Externalising Learner Modelling Representations

Research on inspectable learner models involves the study of representational aspects of the profile that systems maintain about learners. When learners and teachers interact with inspectable learner models several questions arise, such as: What kind of representation is most appropriate for learners and teachers?; How will students and teachers react to different representations?; How should learners and teachers interact with the model?; In order to answer these and other questions, we have developed ViSMod, an interactive visualization tool for Bayesian learner models and ConceptLab, a knowledge construction and navigation system that uses XML-based conceptual maps to represent the learner’s view of the domain. INTRODUCTION In order to provide individualised instruction, educational systems maintain a representation of the learner (learner model). By making the learner model inspectable to students and/or teachers, it is possible to promote reflection, interactive assessment, and collaborative creation of learner models. These provide the teacher with a powerful assessment tool. In order to determine how learners will react to the learner model, it is necessary to consider representational, and visualization issues. Several authors have explored the use of different media in presenting the learner model to students and teachers [Bull & Shurville 1999, Morales et al. 1999, Dimitrova et al. 1999, Müehlenbrock et al. 1998, Paiva et al. 1995]. How to present the model to the students and teachers basically depends on three factors: the domain target, the students’ and teachers’ preferences, and the goals of the learning environment. Supporting reflection can be done by simply exposing learners to a simple textual description of the content of the model, to a 2D or 3D graphical model or even a virtual space in which students can interact with abstract representations of their cognitive states. Should this reflection process be an individual activity, a group based activity, or a guided activity (by a teacher or an artificial agent)? In order to explore these questions, we have developed ViSMod [Zapata-Rivera & Greer 2000], an interactive visualization tool for Bayesian learner models and ConceptLab [Zapata-Rivera et al. 2000], a knowledge construction and navigation system that uses XML-based conceptual maps to represent the learner’s view of the domain.