Creating service expertise from raw data with expert system software

Abstract

Putting expertise into software requires us to first understand how humans become experts and how they process information into the bases in their minds. From this point of view, it becomes readily apparent why field service personnel who are expert in servicing several equipment types will perform much better than novices when a new equipment type. They use their general service and troubleshooting competencies to structure the new information around their proven knowledge retrieval and problem-solving methods. In this article we will explore how we can endow service support software with these competencies in order to drastically shorten the software's learning period, the time it takes to make the knowledge base ready for deployment. This software can process raw design data, and raw service data, into ready-to-use knowledge-all with minimal demands on the time of human experts. The concepts presented have been field proven in problem resolution software for supporting equipment maintenance, service and repair. Based on our experience and findings reported by users of other approaches, we have determined that the combination of case-based and model-based reasoning, and relying on automated import into the knowledge model, leads to both fast deployment and high diagnostics performance.