AuthorIT: Breakthrough in Authoring Adaptive and Configurable Tutoring Systems?
Joseph M. Scandura
This article is Part I in a two part series introducing AuthorIT, an integrated, extensible authoring system for building intelligent (adaptive & configurable) tutoring systems (ITS). AuthorIT is based on an empirically tested theoretical foundation that is both deep and broad; it automates many processes in building ITS, including an elicitation tool, called AutoBuilder, for representing knowledge as Abstract Syntax Trees (ASTs). ASTs serve as input to AuthorIT’s companion, TutorIT. TutorIT delivers learning as prescribed—based exclusively on AST structure, interface and configuration options without reference to content semantics. Section 1 presents an overview of AuthorIT’s major components: AutoBuilder, Blackboard Editor, Configuration Tool and TutorIT, illustrating their use with examples. AuthorIT has been used to develop prototype ITS involving procedural, declarative (structural) and modelbased knowledge in arithmetic and basic mathematical processes. Programmers easily convert AST-based knowledge representations to executables. EZauthor and TutorIT Customizer, respectively, are limited teacher friendly subsets that support development and customization with no programming required. Section 2 describes the underlying theory. It details the close relationships between structural and procedural knowledge and shows how each can be represented at arbitrary levels of detail in terms of a small finite number of refinement types. The section also describes the role learning objects, display, response and evaluation types play in ASTs and the Blackboard Editor, and the diagnostic and instructional logic enabling TutorIT to accommodate any content represented as ASTs. Section 3 summarizes accomplishments to date, limitations of current work and planned extensions. Examples illustrate how ASTs in conjunction with AuthorIT have been used to develop tutorials not only for procedural knowledge, but also for declarative (structural) and model-based knowledge. Existing technology also is sufficient for supporting welldefined problem solving. Finally, current limitations are identified along with possible extensions showing how they may be addressed. Part II in this series will detail theory supporting planned extension of AuthorIT to ill-defined problem solving, with emphasis on higher order knowledge, the role of structural analysis in identifying higher order knowledge and a universal control mechanism.