History has been a qualitative discipline and has often been considered part of the humanities, well, historically, but the emergence of big data is likely to extend the use of quantitative methods in historical research in the long run. Big data projects have aimed at everything from finding out where to pick fruit in your city to mapping the prevalence of AIDS in the United States, but a recent project, Global Database of Events, Language, and Tone (GDELT) has compiled a massive database of print media coverage in over 100 languages including Armenian, Azerbaijani, and Georgian. Originally created by Kalev Leetaru and Philip Schrodt at Georgetown University, the GDELT database contains about a quarter of a billion uniquely coded units starting from 1979.
One simple operation to perform with the data is to look at the intensity of a country’s representation in media reports globally. In order to make comparisons over time, the data is normalized using a special algorithm to account for the amount of media disseminated globally in a given time period. We observe the average monthly intensity of media coverage between January, 1979 and October, 2014. After extracting and normalizing the data, what emerges is a snapshot of climatic moments in South Caucasian history that were reported on internationally. By comparing the timing of intensity of coverage with reporting on events in major media outlets such as the New York Times, we are able to identify which events garnered international attention and caused peaks in intensity of coverage.
What appears and will be unsurprising to any Caucasus watcher is that in many ways, the destinies of Armenia and Azerbaijan are tied together due to the Nagrno Karabakh conflict. This fact appears prominently in the graphs below with peaks 2, 4, 5, 6, and 7 in Armenia and peaks 2, 3, and 4 in Azerbaijan tied to the conflict.
Paths do diverge, however, and most of the peaks on the graphs below are country specific rather than Karabakh related. In Armenia, peak 1 marks the 1983 terrorist attack perpetrated by Armenian nationalists in France at a Turkish Airlines check-in counter, while the 1979 Iranian Revolution seems to be the cause of Azerbaijan’s first peak.
Coups and coup attempts are upheavals each South Caucasian country has experienced. Notably, in Azerbaijan the 1993 coup which brought Heydar Aliyev to power (peak 5) and the 1999 parliamentary shootings in Armenia (peak 8) both created spikes in global news coverage.
Monday, November 03, 2014
The recent history of the South Caucasus as seen by the world’s media – Part 1, Armenia and Azerbaijan
Note: In this graph, the country’s mean monthly share of global media coverage (defined as all media contained within the GDELT database) is shown. The table below gives a summary of events in Armenia according to the peak they correspond with:
In 2009, Armenia appears in the international news with a peak caused by a breakthrough in talks on the normalization of relations between Turkey and Armenia, followed by the failure to actually normalize relations (peak 9). In 2010, Barack Obama’s speech on the Armenian Genocide failed to mention the word “genocide,” despite his election campaign promises to the Armenian American community to the contrary, leading to peak 10. Peak 11 marks domestic protests following presidential elections in 2013, but the event which captured the greatest share of the world’s attention was the 1988 Spitak earthquake (peak 3).
Note: In this graph, the country’s mean monthly share of global media coverage (defined as all media contained within the GDELT database) is shown. The table below gives a summary of events in Azerbaijan according to the peak they correspond with. A quick glance at the Azerbaijan table shows two events marked with questions marks – peaks 8 and 11, and the month and year of their occurrence. Despite our best efforts, we were unable to identify the specific events which these peaks correspond to. If you know what these peaks represent, we would love to hear your ideas, so please discuss them on our Facebook page, here.
In Azerbaijan, peak 6 appears to be reports of the first oil to be piped out of the country based on the 1994 “contract of the century”, which itself appears to garner media attention after peak 5 (it is important to remember that events were identified based on timing, and with relatively small peaks, it is possible to misidentify events). Azerbaijan also has a peak related to the Eurovision song contest (peak 14), but, interestingly, not in May of 2012 when the country hosted the event. The hosting of Eurovision in Azerbaijan did not lead to a spike in coverage, in spite of the fact that it was billed by Azerbaijani authorities as a mega event and generally considered a controversial location for the event due to the country’s poor human rights record and prevalence of homophobic attitudes despite the event’s large popularity among the LGBT community. A year later, in May 2013, a spike in coverage coincides with the Eurovision voting scandal involving Russia and Azerbaijan. While Russian singer Dina Garipova finished second among Azerbaijani voters, Azerbaijan did not award any points to the singer.
This brings us to the question - what else wasn’t being talked about in Armenia and Azerbaijan? Some events which received less attention from the world's media than one might expect (they are there, but comparably small) are the 2008 presidential election protests in Armenia, and the “contract of the century” in Azerbaijan, signed in 1994 by the government with a conglomerate of Western oil companies.
In sum, the GDELT database is an interesting tool. While still not fully explored and having some issues, it is a big data project which will likely spur on future developments in the social sciences. On Thursday, we will explore the global media coverage of the third South Caucasus country, Georgia.