Using AI planning to enhance e-learning processes

This work describes an approach that automatically extracts standard metadata information from e-learning contents, combines it with the student preferences/goals and creates PDDL planning domains+problems. These PDDL problems can be solved by current planners, although we motivate the use and benefits of case-based planning techniques, to obtain fully tailored learning routes that significantly enhance the learning process. During the execution of a given route, a monitoring phase is used to detect discrepancies, i.e. flaws that prevent the student from continuing with the original plan. In such a situation, an adaptation mechanism becomes necessary to fix the flaws, while also trying to minimise the differences between the original and the new route. We have integrated this approach on top of Moodle and experimented with 100 benchmark problems to evaluate the quality, scalability and viability of the system. Copyright © 2012, Association for the Advancement of Artificial Intelligence. All rights reserved.