Join The Conversation With Mishpacha's Weekly Newsletter

Predicting the Next Revolution

Shira Yehudit Djlilmand

If you think a computerized model that can predict a revolution or an outbreak of violence is the stuff of science fiction, think again. Kalev Leetaru’s computer program successfully predicted the Arab Spring uprisings and accurately pinpointed Osama Bin Laden’s final location. Next on the agenda: analyzing the 2012 US election and the chances of an Israeli military strike on Iran.

Wednesday, December 14, 2011

Emotions may have little correlation with computers, but Kalev Leetaru’s model for forecasting future international political unrest is based on just that — an analysis of emotions in the global media. And that’s what makes it unique in the world of computer prediction.

Other computer models have been designed to predict human behavior and even political violence, but only Leetaru’s “Nautilus” model analyzes emotions rather than facts. Leetaru, a 30-year-old computer genius who founded his first web company while in the eighth grade, notes that revolutions are usually caused by the population’s perception of the situation rather than the situation itself.

“Look at China,” he says. “China doesn’t exactly have a great human rights record, but we’re not seeing any signs of revolution there. The people aren’t complaining in large enough numbers yet.”

The Nautilus model, named for the massive University of Tennessee supercomputer it utilizes, is still in the developmental stages. It holds great potential, not only for computer geniuses and prognosticators, but also for governments that need to identify trends and patterns in domestic and international events, among allies and enemies.

Leetaru holds the unwieldy job title of assistant director for text and digital media analytics at the University of Illinois’s Institute for Computing in the Humanities, Arts, and Social Science (I-CHASS), hosted by the National Center for Supercomputing Applications. He already holds three US patents in web-related technology and knowledge management, and reports a great interest in his project from governmental bodies.

Indeed, Leetaru appears to be held in great regard not only in the academic community but also in the political and business worlds. He is in great demand as a lecturer for Fortune 100 companies, national venues such as the Library of Congress, and leading universities. Leetaru was also one of the few academics invited to participate in the Director of National Intelligence conferences.

Regarding US affairs, he hopes the Nautilus model will be able to give forecasts of the approaching presidential election. It has already been able to shed some light on the Occupy Wall Street (OWS) movement, by identifying similar patterns in media reports to those of the recent Egyptian revolution. But today, Leetaru says, things have changed. “Now, we are seeing in the media reporting that the movement, although in the early days it had the potential to explode, has lost its momentum.”

Feelings First

Nautilus’s unique data — emotions or mood — is vital for predicting the kind of political violence that we have recently experienced worldwide. An analysis of surface facts simply wouldn’t come up with the goods, says Leetaru.

“There are lots of models that simply extract surface information — for example, data on what shoes people are buying — and compile a database,” he elaborates. “I’m looking at the undercurrents, and that’s why I’m looking at global tone or mood. I’m not looking for objective facts, but how people feel about them.”

So how is such data gathered and analyzed? First Leetaru, who operates the Nautilus computer by remote from his University of Illinois base, feeds in three massive news archives, totaling more than 100 million articles worldwide. These include the complete New York Times from 1945 to 2005; the Summary of World Broadcasts, which synopsizes news items translated from the local vernacular across the globe from 1979 to 2010; and the Google News English language article archives worldwide from 2006 to 2011.

These media reports are then analyzed according to their “tone,” working with a number of dictionaries used by psychologists that list various words according to their “emotive strength.” For example, “loathe” is classified as a stronger word than “dislike.” The words are then given scores, and the supercomputer begins its job of analyzing global patterns.

The obvious problem here is that media reporting is inevitably deeply biased, depending on political and national affiliations. How can a computer program get around that?

“Of course, all reporting is subjective,” agrees Leetaru. “Even a baseball game will have two reports — one that says it was great and one that says it was terrible. And we have the same thing during war; what’s good for the enemy is bad for us. If South Korea is reporting on North Korea, it will be negative, China’s reports will be positive, and the US’s reports somewhere in between. But that’s all part of the global picture that we’re putting together. What we’re looking for is the tipping point — where the tone of media reports becomes generally more negative — that can indicate an approaching crisis.”

For example, the model found that in Egypt, the media was very neutral, since the government censored media discussion of the uprising. But Nautilus analyzed media reports from across the globe over which the Egyptian censors had no control. Nautilus also analyzed media reports for a long period before the uprisings and found that their tone had gradually become more negative over the last decade. As the revolution drew closer, the media reports got “hotter and hotter.”

“It’s a bit like a weather forecast,” Leetaru explains. “You can give a weather forecast for Jerusalem for next summer, but it’s not going to be very accurate. The closer you get, though, the more accurate you can forecast.”

Indeed, as with weather forecasting, admits Leetaru, the model is not precise or perfect. “With a weather forecast, you can’t say for sure it’s going to rain today, but you can say it’s definitely worth taking your umbrella!”

That’s why Nautilus can’t predict, for example, that “Egypt will fall at five o’clock next Friday morning,” but it can predict a high probability that it will fall at some time in the near future. “Equally, I can’t tell you that Obama will lose the election, but the model can show us that he has lost a lot of global credibility recently.”


To read the rest of this story, please buy this issue of Mishpacha or sign up for a weekly subscription.

Share this page with a friend. Fill in the information below, and we'll email your friend a link to this page on your behalf.

Your name
Your email address
You friend's name
Your friend's email address
Please type the characters you see in the image into the box provided.

Evolution vs. Revolution
Shoshana Friedman I call it the “what happened to my magazine?” response
Up, Up, and Away
Rabbi Moshe Grylak What a fraught subject Eretz Yisrael is, to this day
Where Do You Come From?
Yonoson Rosenblum Could they be IDF officers with no Jewish knowledge?
Heaven Help Us
Eytan Kobre Writing about anti-Semitism should rouse, not soothe
Work/Life Solutions with Chedva Kleinhandler
Moe Mernick “Failures are our compass to success”
An Un-Scientific Survey
Rabbi Emanuel Feldman Are Jerusalemites unfriendly? Not necessarily
Out of Anger
Jacob L. Freedman How Angry Lawyer was finally able to calm down
5 Things You Didn’t Know about…Yitzy Bald
Riki Goldstein He composed his first melody at eight years old
When the Floodgates of Song Open, You’re Never Too Old
Riki Goldstein Chazzan Pinchas Wolf was unknown until three years ago
Who Helped Advance These Popular Entertainers?
Riki Goldstein Unsung deeds that boosted performers into the limelight
Your Task? Ask
Faigy Peritzman A tangible legacy I want to pass on to my children
Are You There?
Sarah Chana Radcliffe Emotional withdrawal makes others feel lonely, abandoned
A Peace of a Whole
Rebbetzin Debbie Greenblatt Love shalom more than you love being right
Seminary Applications
Rabbi Zecharya Greenwald, as told to Ariella Schiller It’s just as hard for seminaries to reject you