New algorithm spotlight: Researchers from the Ohio State University trained an AI to guess a congress member's political party, based on his or her tweets about the coronavirus. According to the findings, Democrats discussed the pandemic more often and stressed public health threats, while Republicans placed greater emphasis on China and the impacts on businesses.
- The algorithm, which was trained on 30,887 tweets from Jan. 17 to March 31, was correct 76% of the time.
- During that time, Democrats sent out a total of 19,803 tweets versus 11,084 for Republicans.
- The correlation between a political party and a person's views on COVID-19 became more polarized over time, peaking in mid-February.
- Once both parties began to determine the political implications, the polarization in tweets became noticeable "pretty quickly,” study co-author Jon Green said.
- The study was published in Science Advances.