The Global Data on Events, Location and Tone dataset contains information on events both good and bad, pulling information from news stories. Natural language processing software works out what one “actor” has done to another, and also adds a location for the action. For example, if a report contains the line “Syrian military units shelled civilians in Homs” the event might be coded “SYRMIL fought CVL” and include the latitude and longitude for that city.
To concentrate on Syria’s civil war, New Scientist selected acts of “material conflict” from within the country’s borders between all domestic actors – including military, government, rebels, civilians and media. We further narrowed the focus to events with a "Goldstein scale" of -9 or less. These are judged to be the most destabilising actions, including fighting, violent repression and destruction of property.
Some events were described only as occurring in Syria, and had no more specific location. These are included in the chart, but do not appear on the maps.
Analysis and graphic by Peter Aldhous; published 13 May 2013.
Source: Kalev Leetaru and Philip Schrodt / GDELTNumber of events:
1 to 49
50 to 999
1000 to 4999
5000 or more