Two professors from the S.I. Newhouse School of Public Communications who are working on the development of technology to detect manipulated media and combat the spread of fake news are supported by a subcontract that now tops $1.1 million, thanks…
Knight Foundation Funds Research on Paid Political Ads on Social Media
Research that sheds light on messaging in the 2020 U.S. presidential campaign ads on social media has received funding support from the John S. and James L. Knight Foundation.
The Foundation has awarded $148,000 to support Illuminating 2020, a project spearheaded by Syracuse University’s School of Information Studies Professor Jennifer Stromer-Galley and Assistant Professor Jeff Hemsley, along with University of Liverpool Derby Fellow Patricia Rossini.
The project aims to provide more comprehensive information and greater transparency about the content and reach of political ads on Facebook and Instagram as a way to help journalists who cover the 2020 presidential campaign. Researchers are using state-of-the-art algorithms and other computational tools to categorize and classify the statements candidates and others make in those ads, then making that information into a usable database and sharing it online to assist journalists.
“This project will empower journalists to leverage technology and better inform the public on important issues, especially with an election cycle in high gear,” says Paul Cheung, Knight Foundation director for journalism and technology innovation. “Journalists want to understand how candidates are using Facebook and Instagram, and given their enormous influences on the electorate and the public debate, it’s also important to learn how bad actors are using political ads on these social platforms.”
While other studies have tracked social media postings of candidates and the structured data surrounding them (such as changes in the number of followers), Stromer-Galley, Hemsley and Rossini have taken a different tack.
Their team of faculty, doctoral, masters and undergraduate students are examining what the candidates are actually saying on social media. Using a variety of machine learning techniques, “we were able to accurately characterize the topic, tone and style of political campaign messages on social media,” Rossini notes. Researchers also classify the toxicity levels of what is being said, then compile and share the data.
As Stromer-Galley details, “With online ads increasingly being microtargeted to segments of the public, journalists need tools to be able to see and report on what candidates and other political organizations are saying to whom during the 2020 presidential campaign behind the closed doors of Facebook and Instagram.” Hemsley adds, “By using the tools and techniques we have developed over several years here at the iSchool, we hope to empower journalists to do that.”
Why Help Journalists?
Political reporters face the challenge of covering not only candidate speeches, campaign events and TV ads, but also what is happening on social media.
The sheer volume of information flowing through social platforms makes it difficult to accurately and comprehensively report on that arena, Stromer-Galley believes.
“Journalists will be able to use the site to find out where a given candidate is currently running negative ads, for example. They could also see or be alerted to when there are major changes in campaign strategy by a candidate, or they can see if political action committees are running especially uncivil, or what we are calling ‘toxic,’ ads in battleground states,” Stromer-Galley explains.
“Our goal is to help journalists in that important work by providing a usable yet comprehensive summary of the content of messages that goes beyond counting likes or retweets. The Illuminating project enables political journalists an insightful yet accessible summation of the important political persuasion that is happening online,” she adds.
Five Years of Message Data
The 2020 research is the newest part of a larger project that Stromer-Galley, Hemsley and Rossini have been conducting since 2014 to study and classify Twitter and Facebook posts by political candidates on their free social media accounts. Past research covered the 2014 campaigns of governors across the United States, the 2016 presidential campaigns, and 2018 U.S. House and Senate and nationwide gubernatorial campaigns.
In addition to Knight Foundation funding, Illuminating 2020 is also supported by the iSchool’s Center for Computational and Data Science (CCDS).
The Center builds on the iSchool’s historic strengths in human language technologies, such as natural language processing and machine learning, and its emphasis on data science research. CCDS researchers are advancing the science of data collection, retrieval, curation, analysis, and archiving and are applying those techniques to important social problems. Stromer-Galley is founding director and Hemsley is co-director of CCDS.
Journalists who are interested in gaining access to the interactive database and further information are welcome to contact the center at ccds.ischool.syr.edu.