The political communication data analysis research of two School of Information Studies (iSchool) faculty members has again been selected for Tow Center for Digital Journalism recognition at Columbia University and funding support from the Knight Foundation.

The work of Professor Jennifer Stromer-Galley and Assistant Professor Jeff Hemsley on “Illuminating 2018” has earned that honor. Their computational journalism project captures and analyzes messaging of candidates and public feedback about what candidates are saying on social media. The research is designed to help journalists and the public look at major trends in political campaign communication on social media channels.

Doctoral student Feifei Zhang and other student members of the iSchool’s Center for Computational Data Science (CCDS) are also involved in the project. The 2018 project continues their earlier reviews of digital political conversations they first began tracking in 2014 and a second effort in 2016 that was a comprehensive review of social media messaging during the 2016 presidential election that won Tow Center/Knight Foundation Innovation Fellows support and funding that year. 

High Accuracy Findings

Zhang noted that the Tow Center’s award’s permitting the continuation of the Illuminating project this year is especially gratifying in light of findings on the 2016 Illuminating project. Machine-learning models that predicted the types of presidential campaign messages generated on Facebook and Twitter in 2016 had an accuracy rate of 75 percent, she said.

That record provides researchers with a high level of confidence about this year’s predictions, too. “According to this previous performance, we are confident that our machine-learning models can be applicable to predict types of campaign messaging in the 2018 gubernatorial and U.S. House and Senate races. The Tow Center grant supports us to continue our machine-learning work to deeply understand online political campaign discourse,” Zhang adds.

The 2018 project has been expanded to include an interactive website that allows users to examine data from the current election cycle as well as archived data from 2016 and 2014 tracking efforts. It is one of 10 projects named in the 2018 Tow Center fellows program. Other funded projects focus on research ranging from artificial intelligence and machine learning, digital security, mobile push alerts, voice assistant news, newsroom diversity, conservative news values and local news capacity and trust.

2016 Tow Fellow 

Stromer-Galley was initially named a Knight Foundation Innovation Fellow in February 2016. She and  Hemsley now join more than 80 current and former fellows whose research is being supported by the Center and the Knight Foundation.

As director of CCDS, Stromer-Galley explores why people talk politics online, what practical addition deliberation can bring to e-government and the development of a coding scheme to assess the qualities of political discussion. Hemsley is involved in CCDS projects and is a founding member of the Behavior, Information, Technology and Society Laboratory (BITS) at the iSchool. His research seeks to understand information diffusion and user interaction in social media, using computational methods to collect, wrangle, visualize and analyze large, heterogeneous datasets.”

Funding for the fellowship projects comes from the John S. and James L. Knight Foundation, adding to the Tow Center’s existing research efforts in journalism and technology connections.