Data Science Study of “Unite The Right” and social media

In response to the violence in Charlottesville, the Data Science Institute at the U. of Virginia is undertaking a unique project to help understand the ways people use social media to physically, and politically engage in the world around them.

By Melissa Moody, U. of Virgina

Former Pres. Barack Obama’s tweet paraphrasing Nelson Mandela in response to the violence of the Aug. 12 ‘Unite the Right’ rally in Charlottesville is the most liked tweet in Twitter history – with more than four and a half million likes and more than one million retweets.

Social media has a major impact not only the world we inhabit online but also the world we physically inhabit and the ways we engage with our communities more than ever before. In response to the violence in Charlottesville, and the social media activity that incited that violence and misdirected first responders, the Data Science Institute at the University of Virginia is undertaking a unique project to help understand the ways people use social media to physically, and politically engage in the world around them.

A capstone project involving four graduate students and three faculty advisors, including DSI Director Dr. Philip Bourne, Founding Director Don Brown, and Dr. Abby Flower, will use the data to ask two key questions:

  • How can the DSI determine the unfolding of the social network connecting various protesters before, during, and after the events in Charlottesville on August 11 and 12?

    Using the request for a Charlottesville event permit as a starting point, researchers will work to determine key indicators in the shaping of the social network. Determine how the network regarding the event evolved. They will be looking for geographic, demographic, and chronological patterns in the network.

  • Can social media smoke screens be detected and avoided?
  • It has now been confirmed that in at least one case, social media was used to mislead/misdirect valuable first responders to events that were not actually taking place. This resulted in less resources at the scene of major activity. Using events that were promoted on social media and that actually occurred as a control, researchers will mine the data to determine if there are language differences that could be indicative of whether an event will actually transpire.

“In the aftermath of the events in Charlottesville, University leaders observed that with more information and analysis, emergency responders could have been better equipped to deal with the rally and related events,” said Arlyn Burgess, associate director of strategic initiatives and operations at the DSI. “By better understanding the events and the people that host them as well as their sphere of influence, we can better prepare emergency responders, and give people important information on how, where, and in what manner to respond.”

The DSI is partnering with IBM and their platform BlueMix, which the students will explore for its capabilities in text analytics and social media analysis. This project is also partnering with the Governor’s Data Internship Program (GDIP), which is run through the Virginia Information Technologies Agency.

The projects pair Universities with agencies and capabilities to help answer grand challenges in the Commonwealth of Virginia. The Data Science Institute has previously partnered with the Department of Motor Vehicles (DMV), the Department of Aging and Rehabilitative Services (DARS) and the Department of Consolidated Laboratory Services (DCLS), among others. We are also working on a project this year with the Virginia Department of Health on the opioid epidemic as a part of the GDIP program.

The UVA Data Science Institute’s innovative project will aid in understanding the predictability of the level of violence via the dissemination of information through social media and other channels. Through this partnership between academia, industry, and government, DSI researchers can use the latest techniques and experts with unique data access. There are opportunities to bring in additional data sets to gain even greater insight with the possibility of resulting platforms, algorithms, and analyses that can be disseminated to government officials and Universities to be used in the case of future incidents.