Designing Intelligent Systems for Teaching Visual Concepts

In IJAIED 8 (1)

Publication information

Abstract

The paper describes how high-level knowledge about visual images could be represented and further interpreted through system-active and systempassive tutorial interactions. The ideas lend themselves to the design and implementation of a category of Intelligent Tutoring Systems aimed at the teaching of abnormalities in highly visual domains, like medical diagnostic imaging. The involved problems are treated through (1) a domainindependent, object-oriented method for managing the complexity of design representations and (2) a model of dialogue interpretation for implementing tutorial interactions. The method and the model are both supported here by implemented, computer-based tools that integrate the multi-layer environment RUI. To evaluate the power of RUI, empirical observations have been carried out, focussing on the generality of the design methodology as well as on the usability of the interface. We draw conclusions about the suitability of object-oriented principles in the context of ITSs evolution.