Experimentation and results for calibrating automatic diagnosis belief linked to problem solving modalities: A case study in electricity

Abstract

Learners increasingly work with virtual laboratories that provide various activities and tools, including sophisticated modeling and simulation systems. The learning environments have to combine traces to establish the most precise diagnosis possible on the learner's activity. This paper presents a diagnosis tool, called DiagElec, establishing a diagnosis on the learner's activity. DiagElec integrates a notion of belief, which is related to the modalities in the generated diagnoses. To analyze our model, we have carried out a two-phase experiment, first with learners and then with teachers. From the corpus of diagnosis done by the teachers, we are looking for the emergence of a model of human behavior to recalibrate the degree of belief defined into the diagnosis rules. © 2010 Springer-Verlag Berlin Heidelberg.