Recent Developments and Summary
The "Traditional Trinity" has formed the basis for a number of intelligent tutoring systems. However, Self (1988) suggests that recent trends in intelligent tutoring research do not map well to it.
These trends have seen tutoring strategies moving away from fact based tutoring towards the tutoring of problem solving strategies and the analysis of these strategies. This meta-level knowledge is less easy to represent within the domain knowledge base.
There has also been a move towards multiple representations of the domain knowledge in order to cater for specific situations and contexts. It has been recognised that the expert and learner work in different ways. The student will go through several phases while attaining expert problem solving skills. Alternative domain representations may be required as these phases are established and progressed through. Multiple domain representation do not map well to the overlay and bug rule based student models.
The emergence of alternative learning environments such as those that allow for experimentation and simulation and an increased range of interaction are not well catered for either by the "Traditional Trinity".
One interaction style that has recently been gaining in popularity is hypertext. There are significant advantage associated with hypertext and a number of tutoring systems have been devised that incorporate a hypertext component. A number of these will be explored in the next chapter.
Machine learning techniques have also been used in a number of recent intelligent tutoring systems. A number of these systems along with the various machine learning techniques they incorporate are discussed in chapter 3.
This chapter introduced the three module model for intelligent tutoring systems. Each of the modules was described and a number of intelligent tutoring systems described.
Although few of these systems have made it into the classroom a number of important lessons have been learned. Perhaps the most significant of these is that knowledge organisation is itself as important as knowledge itself. The reasoning processes in most intelligent tutoring systems do not match with the reasoning of a human expert in real life situations, this is particularly the case with expert system based tutoring systems such as GUIDON. In order to be effective a tutoring system should work in a human-like way (Anderson1988).
Authored by Serengul Smith
E-mail to:
serengul1@mdx.ac.uk
School of Computing Science Middlesex University
Revised: September 1998