Automated course configuration based on automated planning: framework and first experiments

The work presents an application of automated planning to the process of course configuration. Configuring a course means to build a course based on the effective needs of the specific learner. The course is built by selecting and sequencing learning objects (learning components) from a repository. Parameters in that process are the starting knowledge that the learner already possess and the target knowledge that is expected to be gained through the course. In order to have more powerful ways to reason on its properties and structure we propose to treat this problem as a planning problem: a configuration is a plan, a component is an action in the plan; required/acquired knowledge are modeled, resp., as action preconditions and postconditions. The initial state of the plan encodes the student’s starting knowledge, while the goal of the plan encodes the course target knowledge. We have run a short experiment on a class of 26 third-year undergraduate students, regularly attending the course of Algorithms and Data Structures.