Automated knowledge acquisition for knowledge-based systems: KE-KIT

Despite recent progress, knowledge acquisition remains a central problem for the development of intelligent systems. There are many people throughout the world doing studies in this area. However, very few automated techniques have made it to the market place. In this light, the idea of automating the knowledge acquisition process is very appealing and may lead to a break through. Most (if not all) of the approaches and techniques concerning intelligent, expert systems and specifically knowledge-based systems can still be considered in their infancy and definitely do not subscribe to any kind of standards. Many things have yet to be learned and incorporated into the technology and combined with methods from traditional computer science and psychology. KE-KIT is a prototype system which attempts to automate a portion of the knowledge engineering process. The emphasis is on the automation of knowledge acquisition activities. However, the transformation of knowledge from an intermediate form to a knowledge -base format is also addressed. The approach used to automate the knowledge acquisition process is based on the personal construct theory developed by George Kelly in the field of psychology. This thesis gives and in-depth view of knowledge engineering with a concentration on the knowledge acquisition process. Several issues and approaches are described. Greater details surrounding the personal construct theory approach to knowledge acquisition and its use of a repertory grid are given. In addition, some existing knowledge acquisition tools are briefly explored. Details concerning the implementation of KE-KIT and reflections on its applicability round out the presented material