I'm all for it. However, it would have to be a small library, because you'd get into writing a language parser in that it would have so many words to contextualize it would probably bring up more erroneous suggestions than not.

I think it's a good idea though, ++ for you :-)