Comparing media coverage of July conflicts
After experimenting with GDELT data for a single day, I was eager to see whether I could compare media coverage of conflicts over a longer span of time. Arbitrarily, I picked the month of July, which saw the escalation of Israel's Operation Protective Edge, the downing of flight MH17 over Ukraine, ISIS's expansion in Iraq, and lots of other horrible news.
I used a slightly tweaked version of the Ruby script described in the last post to add up the total article count for "violent" events related to eight national conflicts: Israel, Ukraine, Afghanistan, Iraq, Syria, Central African Republic, Honduras, and South Sudan. I keep putting the word "violent" in scare quotes because I suspect my methodology rests on very shaky ground, even ignoring GDELT's various flaws. Let me explain.
As I mentioned in July, GDELT uses algorithms to extract distinct "events" from news reports and then codes each event according to the CAMEO system. I, more or less arbitrarily, designated events with codes fifteen through twenty as "violent" events. Here's what those codes mean, officially.
|15||EXHIBIT MILITARY POSTURE||"Mobilize or increase armed forces"|
|16||REDUCE RELATIONS||"Expel or withdraw peacekeepers"|
|18||ASSAULT||"Use as human shield"|
|20||ENGAGE IN UNCONVENTIONAL MASS VIOLENCE||"Engage in mass expulsion"|
There is lots of room for debate here, obviously. Does code thirteen, "threaten," as in "threaten mass violence," count as a violent event? Probably, but I wanted to impose some restrictions in order to measure coverage of conflicts rather than the entire national life of countries.
My script ran through the GDELT data for every day in July 2014, pulled out every "violent" event related to the countries in question, and added up the number of articles recorded for those events. (Using the numarticles value for each GDELT entry.)
There is a huge spike in Ukraine coverage on the day MH17 was shot down, but the trend drops off quickly, although coverage remained generally higher in the second half of the month than in the first. Meanwhile, Syria coverage remains steady -- and low compared to Israel and Iraq -- despite some of the bloodiest fighting of the civil war taking place in the second half of the month. South Sudan, CAR, and Honduras barely register.
Here are the complete data.
|Date for July 2014||Israel/Palestine||Ukraine||Afghanistan||Iraq||Syria||Central African Republic||Honduras||South Sudan|