A knowledge acquisition framework for an intelligent decision-support system

One of the lessons learnt from the early pioneering expert systems is that excellent decision making in itself is not sufficient to guarantee client acceptability. User involvement is necessary in the development of all interactive expert systems. We propose a framework for facilitating the knowledge acquisition process. We focus on the iteration cycles in expert-client interaction, which avoid mismatches in knowledge process and enhance the verification of knowledge. A combination of model-based and rule-based structure is demonstrated, which provides richer representation and makes the maintenance process much simpler. The proposed framework has been applied to build a successful industry expert system, Respirator Advisory System (RAS), in collaboration with 3M Australia Pty Ltd.<>