Paul Compton's home pageI am am an emeritus professor in the School of Computer Science and Engineering (CSE) at The University of New South Wales. I am a researcher in CSE's Artificial Intelligence Research Group.
Most of my research for the last 20 years has been around the idea of building systems incrementally using a learning or knowledge acquisition stragegy known as Ripple-Down Rules.
Professor Paul Compton
School of Computer Science and Engineering
The University of New South Wales
Sydney 2052 Australia
ph 61 2 93856939
mob 61 425375279
fax 61 2 93854071
room 104a Computer Science (map ref k17)
studentscommercial RDR systems
Ripple-Down Rules summaryRipple-Down Rules (RDR) is a strategy of bulding systems incrementally while they are already in use. When a system does not deal with a case or situation correctly a change is made in such a way that the previous competence of the system is not degraded. The change is mades simply and rapidly and the difficulty of making a change should not increase as the system develops. RDR can be categorised as a type of apprentice learning
Various commercial RDR systems have been developed for a range of applications. There have have been research proofs for a wide range of applications including: various types of classification problem, configuration or parameter tuning, text-processing, image processing, heuristic search, tuning genetic algorithms and multi-agent environments. There have also been machine-learning versions of RDR.
A number of researchers at UNSW and elsewhere have been involved in this research. Current research is aimed extending the range of possible applications of RDR and further integration with machine learning so that a system knows when it cannot deal adequately with a problem and needs further training and can discuss the situation with its trainer.
The focus of this work has mainly been knowledge-based system. The approach seems even more essential when one considers the increasing significance of preference-based systems, either personal or business preferences with applications across web services. The ultimate goal is a general software engineering solution whereby all systems can be easily evolved as requiremente evolve and further requirements emerge.
|1. Compton, P. J. and R. Jansen (1990). A philosophical basis for knowledge acquisition. Knowledge Acquisition 2: 241-257. (This paper first appeared in EKAW 89 PDF)|
|This was the second RDR paper published. It argues from a situated cognition perspective that experts can never explain how they reach a conclusion, rather they justify that a conclusion is correct, and provide this justification in a particular context. Therefore all knowledge acquisition must be incremental and case-based.|
|2. Compton, P., L. Peters, et al. (2006). "Experience with Ripple-Down Rules." Knowledge-Based System Journal 19(5): 356-362. preprint PDF.|
|This paper documents the experience of a pathology laboratory using RDR to add clinical comments to patient biochemistry reports to assist GPs in patient management. The laboratory had built over 16,000 rules and processed over 6,000,00 patient reports. Knowledge acquisition was done by pathologists as a minor extension to their normal duties.|
|3. Compton, P., Cao, T., and Kerr, J. Generalising Incremental Knowledge Acquisition. in Proceedings of the Pacific Knowledge Acquisition Workshop 2004. Auckland: University of Tasmania Eprints repository, p. 44-53, 2004. PDF|
|There are a number of versions of RDR for different types of applications. There are also a number of generalisations of RDR by various authors to apply to a range of problems. This paper is the most recent attempt at generalising RDR|
|4. Cao, T.M. and Compton, P. A Simulation Framework for Knowledge Acquisition Evaluation. in Twenty-Eighth Australasian Computer Science Conference (ACSC2005). Newcastle, p. 353-360, 2005 (this is the version originally submitted PDF)|
|A simulation technique has been developed as a way of evaluating knowledge acquisition. The relevance of this paper as an introductory paper is that it details a few different RDR algorithms.|
|5 Ho, V., W. Wobcke, et al. (2003). EMMA: An E-mail Management Assistant. IEEE/WIC International Conference on Intelligent Agent Technology, Los Alamitos, CA, IEEE pp 67-74. preprint PDF|
|This is an example of an RDR for text classification. It is currently under commercial evaluation|
|6 Prayote, A. and P. Compton (2006). Detecting Anomalies and Intruders. AI 2006: Advances In Artificial Intelligence, 19th Australia Joint Conference on Artifical Intelligence, Hobart, Australia, Springer. pp1084-1088. preprint PDF|
|This is potentially the type of technique that could be used with RDR, or perhaps other types of knowledge-based system so that the system knows when a case is outside its range of experience and expert input is required.|