Showing papers in "Artificial Intelligence Review in 1988"
TL;DR: This tutorial for novice knowledge engineers and managers discusses some considerations involved in using multiple experts, including deciding when multiple experts may be necessary, eliciting knowledge from multiple experts individually or in small groups, and knowledge engineer capabilities and preparation.
Abstract: The already difficult knowledge acquisition process is complicated when the expert system being developed requires interaction with multiple experts. In this tutorial for novice knowledge engineers and managers we discuss some considerations involved in using multiple experts, including (1) deciding when multiple experts may be necessary, (2) eliciting knowledge from multiple experts individually or in small groups, and (3) knowledge engineer capabilities and preparation. Next, we present three specific group-appropriate techniques to elicit knowledge during a knowledge acquisition session: brainstorming, consensus decision making, and the nominal group technique. Finally, we consider the importance and objectives of debriefing following knowledge acquisition from multiple experts
50 citations
TL;DR: This work elaborate on these two approaches to learning and contrast the symbolic search space paradigm with the connectionist paradigm.
Abstract: At a symbolic level cognition can be modelled as a production system where meaning units are represented as condition-action rules. Anderson (1982, 1987) provides a good example of how learning can occur with this type of knowledge representation. At a subsymbolic level cognition can be modelled with a connectionist network where meaning units are represented as patterns of parallel distributed activity. The work of the McClelland and Rumelhart (1986) group is a prototype of this approach. We elaborate on these two approaches to learning and contrast the symbolic search space paradigm with the connectionist paradigm.
36 citations
TL;DR: The history of AISB is certainly of interest and I would like to add some precision to your statement (Artificial Intelligence Review (1987) 1 ,137-138) that ECAI-82 was "retrospectively declared to have been the first ECAi", which could be misunderstood by some nasty minded people.
Abstract: The history of AISB is certainly of interest and I would like to add some precision to your statement (Artificial Intelligence Review (1987) 1 ,137-138) that ECAI-82 was \"retrospectively declared to have been the first ECAI\". Your statement could be misunderstood by some nasty minded people (and such people seem to exist among us scientists) as a hint at tension caused by some underhand appropriat ion of the 1982 AISB meeting by the newly formed European Coordinating Commit tee on AI. It is quite true that during the winter of 1979, when the AISB commit tee decided that I would be the general chairman for the 1982 meeting, it was understood that the meeting would be an AISB event, and everyone knew that the only European AI meeting was the AISB one. However, during the winter of 1981, when we were preparing the Call-for-Papers for the meeting, we all agreed that the name ECAI would be better than AISB. Therefore all the conference anouncements and proceedings were given the name \"ECAI-82\". You should be aware of this as you had a paper, enti t led \"How to write a story\", on pages 259-260 of these proceedings. ECAI and ECCAI are the chi ldren of AISB, their birth took place under the AISB commit tee 's supervision. I would not like to claim that everything has been perfect and rosy. It can be said, for instance, that ECCAI has quickly become more than the s imple umbrella organization it was planned to be. We have also had some benefit sharing problems, the AISB commit tee which inherited ECAI and ECCAI was not the commit tee that decided their birth, and so on and so forth. In the case of ECAI-82, Great Britain wil l ingly offered Europe something which she was holding in her hand. This may seem utterly unbelievable to today's young 'wolves ' but, as far as I have witnessed, this is the process which took place!
27 citations
TL;DR: The notion of metadata and the role it plays in a user's understanding of data in a database is introduced and a metadata knowledge representation (MAKR) is described which relates database metadata to a KL-ONE style meaning representation.
Abstract: We introduce the notion of metadata and the role it plays in a user's understanding of data in a database. We survey the kinds of metadata present in database schemas through an Abstract Conceptual Schema, and also the kinds of metadata associated with frames in knowledge engineering tools. Finally we look at networks used for natural language understanding, (KL-ONE and MEMORY), and describe a metadata knowledge representation (MAKR) which relates database metadata to a KL-ONE style meaning representation. Examples in MARK are used to illustrate the kinds of metadata surveyed in the first part of the paper.
26 citations
TL;DR: The time has come for all those working in AI to take the issue of professionalism seriously and produce a code or codes which will demonstrate that work is being done responsibly.
