Knowledge Acquisition

Abstract

It has been said that the problem with being a parent is thai by the time you become experienced, you are, for the most part, out of work. Similarly, the goal of every knowledge engineer should be unemployment. During implementation of the core problems, the expert becomes more familiar with the techniques used to implement the system and begins to feel more ownership of it. With any luck, the domain expert should be able to directly update and maintain the system. With the mechanism described here, this is possible, since the expert can have extensive, hands-on experience in KBA construction. By far the best situation is when the domain expert can build the KBA directly. This is rapidly becoming the stale of the art for smaller ES and is rapidly becoming more plausible for larger systems. A long range goal of ES research is to capture and reproduce the knowledge engineer's current role - to develop an Expert System that represents the knowledge engineer's expertise. This would allow KBAs to be build much more cheaply and quickly and would allow knowledge engineers to move on to more challenging and complex domains. We believe that our prototype described here has taken the first step at capturing the expertise that the knowledge engineer retains. It is our hope that this type of development is continued by others faced with similar problems so that robust Expert Systems for problems in CPE can be produced quickly and economically to answer the computing challenges being faced in the coming decade.