Ripple Down Rules (RDR) is a knowledge acquisition method which constrains the interactions between the expert and a shell to acquire only correct knowledge. Although RDR works well, it is only suitable for the problem of providing a single classification for a set of data. Multiple Classification Ripple Down Rules (MCRDR) is an extension of RDR which allows multiple independent classifications. The approach has been evaluated in simulation studies where the human expert is replaced by a simulated expert. MCRDR may provide a basis for building a general problem solver for a range of problems beyond classification.

Evaluation of knowledge acquisition methods remains an important goal; however, evaluation of actual knowledge acquisition is difficult because of the unavailability of experts for adequately controlled studies. This paper proposes the use of simulated experts, i.e., other knowledge based systems as sources of expertise in assessing knowledge acquisition tools. A simulated expert is not as creative or wise as a human expert, but it readily allows for controlled experiments. This method has been used to assess a knowledge acquisition methodology, Ripple Down Rules at various levels of expertise and shows that redundancy is not a major problem with RDR. Introduction Evaluation of knowledge acquisition (KA) methods remains an important goal. Many KA methods have been proposed and many tools have been developed. However, the critical issue for any developer of knowledge based systems (KBS) is to select the best KA technique for the task in hand. This means that papers describing m...

Situated cognition is not a mere philosophical concern: it has pragmatic implications for current practice in knowledge acquisition. Tools must move from being design-focused to being maintenance-focused. Reuse-based approaches (e.g. using problem solving methods) will fail unless the reused descriptions can be extensively modified to suit the new situation. Knowledge engineers must model not only descriptions of expert knowledge, but also the environment in which a knowledge base will perform. Descriptions of knowledge must be constantly re-evaluated. This re-evaluation process has implications for assessing representations 1.

A knowledge acquisition methodology, Ripple Down Rules, has been developed which only allows knowledge to be used in the context in which it is acquired and ensures that only valid rules can be added to a knowledge base. This method has now been used to build a large (2000 rule) medical expert system. This system is in routine use in a Chemical Pathology laboratory providing clinical interpretations for laboratory reports. It has been developed entirely by experts with no knowledge acquisition or programming support or skills. This task was a minor extension to their normal duties. This paper describes the resultant knowledge base and concludes that such knowledge acquisition is very simple. It claims that allowing knowledge to be easily refined is a powerful technique which greatly simplifies dealing with complex domains. INTRODUCTION TO RIPPLE DOWN RULES Ripple down rules (RDR) is a knowledge acquisition methodology and a way of structuring knowledge bases which grew o...

Currently, "task analysis" is the dominant paradigm in the knowledge acquisition community. We argue that for performance systems (i.e. systems that do not have to offer a knowledge-level description of their performance at runtime) a simpler "test analysis" approach may suffice. We offer examples were a seemingly-naive testing regime gives rise to competent performance systems. Further, by certain measures, these systems developed via test analysis out-performed systems developed for similar domains using other techniques. Test analysis did not augment some other methodological approach: it removed the need for any other methodology. We speculate that for performance systems, task analysis could be deferred till after the development of a tested performance system. That is, for performance systems, testing replaced task but task analysis could augment test analysis once a system was in production. 1. INTRODUCTION Task analysis evolved from a reverse engineering of existing expert sy...

Situated cognition (SC) and knowledge maintenance are intimately connected. We say that SC motivates research into KM and KM assess SC. If knowledge is situated, then a symbolic KB will require modification as the situation changes. However, if we could demonstrate that the 5 process of change dictated by SC is tamed via current KM tools, then the case for concerning ourselves with SC becomes weak. We argue here that (i) there is enough evidence to support the view that SC is a very strong motivation for KM; but (ii) there is not enough evidence to support the view that KM has tamed the problem of SC. That is, the current gener- 10 ation of KM tools may be an inadequate response to the challenge of SC for symbolic knowledge based systems. 1 Introduction "What is wanted is not the will to believe, but the will to find out, which is the exact opposite." 15 -- Bertrand Russell "We should no longer ask if we should measure, the question today is how." -- Dieter Rombach 20 Proponents of...

: Ripple-Down Rules (RDR) are an alternative from mainstream approaches to the building of knowledge based systems (KBS). RDR use simple, but reliable, techniques for Knowledge Acquisition (KA) and representation which have been shown to support the on-line development, maintenance and validation of KBS. Key features of RDR that affect the approach to V&V are the incremental nature of KA and maintenance, the use of cases for KA and validation, the use of an exception structure for knowledge representation and the development of KBS by experts. This paper describes RDR and its approach to V&V particularly concentrating on recent extensions which use Rough Set Theory for verification and Formal Concept Analysis for validation. Keywords: Verification and Validation, Ripple-Down Rules, Rough Set Theory, Formal Concept Analysis 1 A Different Approach to KA Many approaches to knowledge based system (KBS) development attempt to build complete systems that are mostly considered final before...

Introduction Pathology, St.Vincent's Hospital Sydney [Compton, 1992, Edwards 1993]. PEIRS now has more than 2000 rules. It covers about 25% of Chemical Pathology and is over 95% accurate. The system was put into routine use after 200 rules were added. All other rules have been added by experts as errors occur, without knowledge engineering assistance. It takes about three minutes to add a new rule and the laboratory anticipates continuous on going development as it is trivial to add to the system as domain knowledge evolves. Knowledge acquisition and knowledge maintenance are problems with any expert system. These problems are exacerbated in domains dealing with temporal data such as the example data sets distributed for AIM-94. Knowledge acquisition for such domains requires information about how features in the data are identified as well as how these features are reasoned about. Ripple Down Rules (RDR) is a knowledge acquisition methodology which goes some way towards addr

Abstract With the development of computer games, different game worlds and various game characters are found within them. Various Artificial Intelligence (AI) techniques are usually used to define behavior s of the characters within game worlds, which are controlled by AI algorithms in the computer as well as by the user. The AI techniques defined for these characters are generally developed by the game creators and cannot be changed without going to some effort, which means that if a user wished to control the behavior s of a character within a game, they could not easily do so. Being able to edit the behavior s of AI characters is beneficial as it gives the user extra control over their characters. Therefore a method for allowing a user to easily personalize the AI characters was needed. This goal was achieved by using an incremental knowledge acquisition method, called the MCRDR. The MCRDR allows the user to easily acquire new control knowledge of the AI characters by combining rule-based and case-based knowledge acquisition approach. Our experiment results showed that AI of a character could be personalized with this method of knowledge extraction.