Expert systems technology has been around for a long time, becoming increasingly easy to use, inexpensive and reliable in recent years. It would seem to provide an ideal vehicle for the dissemination of expertise in developing countries, particularly in the field of medicine, which was the focus of much early work in diagnostic systems. Despite the apparent match of a real problem and a credible solution, however, remarkably few AI systems for medicine in developing countries have been researched, designed and implemented. This paper addresses why this might be the case, reviews some of the extant systems and explores some of the design issues. Particular emphasis is placed on the question of “Appropriate Technology”. Various criteria for Appropriate Technology are explored, and an optimal set used to guide principles of design. It is argued that medical AI systems can satisfy these criteria, provided that sufficient care is taken in their design for the country of application.

It is often claimed that it is easy to write expert systems. This claim was examined by monitoring experienced programmers learning to use the S.I knowledge engineering tool. Their achievements and difficulties were examined using a framework that has emerged from previous research into novices learning to use standard programming languages. Even though the experienced programmers all had several years' experience of programming in more than one standard language, there were similarities between their difficulties in learning to use S.I and the difficulties of complete novices learning to program in standard languages.

The experienced programmers were however able to overcome their initial difficulties fairly quickly, but it is argued that complete novices would not find it so easy to do so. Also the experienced programmers did take time to develop a repertoire of schemeta for representing different kinds of factual, judgemental and procedural knowledge. It was concluded that in S.1, as with other programming languages and softwares tools, it is easy to learn how to do simple things, but difficult, even for experienced programmers to learn how to do more complex things.

No criticism of S.1 is implied. S.1 was found to be a suitable vehicle for introducing non-trivial knowledge engineering concepts, and we believe that similar difficulties would occur in learning to use other knowledge engineering tools.

Knowledge acquisition research supports the generation of knowledge-based systems through the development of principles, techniques, methodologies and tools. What differentiates knowledge-based system development from conventional system development is the emphasis on in-depth understanding and formalization of the relations between the conceptual structures underlying expert performance and the computational structures capable of emulating that performance.

Personal construct psychology is a theory of individual and group psychological and social processes that has been used extensively in knowledge acquisition research to model the cognitive processes of human experts. The psychology takes a constructivist position appropriate to the modelling of human knowledge processes, but develops this through the characterization of human conceptual structures in axiomatic terms that translate directly to computational form. In particular, there is a close correspondence between the intensional logics of knowledge, belief and action developed in personal construct psychology, and the intensional logics for formal knowledge representation developed in artificial intelligence research as term subsumption, or KL-ONE-like, systems.

This paper gives an overview of personal construct psychology and its expression as an intensional logic describing the cognitive processes of anticipatory agents, and uses this to survey knowledge acquisition tools deriving from personal construct psychology.

Frame systems occupy an important place among formalisms for computer-based knowledge representation. A common concern about frame systems, however, is that they are not efficient enough. We argue that this is not necessarily true of all possible systems, and that the trade-off between generality and efficiency has not been fully explored. While many systems provide generality at the expense of performance, systems closer to the low end of the spectrum have not been investigated nearly as much. Those systems are well suited for applications that need flexible knowledge representation but cannot afford the high performance price.

We describe in detail KR, a very efficient frame system that provides mechanisms for knowledgerepresentation including user-defined inheritance and relations, object-oriented programming, andconstraint maintenance. The system is simple and compact and does not include some of the more complex functionality, but it is highly optimized and offers excellent performance for a variety of applications.

Expert critiquing systems present an interesting and important complement to conventional expert systems. The basic idea is to emphasize the support for the user's own decision making rather than to let the system independently suggest a solution to a given problem. Thus it is assumed that the user initially proposes a decision or course of action. The system then reviews this suggestion relative to known circumstance, tries to evaluate the proposed solution, provides suggestions for improvements, draws attention to possible risks, indicates alternatives, and evaluates their merits, etc. A critiquing system is characterized both by its approach to problem solving and by its style of interaction with the user, as will be detailed below.

Learning is one of the important research fields in artificial intelligence. This paper begins with an outline of the definitions of learning and intelligence, followed by a discussion of the aims of machine learning as an emerging science, and an historical outline of machine learning. The paper then examines the elements and various classifications of learning, and then introduces a new classification of learning based on the levels of representation and learning as knowledge-, symboland device-level learning. Similarity- and explanation-based generalization and conceptual clustering are described as knowledge level learning methods. Learning in classifiers, genetic algorithms and classifier systems are described as symbol level learning, and neural networks are described as device level systems. In accordance with this classification, methods of learning are described in terms of inputs, learning algorithms or devices, and outputs. Then there follows a discussion on the relationships between knowledge representation and learning, and a discussion on the limits of learning in knowledge systems. The paper concludes with a summary of the results drawn from this review.

The computer industry is in the early stages of a revolution. Its capability for processing natural languages will advance dramatically over the next few years. As a result, computer systems which deal with human language will begin to play a central role in the industry. Computers will becooome significantly more accessible to many people, and many tasks which are wholly manual will be handled electronically.

The oft quoted inadequacy of von Neumann architectures for AI applications has regularly been used to justify the design of special purpose parallel machines. In particular, the von Neumann computational model has been criticized as being unsuitable for parallelism because of the memory access bottleneck. For the design of a new machine both top-down and bottom-up methodologies have drawbacks. The middle-out strategy, working both up and down from an intrinsically concurrent high-level programming language as a means of both representing and processing knowledge provides an attractive way of providing a symbiosis between software and hardware. The longest established and most well-founded symbolic method for the representation and manipulation of knowledge is logic. A notable result of the last decade, work on mechanical theorem proving was that a subset of predicate logic, Horn Clauses, can form the foundation of a computational model. The execution model of Prolog, the first popular language based on Horn Clauses, was designed for efficient evaluation on von Neumann architectures. An alternative computational model, more suitable for expressing reactive systems but retaining Prolog's affinity for metaprogramming, has given rise to a new class of languages, concurrent logic languages. One among many of these languages, FGHC (flat Guarded Horn Clauses), was developed by the Japanese as the kernel of their Fifth Generation initiative. A background familiarity with Prolog would be helpful in understanding this article.