Skip to main content
Figure 1 | EPJ Data Science

Figure 1

From: Activism via attention: interpretable spatiotemporal learning to forecast protest activities

Figure 1

Spatiotemporal occurrence for different social movements by date (x-labels) and by location (y-labels). A red circle indicates at least one offline protest event happening on a particular day and state; the blue shade indicates the volume of tweets posted on the corresponding day/state. Charlottesville counterprotests exhibited burst patterns, in which most of the activities were sparked by a deadly violence attack and President Trump’s statements on Aug 12th, 2017. In the first few days following the attack and the statements, more protest events occurred nation-wide and larger tweet volume was observed. The Ferguson I protests appeared to have a gradual build-up process, in which the activities were initially local (around Missouri and few states) following the shooting of Michael Brown on Aug 9th, 2014 and later received global attention. A global increase in tweet volume was observed until Aug 20th, 2014. The Ferguson II protests started on Nov 24th, 2014, with the announcement of the jury decision not to indict the police officer, and garnered global attention. The tweet volume for each state was greater in the first two days after the jury decision compared to the other days

Back to article page