A problem solving strategy based on knowledge-based systems

• Main Author: Gillies, Alan Cameron
• Published: University of Central Lancashire 1992
• Subjects: 005; Technologies not elsewhere classified
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.314789

The historical development of knowledge based systems (KBS) from artificial intelligence (AT) has led to a number of characteristics which isolate knowledge based systems from the rest of software development. In particular, it has led to the growth of 'stand alone' systems. This thesis argues that this has restricted the use of KBS to a narrow range of problems, and has reduced the effectiveness of the consequent solutions. By considering first a specific problem in some depth, the thesis seeks to develop an alternative approach, where KBS is considered as simply another software technology to be used within an integrated solution. The problem considered is the automatic analysis of photoelastic fringe patterns, and KBS methods are employed alongside conventional image processing techniques to produce an integrated solution. The conventional algorithmic solution is first constructed and evaluated. This solution, having proved partially successful, is then enhanced by the use of KBS techniques to provide a full solution. From this specific example, a framework for integration is derived. This framework is tested in an unrelated application to consider whether the approach adopted has more general utility than one specific class of problem. This problem was the provision of decision support for business planning based upon market research. The resulting strategy and design is described together with details of how the system was implemented under the supervision of the author. The thesis concludes with an evaluation of the work and its conthbution to knowledge in the twin areas of the specific solutions and the underlying methods.