I spoke to Bill Hunt at the Silicon Valley Hackers and Founders meetup last night, CEO of StockMood. StockMood, a recent TechCrunch 50 finalist, is scouring the Web and determining whether the buzz around company X is hot or not. Based on this information, alongside trend data of company X’s stock, they hope to provide valuable insight to consumers and trading firms alike. Sign up for their beta–you’ll get invited promptly.
How are they gathering this information? Humans? That is so 20th century. Machines, of course! They are using sentiment analysis, an up-and-coming trend.* Sentiment analysis, according to Wikipedia, is (I have technical research papers, if you’re interested in more elaboration):
“…a broad (definitionally challenged) area of natural language processing, computational linguistics and text mining. Generally speaking, it aims to determine the attitude of a speaker or a writer with respect to some topic. The attitude may be their judgment or evaluation (see appraisal theory), their affectual state (that is to say, the emotional state of the author when writing) or the intended emotional communication (that is to say, the emotional effect the author wishes to have on the reader).”
This is very exciting news, indeed, as I expect this to be one of the many factors in an upcoming AI revolution. No revolution is without a complete list of players, of course, and StockMood isn’t the only one.** In fact, I recently blogged about a web application that psychoanalyzes blogs.
Collective Intellect, out of Boulder, CO, made StockMood’s exact value proposition approximately two years ago. However, they have moved toward becoming more of a marketing and analysis tool for companies based around that the buzz they get in social media applications.
Jodange (pronounced ye-dah-nj), is a NY-based spinoff of Cornell’s research in the area. They are a broad-based opinion collector that act as a search engine for sentiment. Check out their demo or even try their iGoogle gadget!
Happy Tweets analyzes the sentiment of a user’s recent tweets. It ranks each user from 100 to approximately 750 (the current score of the happiest tweep). Amusingly, I rank quite poorly with a score of 457:
The current Happyscore for abossy is:
457
Generally, people who have followed this person on Twitter lately will perceive this person as
Not Very Happy
While the author, Tim, claims to have built the website for his work in computational linguistics, this is a prime example of sentiment analysis. The two categories of research aren’t mutually exclusive, by any means, and have quite substantial overlap.
The beauty of such fields–as well as others that will partake in the AI revolution–is that they are largely invisible to the end user. This will result (as my Dad predicts) in a highly fragmented market with a huge multitude of niches, unlike the social networking and e-mail giants we have come to know (Facebook, Myspace, Gmail, Yahoo! Mail, etc.). This is due primarily to the network effect, where a service’s value is determined largely by the number of users, and each user being dedicated to only one or two services. The uses and value of sentiment analysis is still unclear, and Bill Hunt is certainly not trying to sell StockMood as a one-stop-shop, but as an additional tool for the arsenal of big traders.
I am eager to partake in the events as they unfold.
* Not enough of a trend for it to appear on Google trends, however. That doesn’t help my case at all.
** See a more exhaustive list than my few examples on the now-thrice-cited Wikipedia page on sentiment analysis.
EDIT: Added HappyTweet description. Added sentence about the network effect for clarity.
