Towards Knowledge Acquisition by Experts

From the three basic knowledge acquisition types for expert systems — indirect knowledge acquisition with a knowledge engineer asking an expert, direct knowledge acquisition mainly by the experts on their own, and automatic knowledge acquisition with machine learning techniques — we currently view direct knowledge acquisition as the most promising approach with respect to total project costs and the technical state of the art. Prerequisites are knowledge representations as well as problem solving methods easily comprehensible for experts and comfortable and easy to learn knowledge acquisition components. In this paper we give an overview on our research to achieve both requirements by “strong” problem solving methods and graphical knowledge acquisition facilities. This is demonstrated with a successful implementation for the well known problem class heuristic classification.