The study focused on Twitter and using tweets to help identify dangerous situations, with the goal being to learn if social media trends could be used to identify potentially dangerous developing situations faster than police reports, which have been the longstanding standard.
It turns out that the answer is a resounding yes.
The researchers combined a dataset of 1.6 million tweets from the London riots in 2011 with a machine learning algorithm which automatically scans Twitter for potential threats. The three primary variables taken into account were street name, time of tweet and key words, which vary from one situation to the next, depending on what someone is looking for.
The results from the Cardiff research were confirmation that data drawn from Twitter can predict violent threats up to an hour faster than conventional methods that rely on police reports and official data sources.
The fact that social media is so much faster is a bit surprising. On the other hand, there are some companies in business today selling their ability to do that very thing, which is what makes the results less of a surprise and more of a confirmation. After all, if these companies weren’t successful at making predictions using something close to real time social media data mining, then they wouldn’t still be in business.
The lesson to be learned here is simple. We’re getting increasingly adept at handling very large datasets, and that data can be mined in real time (or close to real time) to produce actionable intelligence.
The same algorithm that can be used to predict violent outbursts in a large city can be tweaked for use by businesses to provide a variety of intelligence. If it’s not something you’ve considered before, now is the time to factor it into your thinking.