| Criteria to consider for selecting the expert whose expertise you will encode: |
 |
You want an expert with experience in actually solving the problem.
If your goal is an expert system that will diagnose why a car won't start,
a mechanic is probably going to be more useful than a mechanical engineer.
The expert system will encode rules-of-thumb for providing the advice,
not theoretical explanations. |
|
You want the expert nobody wants you to have.
Experts tend to be busy people, and convincing management that you need access
to their time may be difficult. This is the knowledge engineer's equivalent
to the problem information systems analysts often face: to design an application to
automate a process you need access to the people who actually accomplish the
process. But, the people who best understand how the process works are
overworked -- that's why the process is a candidate for automation in the
first place.
|
|
Continued... |