True sentiment analysis requires more than just picking out keywords and pairing them with obvious indicators of emotion like "hate" or "love." In addition to the words themselves, language contains sarcasm, irony, humor, and a slew of other nuances determined by context. Parsing sentences to understand exactly what they mean seems to require a human element, but researchers are getting closer to making it an exact science.

Engineers at IBM have designed a computer program to compete against humans at a game of Jeopardy, named Watson. While Watson is being called a first step toward realistically achieving artificial intelligence, what the program can already do is where the future of sentiment analysis programs is headed.

Jeopardy questions (answers) can be phrased a variety of ways, from analogies to quotes, much like customers' comments. Facebook status updates are rarely grammatically correct, relying mostly on sentence fragments to make a point. Deciphering what users mean en masse will take at least all the analytical skills a program like Watson possesses.