MIT’s Luminoso Claims It’s Cracked the Code on Text Mining
Luminoso, a startup hatched from MIT's Media Lab that is officially launching this month. Luminoso's not the only company analyzing text on the Internet in order to mine such relationships, but it claims its technology is a quantum leap over competitors like Lexalytics, Crimson Hexagon and Clarabridge.
How?
Catherine Havasi, Luminoso's co-founder and CEO (pictured, third from right), says the company's technology is based on 14 years of research at MIT that applies artificial intelligence, probability and a database of 17 million facts (many taken from Wikipedia) to appropriate common sense knowledge to text mining.
A video on Luminoso's website explains how the technology works. First a word cloud appears of all the terms related to a business. In the video's example, Luminoso scoured reviews for a Boston-based hotel. The software then lets its users cross-search for the word "loved" and then colored those terms blue. In that case, the name was blue, so you could surmise that people love the name of the hotel.
So far, Luminoso at least has been successful in convincing some top companies — who pay in the five figures for its services — that it has built a better linguistic mousetrap. Partially this is because of the rigor that Luminoso (the name is a musical term meaning "to play luminously") has applied to its searches. While the conventional wisdom is that Twitter is the mother lode of such knowledge, Havasi says that YouTube comments are often richer sources and "don't get me started on mommy blogs." The technology's efficacy has let Luminoso exist without funding since 2009. The company has also attracted MIT Media Lab alums Jason Alonso and Rob Speer as well as Kenneth Arnold, a doctoral candidate at Harvard.
See the full story here: http://mashable.com/2013/04/12/luminoso/?utm_medium=feed&utm_source=feedburner&utm_campaign=Feed%3A+Mashable+%28Mashable%29&curator=MediaReDEF