Microsoft Predicts The Future Using News Headlines Of The Past

Scientists from Microsoft Research and the Technion-Israel Institute of Technology have developed software, which could predict future events such as disease outbreaks with 70%-90% accuracy.

In their research paper, the two scientists say that using a mixture of archived news reports and real-time data, they were able to make connections between droughts and storms in parts of Africa and cholera outbreaks, split by years of time.

This prototype uses archived material from the New York Times, combined with real-time data from other websites (including Wikipedia) to spot links in past events to predict the future.  The experts focused on riots, deaths and disease outbreaks in their research, and achieved the high accuracy you see above. The experts focused on predicting riots, deaths and disease outbreaks and say their accuracy was between 70%-90%.

As an example they identified, in 1973 the New York Times reported on a drough in Bangladesh, and on a cholera epidemic in 1974.

The same events occurred 10 years later in the same country, as the paper reported again on a drought in '83 and cholera deaths in '84.

"Alerts about a downstream risk of cholera could have been issued nearly a year in advance," wrote researchers Eric Horvitz, director of Microsoft Research, and Kira Radinsky, PhD student at Technion-Israel Institute of Technology.

Of course, we're still speaking in theoretical terms, as this research has tended to be retrospective and not actually used to predict an event; but it still shows a clear pattern and strong success rate of prediction.

Source: Microsoft Research

Jason England