Using Bayesian networks to better understand how users of educational software acquire skills

Joseph E. Beck

Research Scientist

WPI -Computer Science Department

Friday, February 27, 2009

11:00 a.m. – 12 p.m.

Fuller Labs, Perreault Hall

Abstract:

Educational data mining (EDM) is a new approach to answering questions about how learners acquire skills and best learn material. EDM is analogous to bioinformatics in that both fields rely on large statistical samples and computational techniques to perform research. For EDM, the recent wide availability of intelligent educational software, and its ability to log student interactions, provides a gold mine of information; what is missing is the toolkit to mine it. EDM draws on analytic techniques in computer science and statistics to answer interesting questions in cognitive psychology and education. This talk will provide an overview of the types of research questions and techniques for answering them that I find interesting. It will also provide an in-depth look in the use of Bayesian networks as a framework for analyzing complex interactions between students and educational software. Specifically, this talk will discuss how knowledge of the domain being taught can be used to instantiate Dirichlet priors, resulting in a better understanding of how students learn the domain. It is also possible to use Bayesian networks to answer questions such as whether, and (perhaps more importantly) how, the assistance provided by educational software helps students.

______

Joseph E. Beck is a Research Scientist at Worcester Polytechnic Institute. His work centers on using EDM, a field he founded in 2000 with a workshop series. After growing to having workshops at four conferences in 2007, we held the first International Conference on Educational Data Mining in 2008. Since then, we have launched the Journal of Educational Data Mining, where Beck is an associate editor. His research focuses on developing techniques to better model, and to consequently understand what impacts, student learning. He is primarily interested in fine-grained measures of learning, as these permit the most leverage for examining hypotheses of possible influences on learning.

Host: Neil Heffernan

Refreshments will be served.

Maintained by webmaster@cs.wpi.edu
Last modified:February 23, 2009