Do the words journalists choose reveal unconscious political biases or create/reinforce similar biases? The 2016 and 2020 presidential elections made vicious and fierce rhetoric the norm and strong political identification a fact of life. The accusations of “fake news” and bias ring loudly, destroying the perception of a free and fair press and posing a threat to democracy. The challenge for this project is to develop a human-centered process and natural language processing tools to help journalists make less freighted word choices; that is, a detector for partisan language when covering political news.
Jeremy Gilbert is the Knight Professor of Digital Media Strategy. Both his work and teaching focus on the content and revenue strategies of existing and emerging media companies. He explores the intersection of technology and media, examining how new tools and techniques will affect the creation, consumption and distribution of media.
What does biased language look like?
Whose viewpoints determine partisan words and phrases?
Can more neutral terms be found that make news seem less partisans?
Can a detector be built that helps journalists create news stories without perceived bias?
Weeks 1-3: Reading & research; interviews with journalists, researchers and news consumers of differing political viewpoints
Weeks 4-7: Create and test different word choices and design interfaces for a potential tool
Weeks 8-9: Test the prototypes and identify next steps
Through this project, we'll better understand how journalists and news consumers perceive bias via word choice. We'll prototype, build and test potential tools to help journalists find neutral terms for partisan issues.