Intervening vs. Non-Intervening Actions for INQPRO's Dynamic Learner Model

C.-Y. Ting and S. Phon-Amnuaisuk (Malaysia)

Keywords

Learner model, Intervening and nonIntervening, Dynamic Decision Network

Abstract

Providing pedagogical interventions under uncertainty has always been a challenge for decision-theoretic Intelligent Tutoring Systems (ITSs). Studies have shown that apart from the challenges in structuring the domain problem, the issue of how and when to intervene contributes to the usefulness and overall performance of a decision-theoretic ITS. In this study, two Dynamic Decision Network (DDN) models were proposed: "intervening" and "non intervening". Field tests, which involved 31 learners and 6 domain experts, were conducted to identify the models' classification accuracies. The empirical results showed that the "non-intervening" outperformed the "intervening", implying that the "non-intervening" approach for designing pedagogical interventions is more appropriate for INQPRO, a scientific inquiry learning environment developed within this research work.

Important Links:



Go Back