Using drawings in knowledge modeling and simulation for science teaching

Abstract

Modeling knowledge in simulation-based inquiry learning requires a model of the domain that is executable, as well as a model of the learners' knowledge about the domain. An intermediate level is formed by models of the domain that are created by students as is done in modeling environments. An approach is presented for generating student created models from drawings. This approach requires drawing segmentation, shape recognition and model generation, which is done based on density-based clustering, elementary shape recognition combined with a shape ontology and model fragment composition respectively. The final result is an executable model that can be used to generate simulation outcomes based on learners' conceptions. The role of such a system is discussed, especially with respect to the diagnosis of misconceptions and the generation of tutoring interventions based on confronting learners with the consequences of their conceptions. © 2010 Springer-Verlag Berlin Heidelberg.