This project monitors the activity of a network of accounts that engage in hateful activity on Twitter.
The project monitors in near real time the content, themes, and activity of over 1,000 accounts on Twitter. Our construction of the set of tracked accounts began with the identification of 40 “seed accounts” that regularly engage in spreading hateful content against protected groups. The seed accounts were identified by a group of independent experts. We then used Twitter’s public APIs to generate a list of followers for each of the seed accounts, and we subsequently grouped all of those “follower accounts” in a larger set, ranking them by the number of seed accounts that they follow. Through a combination of algorithmic methods involving statistical feature selection in combination with human curation, we generated a list of the top accounts engaging in hateful conduct amongst these follower accounts. The resultant set of accounts is tracked in this dashboard and includes over 1,000 accounts, though the exact number of accounts may vary because of suspensions and takedowns as administered by Twitter. Finally, we utilize a combination of publicly available and proprietary APIs to analyze the data associated with the follower accounts in real time. This analysis powers the data visualizations available on the dashboard.