Cognitive Diagnosis in ITS Using General Pedagogical Rules

Abstract

This paper presents an approach to cognitive diagnosis in a large-scale ITS. The key ideas are that knowledge is learned only in contexts, i.e. in relation with other knowledge, and that to understand new knowledge the learner has to generate links to old knowledge; if links cannot be generated, impasse occurs. We first describe the learning context in a session, and then summarize some theories about knowledge categorization, knowledge formation and reasoning. Based on these theories, four potential causes of theoretical learning impasses are identified, which are: incoherence of incoming knowledge, insufficient knowledge, misinterpretation of the learning context, and weakness in reasoning. According to the state of the learner’s impasse (i.e. the first time, the second time, etc.), we do one of the two kinds of diagnosis as follow: (1) In the shallow diagnosis, the underlying capability category (which results from the content analysis) and the existing links with other capabilities (such as: deviation, generalization, aggregation, etc.) are first verified, and then appropriate didactic resources will be used to help the learner, (2) In the deep diagnosis, the learner’s cognitive profile is analyzed to find out the prerequisite capabilities which are weakly acquired or acquired a long time ago; these capabilities are then added to the current plan or a recall resource is used respectively. In the weakness reasoning cases (i.e. prerequisite capabilities are well acquired), reasoning support resources will be used. These resources are created based on the fundamental logical reasoning modes that correspond to deduction, induction and abduction. In our ITS, general pedagogical rules are formulated according to the expertise mentioned above.