Using Semantic Concepts to Characterise Various Knowledge Representation Formalisms: A Method of Facilitating the Interface of Knowledge Base System Components

Open Access
Publisher Website
Google Scholar
Abstract
Currently, there are a number of research groups working on various components for knowledge base system (KBSs). As example: (a) novel hardware is being developed for mass storage of simple facts, (b) machines are being built to speed up reasoning with rules expressed in languages such as PROLOG and LISP, (c)algorithms have been designed for automatic maintenance of semantic integrity and for deductive question answering, (d) logical systems are being axiomatised which can accommodate time, beliefs, non-monotonic reasoning and other aspects of knowledge which cannot be handled by classical truth-functional predicate logic, (e) methods are being developed to support multiple user-views of knowldege stored in some canonical from, and (f) some progress has been made in providing natural-language interfaces to knowledge base systems. Integration of such components is problematical for a number of reasons, not the least of which is due to the different terminologies and knowledge representation formalisms which are used by the various components. A possible solution to this problem is to identify a commonly used set of semantic concepts and then employ this set of concepts to characterise the type of knowledge which is processed by the various components. An example of a semantic concept is logical negation (i.e. not). Some knowledge representations, such as those used in classical logic, can accommodate logical negation whereas those used in conventional database systems are unable to represent logical negation other than by omission in conjunction with the closed-world assumption. Choice of an appropriate set of semantic concepts should be based on pragmatic criteria rather than philosophical argument, otherwise it is unlikely that agreement will be reached on what concepts to include. In this short paper we present a version (0) set of concepts which was chosen intuitively. We illustrate how this set might be refined by application to example components of KBSs. This paper is a revised version of a paper presented at the 2nd Alvey-sponsored Workshop on Architectures for Large Knowledge Bases (WALKB2) held at Manchester University and organised by Simon Lavington. At that workshop it was agreed to pursue the approach outlined in this paper by setting up a study group consisting of representatives from industry and academic institutions. The remit of this group is to refine the set of semantic concepts by application to a range of knowledge representations including those used in database models, various formal logics, semantic nets, production systems, logic programming languages, hardware-based systems and so on. The initial output from this group will be the version (1) set of well-defined semantic concepts which all knowledge base research groups will be encouraged to use to characterise the particular components which they are developing. The version (1) set of semantic concepts is scheduled to be available mid 1985.

This publication has 0 references indexed in Scilit: