Social media can predict the future!
Analysis of tweets produces accurate forecasts on certain topics
- By Kevin McCaney
- Oct 28, 2010
The word on the street has moved to cyberspace. If you want to know what’s happening — or in some cases, what’s going to happen — forget the barber shop, barstool or backyard fence. It’s apparently better to analyze search engines and social media.
Researchers at Indiana University and the University of Manchester have found that future direction of the stock market can be predicted by analyzing the millions of tweets each day on Twitter, Bloomberg Businessweek reported. The researchers looked for keywords and ran them through an algorithm to determine people’s moods and said they had an 87 percent accuracy rate with their predictions.
Earlier this year, Hewlett-Packard’s Social Computing Lab took a similar approach to predicting the box office performance of as-yet-unreleased movies, according to a report in The Technium. In fact, merely tracking mentions of upcoming films proved a pretty good predictor of their performance. But when researchers ran the tweets’ positive or negative sentiments through an algorithm, the forecasts became practically dead-on.
Government agencies also have seen the wisdom of tracking the crowd. During a session at the recent Executive Leadership Conference in Williamsburg, Va., Todd Park, chief technology officer of the Health and Human Services Department, mentioned that during last year’s flu outbreak, Google searches provided more up-to-date information than traditional reporting methods on how and where the flu was spreading. It seems like a logical result: People get sick, they want to know what it is, so they Google flu symptoms to see if theirs match. Google has a Flu Trends website based on search terms.
The U.S. Geological Survey also draws on tweets for its Twitter Earthquake Dispatch, which tracks tweets about temblors.
Social media wouldn’t work as a predictor for everything, of course. If people don’t have a role in determining an outcome, their sentiments wouldn’t apply, so you can skip looking for tweets about the sixth race at Belmont. But where the public’s attitudes or experiences come into play — and campaign leaders everywhere are surely working on this — following the crowd could be the smart thing to do.
Kevin McCaney is a former editor of Defense Systems and GCN.