Soft computing for intelligent knowledge-based systems

Knowledge-based systems are founded on the idea that knowledge should be declarative, so that it can be easily read, understood, and altered by a human user as well as by a machine. Logic fulfils these criteria, and logic programming has been widely used for implementing knowledge-based systems. One major shortcoming of logic programming is the lack of a mechanism to deal with the uncertainty inherent in many knowledge-based systems. Soft computing is a key technology for the management of uncertainty, although so far its major successes have been centred on fuzzy control rather than higher level information management. This paper outlines some of the issues related to the area of soft computing in knowledge-based systems, and suggests some simple problems to test the capabilities of software. Fril is discussed ns an implementation language for knowledge-based systems involving uncertainty, and some of its applications are outlined.