AI tracks partisan media criticism spikes during polarizing events
New weakly supervised model reveals 2016 Russia probe and Charlottesville drove distrust...
Deep Dive
Researchers developed a weakly supervised learning method to automatically identify tweets criticizing partisan news sources. Using noisy labeling functions based on tweet content and users' historical sharing behavior, they found media criticism spikes during polarizing events like the 2016 Russian interference investigation and the 2017 Charlottesville rally. This type of criticism is also more likely to occur after users have been exposed to unreliable and hyperpartisan media.
Key Points
- Weakly supervised classifier uses twin inputs: tweet content and user's news-sharing history to label criticism of partisan media
- Criticism spikes sharply during politically polarizing events like 2016 Russian interference probe and 2017 Charlottesville rally
- Exposure to unreliable and hyperpartisan media is a strong predictor of subsequent criticism toward news outlets
Why It Matters
Publishes a scalable method to monitor media trust erosion and filter bubble reinforcement in real-time