Photo: Business Insider
A team of IBM researchers spends their days sifting through Twitter. They use live streams of tweets to develop machines that are smarter than the typical computer, an area of study known as “machine learning.” Using these tweets, they’ve developed technology that allows a machine to understand that some tweets are just background noise and others are newsworthy and important.
For instance, a tweet that says “I urgently need my cup of Starbucks and a scone and before I head over to Staples” is distinctly different than a Tweet that says: “URGENT: I just bit into a scone from @starbucks to find over 10 staples baked into it. Please RT and be careful.”
IBM scientists have also come up with ways to measure “sentiment” … to identify which tweets are saying something good about something important and which are saying something negative.
After two years of studying Twitter, their work wound up in an IBM social media monitoring product, Cognos Consumer Insight.
But it also led to lots of funny stories and interesting facts about Twitter. Rick Lawrence, who leads IBM’s Machine Learning Group at IBM Research at Yorktown Heights NY, shared some of these stories with Business Insider.
IBM can predict wait times at airports by crowdsourcing information from tweets. They search tweets for mentions of airports, then send an @reply to the tweeters and ask them to reply with wait times.
Your tweets give away where you're from. Scientists can tell with amazing accuracy whether you're from the West, East, Midwest, or South, just by the words you use in your tweets.
Because of its 140 character limit and use of abbreviations, Twitter-speak is like its own language complete with swear words.
People are more inclined to Tweet something negative than positive. It's like the Eeyore of the social media world
No one leaked that Watson won Jeopardy, even though the show was witnessed live by dozens of IBMers and others, and aired weeks after it was taped.
Every public tweet since Twitter's inception in March 2006 will be archived digitally at the Library of Congress. IBM plans to map every archived tweet to Wikipedia, and tag it with sentiment, to make them more digestible.