Flexible ILE Development: Case Studies of an Agent-Based Software Engineering Approach

Abstract

The problem of the increasing cost of constructing Intelligent Learning Environments (ILEs) has been highlighted by a number of researchers. Agent-based software engineering (Genesereth and Ketchpel 1994) has been proposed as the starting point for a methodology for constructing agent-based ILEs. We have developed a methodology, called Flexible ILEs via Agent Systems (FILEAS) which seeks to meet the requirements of researchers by supporting incomplete specification, changes to specification, removal of components, incorporation of existing software and extendibility of component use. A case study was conducted with good ecological validity designed to investigate the strengths and weaknesses of the FIILEAS methodology. The context adopted was the development of an ILE to investigate issues in learning by analogy, and the focus was upon building individual agents. Studies of the kind that have been presented here are rarely made but are needed both to support the research community and to help identify areas for further development of methodologies and tools. The results indicated that the FILEAS methodology would provide much of the flexibility required by ILE researchers. The process of incorporating software was straightforward. Those changes that were required were in the main not too expensive to make. Further work is outlined to carry out an extended case study to examine the methodology.