Expert systems for management control: a multi-expert architecture

Abstract

The use of Expert Systems technology in management decision making domains is increasing rapidly as business environments worldwide grow more turbulent and as the cost of development tools decrease. Research effort in this field however, is concentrated largely on confined areas such as market analysis, financial diagnosis and production scheduling. The development of an Expert System to support a wider management area presents problems of both size and complexity since such a system would require a large monolithic knowledge base which would exhibit the associated problems of maintainability, consistency and reduction in inference speed. This paper describes a blackboard based Multiexpert architecture that is capable of integrating the problem solving capabilities of a range of confined expert systems in order to provide problem solving support for a wide area such as management control at the strategic level. The system consists of several dedicated expert modules in the area of marketing, finance, production and so on as well as a control module that handles problem decomposition, task allocation and dynamic scheduling. A prototype version of such a system has been successfully implemented in Prolog.