Two other approaches, a bit messier but with bigger coverage:
* ConceptNet (http://conceptnet5.media.mit.edu/), I think the first major attempt at a "messier" crowdsourced commonsense-facts database. Not a carefully designed hierarchical ontology like Cyc, but rather a large set of lightly curated statements about natural-language nouns and verbs. Has the pros and cons you might expect: better coverage, and easier to use, but less precision and a lot of ambiguity due to polysemy.
* CMU's "Read the Web" (http://rtw.ml.cmu.edu/rtw/) and U. Washington's "Open IE" (http://openie.cs.washington.edu/), both projects that combine a shallow-parsing fact-extractor with web-scale crawling. They are truly gigantic, but the accuracy is not very high.
Cool - it's a long time since I paid much attention - I did research on commonsense qualitative models of engineering domains and that led to a personal interest in the wider issues of commonsense reasoning but at that time (early 90s) I ended up deeply skeptical of the approaches that were common then (in my case mostly various qualitative/sorted/temporal logics).
Worth noting that this approach has already been tried:
http://en.wikipedia.org/wiki/Cyc
"Commonsense reasoning" has been around for a long time as an AI topic - like many AI topics it appears straightforward but is actually full of complexity.