Using a supercomputer to predict major world events
Is conflict is so unpredictable, that it is strangely predictable?
University of Illinois scientist Kalev Leetaru thinks so.
Leetaru used a supercomputer to analyze 30-years worth of news reports. From those reports, Leetaru could identify the signals of conflict from revolutionary uprisings in Egypt to narrowing down Osama bin Laden's location. Using 100 million news stories from the US Government's Open Source Center, BBC Monitoring, New York Times' archive, and other online media outlets, researchers could see clear spikes of tone around the world.
Of course, news sources are inherently biased. But it's the only real-time record we have of human behavior. The results are published in a paper called "Culturnomics 2.0: Forecasting large-scale human behavior using global news media tone in time and space."
The Nautilus SGI supercomputer, which is located at the University of Tennessee, crunched 100 million articles. The stories were analyzed for tone and geography, using a 30-year archive of global news. The data set included 10 billion people, places, things, and activities - totaling 100 trillion relationships.
“Almost every Fortune 500 company monitors the tone of news and social media coverage about their products,” Leetaru said in a statement. “There’s been a huge amount of research coming out of the business literature on the power of news tone to predict economic behavior, yet there hasn’t been as much work in using it to predict social behavior.”
Not only did the supercomputer predict the revolutions in Egypt, Tunisia, and Libya, it showed that location is also important, especially in the case of locating Osama bin Laden.
“I never expected to pinpoint him so accurately,” Leetaru said in a statement. “But it’s fascinating—if you make a map of all the cities mentioned in articles about him over the last decade it leads to a 200-kilometer radius around where he was found.’”
Crowdsourcing bin Laden's location worked. As it turns out, the 200km radius actually included the city he was eventually captured in.
The graph above shows the media sentiment of Egypt from January 2006 to March 2011, with a clear spike when the revolution began earlier this year.
Looking at the tone of all news mentions of a country over time can help predict future conflict.
Quid co-founder Sean Gourley also wants to understand conflict. In February 2003, before the United States invaded Iraq, Gourley was a Ph.D. student, studying the mathematics of war, at the University of Oxford.
Gourley used public information culled from cell phones, texting and attack data from media reports, nonprofit organizations and the United Nations to look for signatures of insurgency. He ran his statistics through a computer program and discovered universal patterns of war. The characteristics looked the same across Afghanistan, Sri Lanka, Columbia, Rwanda and Peru.
via BBC [Supercomputer predicts revolution] and Tennesseeor Full paper here
This post was originally published on Smartplanet.com