Yang, K. C., Torres-Lugo, C., & Menczer, F. (2020). Prevalence of low-credibility information on twitter during the covid-19 outbreak. arXiv preprint arXiv:2004.14484. https://arxiv.org/abs/2004.14484 

University of Michigan - Ann Arbor

Low-credibility information is not that common compared to high-quality and factual information found on Twitter. Social bots are significantly more active than actual users in spreading and posting low-credibility information on Twitter.

Prevalence of low-credibility information on twitter during the covid-19 outbreak

Identification of low-credibility information 
- Studied a list of 570 low-credibility sources
- First dataset consisted of public tweets from March 9 to 29 containing the hashtags #coronavirus and #covid19 obtained through API from the Observatory on Social Media that can be traced back to government and other authoritative websites through their URLs to compare Tweet domains 
- Second dataset consisted of Tweets from February 1 to April 27 that can be traced back to low-credibility sources

Bot detection
- Use BotometerLite, which is a bot detection model for large-scale bot detection that gives social media accounts a number between 0 and 1, with 1 being very bot-like

Learn Before

Related