Towards new learning strategies in intelligent tutoring systems

Abstract

Co-operative tutoring systems replace the prescriptive approach developed by traditional intelligent tutoring systems with a constructive one based on the use of the computer as a way to exchange, control and build knowledge. This paper proposes two new learning strategies, learning by disturbing and learning by co-teaching, that extend the spectrum of possibilities in terms of co-operation and place the learner into a higher degree of abstraction. Learning by disturbing method allows to check the ability of the learner to distinguish between wrong and correct solutions. Learning by co-teaching provides an example of discussions between the teacher and the co-teacher that is useful for inducing correct solutions presented in a pedagogical form. Co-operation can be improved using elicitation techniques that can serve to extract learner's knowledge which can be further compared with the expert solution in order to identify his knowledge level. These techniques strengthen the efficiency of the learning strategies and serve as a basis for developing tutoring systems or learning environments including their co-operative aspects. We show how these strategies can be dynamically selected in an architecture of an intelligent tutoring system in which the knowledge level of the learner is frequently evaluated. We give an example of eliciting dialogues in a medical environment.keywordsIntelligent tutoring systemsElicitationTroublemakerCo-teacherLearning strategyCo-operative system