A plan recognition process, based on a task model, for detecting learner's erroneous actions

Abstract

When a tutoring system aims to provide learners with accurate and appropriate help and assistance, it needs to know what goals the learner is currently trying to achieve, what plans he is implementing and what errors he is making. That is, it must do both plan recognition and error detection. In this paper, we propose a generic framework which supports two main issues (i) the detection of learner's unexpected behavior by using the Hollnagel classification of erroneous actions and (ii) a recognition process based on a task model METISSE that we propose. This model, which is used to describe the tasks the learner has to do according to pedagogical goals, allows learner's unexpected behavior to be detected. The solutions proposed are generic because not dependent on the domain task, and they do not relate to a particular device. © Springer-Verlag Berlin Heidelberg 2006.