Knowledge acquisition issues in the design of decision support systems in quality control

Implementation of formal quality policy in manufacturing environments requires extensive knowledge in many different fields: (1) knowledge of the problems that can be found; (2) knowledge of the methods and procedures that can definitively improve the process; (3) knowledge of the quality techniques that can be used; and (4) how to implement these techniques in manufacturing environments. Different experts are able to provide such knowledge, ranging from operators to quality engineers, who for example provide knowledge about specialised quality tools. Intelligent decision support systems (DSSs) can be used to make quality expertise available to people who face quality problems every day. However, these systems are difficult to elaborate especially due to the acquisition and modelling tasks, considered as the bottleneck of intelligent systems development. Based on our experience of DSS development in the quality area, we propose here to analyze and classify the knowledge necessary to develop advisory systems for quality applications (e.g. process diagnosis, selection of quality tools and configuration of quality tools). Elicitation methods which have been proved successful for extracting quality knowledge are presented as well as guidelines for the design of acquisition sheets. Then we propose to describe the subsets of knowledge in a unified structure, which can be further implemented by the use of object-oriented formalism and rules. The structure greatly facilitates the design of advisory quality systems in manufacturing environments.