An Approach for Representing Administration Processes: Experience From an Expert System Application

Abstract

Administration processes are a promising field for applying expert system technology. A key problem is the representation of the knowledge in this application domain. Commercial rule- oriented expert systems offer only knowledge representation on a low level. But an expert describes his tasks on a higher abstraction level. Until more powerful and suitable Al-tools or languages are available application development will concentrate on mapping the knowledge into current low-level representations. We present a structured approach supporting a stepwise mapping of knowledge into a rule-based expert system shell. The knowledge engineer uses intermediate semi-formal knowledge representations, which can presumably be transformed automatically into the target knowledge base. This knowledge engineering process is based on a model for administration processes. Our approach was applied (but is not limited) to the development of an expert system concerning the European Materials Logistics System (IBM), which is a system of computer programs and administrative activities for the parts requirement generation of all European IBM plants. Controlling the course of this system is mental routine work for highly qualified people. They can be supported by the developed expert system.