Development of automatic knowledge-acquisition expert system

The authors have developed an automatic knowledge-acquisition expert system in order to reduce the time involved in applying expert systems to the automation of steel industry processes. This system automatically gathers a large body of case data and generates a decision tree by using information entropy. The application of the system to control of the material flow at the cooling bed of a steel plate mill confirmed the effectiveness of this knowledge acquisition method. Noise case data, conflict case data, and bad attributes, which could be identified by reference to the decision trees, should be filtered out to optimize the quality of the acquired knowledge