Mining User Behaviors for Providing Personalized Support to Learning from Interactive Simulations

The field of Intelligent Tutoring System has successfully delivered techniques and environments that  provide adaptive coaching and feedback  for problem solving in variety of  domains. There are, however, other educational activities that can help learners acquire the target

skills and abilities at different stages of learning including, among others, exploring interactive simulations.

Like for problem solving, learners can benefit from having individualized pedagogical support during these activities, especially as they are bound to become increasingly important with the advent of on-line courses and self-directed instruction. 

 

However, providing real-time personalize support for interactive simulations  rises unique challenges, because it requires modeling   and responding to student behaviors and skills often not as structured and well-defined as those involved in traditional problem solving.  We have developed a user modeling framework that mines student interaction data to perform behavior discovery and user classification suitable for providing real-time personalized support. In this talk. I will summarize results we obtained for modeling students interacting with two different types of simulations, including a  formal evaluation showing that a simulation that embeds adaptive support based on