I've used this. It's interesting for a while. For a long time I had wanted to participate in David Lenat's Cyc project which set out to do the same thing, except they didn't involve massively volunteered participation. They were having a few highly trained and trusted volunteers do it all, and it was crawling along.

The ultimate way to build a semantic network would be to spider semantic data on the web as described in this fictional article, written in 2002, purporting to be from 2008.