Learning environments based on inspectable student models
Jim E. Greer, Juan Diego Zapata Rivera · 2003
Inspectable student models have the potential to enhance computer-based learning environments. Researchers in the area have begun to explore the potential of externalizing the model to students and teachers in order to support reflection, knowledge awareness, and collaborative and negotiated assessment. A framework for learning environments based on inspectable student models has been proposed Human and computational components in the framework can interact with student models through a technological platform that includes a set of inspectable student modelling tools (i.e. VISMod, ConceptLab, and SMV), a Bayesian student modelling server (SModel server) and several guidance mechanisms (i.e. interacting with a human peer, interacting with an artificial guiding agent, or negotiating the student model with a teacher) that were used to guide student interaction with the student model. A learning environment based on this framework (The Learning Game - Java) has been implemented and used in a study focussed on exploring student model accuracy and student reflection. Results showed that using a Bayesian student model, neither the student view nor the system/teacher view was as accurate as the aggregate view of the model (higher correlation coefficients with a group of experts). In addition, it was found that there was evidence of different degrees of student reflection. Although the initial results are promising, there are still many challenges to face both in the lab and in real deployments. This research shows that inspectable student models have an important role to play in modern computer-based learning environments.