ABSTRACT

Some knowledge engineers favour a method which requires the development of a prototype system based very often on the prior knowledge of the knowledge engineer. The prototype is demonstrated to the domain expert who suggests modifications and amplification. The changes are made and the revised system demonstrated again. The iterations of this process continue until the domain expert is satisfied and the model is acceptable. If a good initial model is produced this method can be very productive. However, it can have the effect of prejudicing the responses of the expert and thus diverting him or her from some of the subtle, more intuitive knowledge, that might be of crucial importance to the operation of the system. An alternative is to start with a blank sheet of paper and ask the domain expert to tell you what he or she knows. A fairly extensive set of knowledge is then assembled before the initial system is coded. This approach is fundamentally better but its success is critically dependent on the time that the domain expert can devote to the process.