Ripple down rules (RDR), that is rules with hierarchical exceptions, are used in knowledge acquisition because they provide a well intelligible and modifiable representation for even very large expert systems. In this paper a formal semantics for RDRs is proposed, that covers first order rules as well as attribute-value based rules. An algebraic foundation is proposed, including simplification of RDRs and transformation of RDRs into flat lists of rules and ripple down rule sets, hence these knowledge representation schemes are put into perspective. It is shown, that a RDR has a shorter description length than an equivalent decision list. Induction of rules with exceptions is characterized as bidirectional movement in the hypothesis space, while known algorithms for learning rules or decision trees either perform a top-down specialization of the most general or a bottom-up generalization of the most special hypothesis. Known algorithms for induction of RDRs are summarized and compared a...

. This paper suggests that a distinction between knowledge acquisition methods should be made. On the one hand there are methods which aim to help the expert and knowledge engineer analyse what knowledge is involved in solving a particular type of problem and how this problem solving is carried out. These methods are concerned with classifying the different types of problem solving and providing tools and methods to help the knowledge engineer identify the appropriate approach and ensure nothing is omitted.. A different approach to knowledge acquisition focuses on ensuring incremental addition of validated knowledge as mistakes are discovered (validated knowledge here means only that the earlier performance of the system is not degraded by the addition of new knowledge). The organisation of this knowledge is managed by the system rather than the expert and knowledge engineer. This would seem to correspond to human incremental development of expertise. From this perspective...

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.

maintenance with the author. The role of a discussion paper is to provoke discussion. Hence, my comments will be heretical rather than theoretical. For the theory, see [19]. Today I want to say two things. Firstly, the knowledge maintenance (KM) problem is our next big challenge. A common method for tackling the KM problem is to use some type of meta-knowledge to represent expected/unexpected or good/bad knowledge. This metaknowledge takes many forms including architectures [26], patterns [12], problem solving methods (PSMs) [29, 31], or ontologies [13]. Elsewhere [20], I have argued that patterns of design and architecture are very similar to PSMs and ontologies: Both represent abstract descriptions of supposedly common parts of many designs.

Abstract. The goal of this work was to develop an adaptable computer vision system that refines itself to the specific task of extracting lung boundary in High Resolution Computed Tomography (HRCT) scans. We have developed an incremental learning framework called ProcessRDR that allows the underlying procedures of a computer vision system to learn knowledge pertaining to their control. This approach to learning control knowledge provides a systematic mechanism to customisation of the procedures for a domain, whilst the system is in operation. 1