Hi Edward,
I've asked myself the same question. My approach was data-driven and similar to what @sitkack recommends you.
I really like that people like you exist and ask questions like that. The only way to a better world imho.
I dislike @waxy's idea, because by putting a registration barrier he locks ideas. I've the picture of a kobold hiding a sack of gold, hehe =) Don't get me wrong, it's fun.
However, I was thinking that when I get enough data I can use prediction algorithms for future trends, add a list of common problems to the fields to be resolved. Then it should be possible to solve it using algorithms and input of people only. Those people would be the modifier for the algorithm, by telling their location specific problem with a question/answer game.
It should be possible to solve loghoritmically. You could ask people questions, similar to the iphone game that guesses which artist you think of by asking questions. log(2) 100 ~=7 (yes/no)
It could use Google-Trends/Google Prediction API or the new startup that google bought to forecast future trends.
I've chosen my last idea using that method. But I haven't written the algorithm down yet, so I followed it manually..
But the best way is to just go into a company and ask the CEO/Engineer/Developer/Worker questions and do so for as many componies as possible. Process that information into clean datasets and you can start to work with that.
That's Business Intelligence.
hope it doesn't sound too mad. If so, just forgot what you read. :P