Partially Observable Sequential Decision Making for Problem Selection in an Intelligent Tutoring System

A key part of effective teaching is adaptively selecting pedagogical activities to maximize long term student learning. In this poster we report on ongoing work to both develop a tutoring strategy that leverages insights from the partially observable Markov decision process (POMDP) framework to improve problem selection relative to state-of-the-art intelligent tutoring systems, and evaluate the computed strategy in the classroom. We also highlight some of the challenges in data mining related to automatically constructing pedagogical strategies.