In your position, which I think is:
- Must create Expert System.
- Knowledge Domain is not fixed.
- Happy with MS technologies.
I’d forget the technologies altogether to start.
I’d find an expert. Interview him. Figure what he knows, what makes it special… Write it down. Be sure you have something real.
Then look at ways to do it.
I’d probably sketch out the idea writing some code that might use a database at the back end.
A lot of ES work kills itself before it starts by deciding on an inference engine up front. I say map out your knowledge before you decide how to represent it.
There are a lot of red herrings in the ES field. Like ES shells that are flaky. Be careful. My observation is that a lot of the prior art hasn’t endured!
That’s enough for now.