Using Big Data and BKT to Evaluate Course Resources (Abstract Only)

Now finding footing in objective research methodology, MOOCs have made significant strides toward developing into mature platforms for not only offering educational materials but also performing exploratory analysis of educational methods. Bayesian Knowledge Tracing (BKT) has been repeatedly shown to be successful at providing an accurate model of student knowledge in more traditional classroom settings, and recent research has explored the application of BKT to MOOCs with promising results. Using data from several MOOCs run by Stanford university, we propose to extend earlier research into the application of BKT to MOOCS by developing a framework within which the use of course resources and student performance can be leveraged both to increase the predictive accuracy of BKT modeling and to provide an evaluative metric for the utility of those resources. We hope that such a framework can contribute not only to MOOC courses, but traditional classrooms as well.