Simulated Students, Mastery Learning, and Improved Learning Curves for Real-World Cognitive Tutors.
Stephen E. Fancsali, Tristan M. Nixon, Annalies Vuong, STEVE RITTER · 2013
Abstract. We briefly describe three approaches to simulating students to develop and improve intelligent tutoring systems. We review recent work with simulated student data based on simple probabilistic models that provides important insight into practical decisions made in the deployment of Cognitive Tutor software, focusing specifically on aspects of mastery learning in Bayesian Knowledge Tracing and learning curve analysis to improve cognitive (skill) models. We provide a new simulation approach that builds on earlier efforts to better visualize aggregate learning curves.