Thanks for the interesting post - it's been making the rounds on Twitter as well and I've been pondering on it just a bit. NLP seems to have found niches of late in ...
1) reputation management - tracking bad or good press in social media (in the general realm of sentiment analysis) 2) targeting ads to users - get users to click on links that they might find interesting, in the hopes they purchase something I must confess that I don't find either problem terribly inspiring. Suppose we solve 1) this means that celebrities or politicians or corporations or perhaps you and I will be able to recognize bad/good press trends and do something based on that, presumably something that corrects bad trends or capitalizes on the good ones. But of course once we solve 1) then there's the related problem of detecting when someone tries to game that, which they surely will. Getting users to click on ads, well that's what makes the web and e-commerce race along, so there's perhaps a greater good achieved by supporting Google and Facebook and all the rest... Nothing here feels like landing a man on the moon or curing cancer though. But, if we can help Justin Bieber maintain his squeaky clean image ... well then it all seems worthwhile. ;) Enjoy, Ted -- Ted Pedersen http://www.d.umn.edu/~tpederse