Using Big Data to map Ukraine's protest violence.
- By Kalev LeetaruKalev H. Leetaru is the Yahoo! Fellow in Residence at the Institute for the Study of Diplomacy in the Edmund A. Walsh School of Foreign Service at Georgetown University. His work centers on the application of "big data" towards understanding global human society in new ways and he is the creator of the GDELT Project.
After nearly a week of bloodshed, Ukraine seems to have found a moment of peace. What the map above offers is a glimpse of what big data can tell us of the unrest through the eyes of the world’s news media. Contrary to the image that has emerged in Western media centering on a single square occupied in the capital city, we see instead a conflict that reaches to the farthest corners of a nation, not only between the police and protesters in Kiev, but in protests that have spread to other cities. In short, big data allows us for the first time to map quantitatively how large-scale societal unrest brings a nation together, even as it tears it apart.
In order to give a full picture of what the unrest in Ukraine looked like from November 2013 to the present, we combined three cutting-edge big data tools. The first was Global Database of Events, Language, and Tone (GDELT), which monitors the world’s news media each day to compile a list of what’s happening around the globe in over 300 categories from protests to peace appeals. Google’s BigQuery system was used to scan in just a matter of seconds all quarter-billion GDELT records to identify the location and intensity (as proxied by media volume) of protests (indicated by the pink dots) and violence against civilians (red circles). And, finally, CartoDB was used to put it all on a map. In all, GDELT processed more than 1.4 million news stories from around the world reporting on events in Ukraine during this period — all protests and violence against civilians from Nov. 1, 2013 through Feb. 19, 2014.
You can click on each location to see an example of the data indicating unrest or violence there. Some indicators include links to news articles about localized unrest; others cite broadcasts or local news sources not available online. A location might show a convoy of 20 police buses heading towards Kiev from Vasylkiv, or a protester traveling from Berdyansk — revealing to us how the protests spread to other cities such as Ternopil, Lviv, Ivano-Frankivsk, Kirovograd, and Rovno.
While this map offers only a post-mortem of the violence in Ukraine to date, in a few weeks GDELT will be releasing a new global map of protests and violence against civilians that will update each morning, making it possible for the first time to track global unrest in every country of the world as it happens. And though it is far from perfect, this map provides a first look at the ability of big data to allow us to look at conflict through the eyes of millions of news articles, and, shortly, to watch it as it unfolds before us, in real time.