"Qualitative simulation and knowledge representation for intelligent tutoring"

Abstract

This paper discusses the strong points as well as some consequences of a newly emerging view on (computer assisted) learning (30). In this view learning should be seen as acquiring mental models by students. The models that are to be acquired should be the same models that drive the simulations used in computer assisted learning. In order to facilitate the process of acquiring the target model intermediate models should be implemented as well, in particular causal qualitative models. These should be the first models to be acquired by the student. This view forms a natural answer on criticisms with respect to the use of computer simulation in education and at the same time enables a better articulation of an existing paradigm for intelligent tutoring. The important implication is that the student should acquire at least certain qualitative causal models. A study of the literature on causality however shows that causality has been a much debated concept for ages and that different forms of qualitative reasoning in artificial intelligence are connected with different causality concepts. Thus the "guess the model in the computer" view on computer assisted learning leads to very fundamental philosophical questions, because in order to implement a qualitative causal model we have to define very precisely what we mean by causal reasoning. These same questions are encountered in expert systems research where the QSIM approach (16, 17) was developed for representation of "deep models". Therefore this paper discusses those aspects of the QSIM approach that are relevant in the context of intelligent tutoring and representation of deep models.KeywordsExpert SystemGeneric KnowledgeKnowledge RepresentationInference EngineCausal ReasoningThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.