Natural Language Processing

Recently I have been dabbling in the world of computational linguistics and have come to the conclusion its a highly saturated buzz word that has a lot to do with 3.0. So I ask, what could I do that's new and novel? Nothing much. Even though I have had a few ideas that were totally authentic that came from my brain, after googling around for a couple hours, their is a pretty good chance your novel idea has already been done or talked about. My boss says new and novel is defined by greater output then the original means. So i thought of an idea to create a corpus of bad assness that uses hackerish techniques to scrape data from the web and apply Vector Space analysis. Doing so would populate a massive corpus to help disambiguate words in worse sense disambiguation. maybe I have solved NLP and given birth to web 3.0.