SimStudent: Improving Tutor Quality and Reducing Authoring Costs
Workshop Proceedings of the 12th International Conference on Intelligent Tutoring Systems
2014
Abstract
Intelligent Tutoring Systems are effective at improving learning, but are costly to produce. In this paper, we review SimStudent, a system designed to aid authors in improving the quality of their tutors while simultaneously reducing authoring costs. This system works by inducing general production rule models from user demonstrations and feedback. We discuss results demonstrating that SimStudent learned models better fit student data than expert authored models and that it is efficient to use. In addition, we provide an example of how a simple algebra tutor would be constructed using this paradigm. Lastly, we review future directions for this work.
BibTeX
@inproceedings{maclellan-its-2014-ws,
title = {SimStudent: Improving Tutor Quality and Reducing Authoring Costs},
author = {MacLellan, Christopher J. and Stampfer Wiese, Eliane and Matsuda, Noboru and Koedinger, Kenneth R.},
booktitle = {Workshop Proceedings of the 12th International Conference on Intelligent Tutoring Systems},
year = {2014},
}