Common sense reasoning is one of the hardest parts of AI. I don't think top-down solutions will work. Look to a project like Cyc for one of the best efforts, and most wasted money
http://www.cyc.com/
You can't build a top down taxonomy of ideas and expect everything too work. You can't just "hard code" the ideas.
I think building tools from the ground up, with increasingly complicated and capable recognition and modeling, might work. For example, a visual object class recognition suite that first learned faces, phones, cars, etc. and eventually moved on to be able to recognize everything in a scene, might be able to automatically perhaps with some training build up the taxonomy for common sense.
On the last point, I'd start with even lower level - objects, how many, how large, etc, and only then try to figure out what those objects are.
Also, I think it's interesting that it appears that the first successful AI will come from analysis of visual information (computer vision) not textual (like the recent Twine, or semantic web in general).
A point of relevancy to this site is that Serge Belongie started a biometrics startup while an undergraduate at Caltech. Failure for biometrics companies is even higher than normal for technology startups.
You can't build a top down taxonomy of ideas and expect everything too work. You can't just "hard code" the ideas.
I think building tools from the ground up, with increasingly complicated and capable recognition and modeling, might work. For example, a visual object class recognition suite that first learned faces, phones, cars, etc. and eventually moved on to be able to recognize everything in a scene, might be able to automatically perhaps with some training build up the taxonomy for common sense.