According to a new study by Cardiff University, Twitter can identify dangerous situations up to an hour faster compared to police reports. Researchers used a series of machines that learns the algorithms from a dataset of 1.6 million tweets of the London riots in 2011, which was then used to identify threats.

The system took a number of details into account such as the location, timing and frequency of the tweets, which could contain certain words (or variants of the words) that would hint as a possible threat.

As the report notes, many existing methods of event detection aims for large-scale threats like terror attacks, so it would be much harder to be alerted to smaller incidents like fires or car accidents. However, leveraging social media data can solve this problem while enhancing intelligence-gathering techniques that are already used by the authorities.

Story source: The Verge

Image source: Swansboro