Two-Phase Updating of Student Models Based on Dynamic Belief Networks

When a belief network is used to represent a student model, we must have a theoretically-sound way to update this model. In ordinary belief networks, it is assumed that the properties of the external world, modelled by the network, do not change as we go about gathering evidence related to those properties. I present a general approach as to how student model updates should be made, based on the concept of a dynamic belief network, and then show this work relates to previous research in this area.