2016
DHSS 2016 Proceeding
Representing adaptive course navigation in the generalized intelligent framework for tutoring
Authors: R. A. Sottilare
Abstract
This paper explores the use of Markov Decision Processes (MDPs) in support of adaptive course navigation in the Generalized Intelligent Framework for Tutoring (GIFT). GIFT is an open source architecture for authoring and eval- uating Intelligent Tutoring Systems (ITSs) and adaptive course navigation is an AI-based technique which considers attributes of the learner and the instructional context to se- lect actions which will optimize learning. GIFT?s current adaptive course navigation model is decision tree-based. Other ITSs primarily use performance as a driver for navi- gation without consideration for other learner states. The adaptive course navigation model presented aligns closely with the principles of MDPs where a user?s current state, possible actions and a reward function determine movement to a future state. Unlike decision trees used which are cur- rently used in GIFT, MDPs also account for multiple states to determine future states and also consider uncertainty in the assessment of learner states.
I3M Scientific Sponsors
I3M Industrial Sponsors
I3M Media Sponsors
