To realize the promise of fielding an intelligent tutoring system (ITS), the ITS requires a knowledge base from which to draw instructional source content. That knowledge must be first acquired and then represented in a tractable form; useful from both a computing standpoint and the point of view of presenting that knowledge to a student.

In general, intelligent tutoring systems research to date has focused on the student and on methods for representing the student knowledge. From student models to learning schemas to presentation methods, comparatively little attention has been paid to the problem of educators attempting to build viable curriculum plans for use within an ITS environment. What is needed, before attempting to design and develop an ITS, is a methodology for defining and developing student curricula in a form directly related to ITS implementation. This methodology should be quantifiable both in terms of content and applicability, and able to accept feedback metrics on a given student's progress to modify the lessons and the curriculum plan. In addition, the tool should be useful without requiring excessive training.

In the literature, one finds numerous examples of knowledge representation schemes, from the idea of concept mapping to the hierarchical databases used in the Air Force's Instructional System Development (ISD) project. Even when projects provide an automated tool capability, educators face steep learning curves, a wide array of user interfaces, and a significant amount of manual development when constructing student curricula.

Our approach employs an automated knowledge acquisition tool, PESKI, to acquire the necessary information for student curriculum generation, utilizes concept mapping to represent that knowledge, and then maps that representation into concept vectors. We developed a prototype system, based on concept vectors, that accepts inputs from an educator via a world wide web (WWW) interface and returns a dynamic lesson plan.