General Principles

Abstract

In this chapter, I will identify and describe general principles that apply to course generation as formalized in Paigos, independent of both the underlying learning theory and the learning goals the course is generated for.The chapter starts with metadata. In Section 4.1, I show that existing learning object metadata standards fail to describe educational resources sufficiently precise for their automatic integration into learning processes by intelligent components. I then describe an ontology of instructional objects that contains specifically this previously missing information. The ontology facilitates the process of making repositories available to the course generator. On the one hand, this process helps to assemble a course from resources of different repositories. On the other hand, the course generator can provide its functionalities as a service to other systems that plug-in their repositories. The mediator architecture that was developed for this purpose is described in Section 4.2. A further question concerns the learning goals that Paigos processes. In traditional course generation systems, learning goals consist only of educational resources, which represent the target content that is to be learned. Such an approach ignores that different purposes require different course of actions. For instance, a course for preparing an exam should consist of different educational resources from a course that provides a guided tour. Section 4.3. tackles this question and provides a general representation of learning goals. The subsequent sections form the main part of this chapter. They explain how course generation knowledge can be formalized as an HTN planning domain. This chapter describes general axioms, operators and methods used in that domain (Chapter 5 describes how these general techniques are used to generate course for different learning goals in different learning theories). The remaining sections of this chapter cover additional features that arise from the need of the application and are possible in Paigos: Section 4.8 describes how to generate a complete course and still allow for adaptive selection of resources when needed. Section 4.9 focuses on the problem that automatically generated sequences of educational resources can lack coherence and explanations about the learning goals, and describes how to use the information about the learning goals available during course generation to generate such information.KeywordsEducational ResourceGeneration DomainLearning GoalPedagogical TaskWorld StateThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.