on computer programming education using process mining methods. A web-based tutorial addresses the topic of agent-based modeling by introducing a guided exercise with NetLogo, a widely used tool for modeling natural and social phenomena. A goal of the project is to analyze the goodness of the learning process, also through appropriate tests placed between the pages and at the end of the tutorial. Actual data extracted on student behavior (e.g., length of time spent on different parts of each web page, movements on the page, mouse position and mouse clicks) are examined using process discovery technique. Special attention is given to the return of student learning outcomes through visualization. Our solution includes heatmaps and direct-follow graphs of the real processes. Initial results are encouraging on the possibility of improving the assessment of learning processes by relying on techniques from the discipline of process mining, as shown by the case of web-based behaviour tracking data.