Abstract: The time has come for all those working in AI to take the issue of professionalism seriously. Professional standards will be difficult to establish in AI. However, there will be pressure from various directions to produce a code or codes which will demonstrate that work is being done responsibly. Such codes will be largely worthless unless they are produced by people actually working at the ‘sharp end’ of AI.
4 citations
TL;DR: It is argued that formal semantics, in the model-theoretic style pioneered by Tarski, is appropriate for specifying the meanings of the compositional component of artificial formal languages but not of natural languages.
Abstract: It is argued that formal semantics, in the model-theoretic style pioneered by Tarski, is appropriate for specifying the meanings of the compositional component of artificial formal languages but not of natural languages. Since computer programming languages are both formal and artificial, formal semantics has a clear application to them, but this does not mean that it is in any way relevant to the problem of meaning in AI. The distinction is drawn between what an expression in a language means, and what a person means by using it. The former is the only kind of meaning that formal semantics can ever explain, whereas for AI to succeed it is essential to elucidate, and then to recreate, the latter. No verdict is offered on whether or not this may ultimately be possible; but it is argued that formal semantics would be an inappropriate tool to use to this end.
2 citations
[...]
TL;DR: It is argued that AI is best defined in terms of its implicit goal—the creation of thinking beings by means other than biological reproduction, and can be seen as a devious and sophisticated form of birth control.
Abstract: This paper argues that AI is best defined in terms of its implicit goal—the creation of thinking beings by means other than biological reproduction. Thus AI can be seen as a devious and sophisticated form of birth control. A discussion of the prospects for AI research from this viewpoint follows. It is concluded that the total extermination of the human species would be a better test of whether AI has succeeded than Turing's celebrated ‘imitation game’.
2 citations
TL;DR: It is argued that cognitive science needs to distinguish between ‘competence’ and ‘performance’ in order to clarify this point, and that such a distinction has far-reaching effects on how the authors view computer simulations of behaviour.
Abstract: Cognitive science may be loosely described as the activity of trying to model aspects of human behaviour upon a computer. It has emerged as a blending of the techniques of artificial intelligence and cognitive psychology but these two disciplines have different, and incompatible, philosophies. Searle has detected elements of behaviourism and operationism within artificial intelligence, whereas cognitive psychology is essentially anti-behaviourist. It is argued that cognitive science needs to distinguish between ‘competence’ and ‘performance’ in order to clarify this point, and that such a distinction has far-reaching effects on how we view computer simulations of behaviour.
2 citations
TL;DR: John von Neumann died a little over thirty years ago, and this short essay is a personal tribute to him.
Abstract: John von Neumann died a little over thirty years ago. This short essay is a personal tribute to him.
1 citations
1 citations
TL;DR: In this paper, Fodor & Pylyshyn's (1987) critique of Connectionism will be examined and an argument on how these two approaches could be reconciled through the notion of ‘movement between levels‘ in a cognitive system is presented.
Abstract: In this paper, Fodor & Pylyshyn's (1987) critique of Connectionism will be examined. It is worth considering this paper in some detail, since it represents the first comprehensive attack on Connectionism from within Cognitive Science. We shall begin by presenting a summary of the disagreement between Classicists and Connectionists. We shall conclude by presenting an argument on how these two approaches could be reconciled through the notion of ‘movement between levels‘ in a cognitive system.
TL;DR: It is claimed that in order to provide a model of the human mind, a robot would have to possess not only intelligence but also a use for that intelligence; and for that to be so it would has to possess purposes which were its own, not simply those of its manufacturer.
Abstract: This paper traces the origin of intelligence in biological organisms, thought of as negative entropy systems, relating it to both perception and mobility. A distinction is made between behaviour which is a function of an animal's beliefs about its circumstances, and behaviour which is related to those circumstances more directly, only behaviour of the former sort being regarded as intelligent. It is claimed that in order to provide a model of the human mind, a robot would have to possess not only intelligence but also a use for that intelligence; and for that to be so it would have to possess purposes which were its own, not simply those of its manufacturer. There would, that is, have to be things which it was not only able but wanted to do. And for that to be so, its experience of the world would have to possess not simply an epistemic but also a phenomenal aspect. It could not be assumed that that would be so, however, since a robot, even if functionally equivalent to a person, would presumably be structured differently and made of different materials.