When I was 18 (10 years ago) I did loads of research into that stuff and knew just as much about information retrieval, vector space tf-ids models, latent semantic indexing, wordnet analysis, etc. At the time it was fairly cutting edge research. This stuff isn't anything new now, I was actually forced to decipher some research papers instead of reading popular books on the subject. It was fairly obvious to me back then that none of these techniques worked well for general web search. I did end up building a system that clustered google search results (in realtime) into DMOZ categories letting you refine your search results by clicking a category (which was actually useful and worked quite well in case you were searching for something ambiguous like "jaguar").
None of these techniques are new to anyone working in information retrieval. Just looking at co-occurrence of words in tweets and expanding the query with some related terms (weighted appropriately) would probably achieve what he has done (weekend project for an average dev).
I'd call this kid really smart if he'd actually figure out how to improve general web search, or could think of a useful application at least. Talking about existing research and making it look like your own isn't great form in my opinion. Coming up with your own definition for a "word" just makes you look stuck up. Much better off acknowledging work of other researchers and quoting them, although that would never generate as much press I guess.
Sorry if the rant is quite negative, I'm just getting a bit fed up with all the marketing surrounding "young geniuses" and teenage entrepreneurs these days. If I wanted to read that stuff I'd get the local paper.