TITLE> International Conference on Intelligent Agents Web Technologies and Internet Commerce

Building a knowledge system while it is in use
Professor Paul Compton
School of Computer Science and Engineering
The University of New South Wales
Sydney 2052 Australia
http://www.cse.unsw.edu.au/~compton/

Abstract
We have an increasing range of tools and techniques at our disposal that enable us to build knowledge systems of many types and for many different applications. However, knowledge shifts, changes and only gradually becomes clear and it can be difficult to change the system we have built. This paper describes research where the main focus is on making it as simple as possible to change a knowledge system so that a domain expert or other user, can change and evolve the system while it is in use as a minor part of their routine duties. The family of techniques that enable this incremental development are known as Ripple-Down Rules (RDR). There are a number of commercial RDR versions, and RDR systems with thousands of rules, built by domain experts, are in routine use. Current research includes extending the range of tasks that can be addressed (e.g. image processing, multi-agent and service composition applications) and integrating RDR with machine learning to provide generalisation and ontology development. Such techniques are likely to be of increasing importance with the growing use of knowledge technologies for handling personal and business preferences as well as expertise.

Biographical Statement Paul Compton is professor and head of the School of Computer Science and Engineering at the University of New South Wales. He originally trained in philosophy and physics and then worked in medical research for 20 years before moving to the School of Computer Science and Engineering in 1990. He worked on GARVAN-ES1, one of the first medical expert systems to go into routine clinical use and has been one of the few researchers to investigate the on-going maintenance problems of such systems. This led to his research focus on how systems can gradually be evolved over time, built by end-users.

Last Modified: 22/11/2006