Misleading Repurposing on Twitter

Abstract

Twitter allows users to change their screen name and other profile attributes, which allows a malicious user to change their account's identity or purpose while retaining their followers. We present the first large scale and principled study of this phenomenon of misleading account repurposing on Twitter. We analyze two large datasets to understand account repurposing. We find 3,500 repurposed accounts in the Twitter Elections Integrity Datasets. We also find more than 100,000 accounts that have more than 5,000 followers and were active in the first six months of 2020 using Twitter's 1% real-time sample. We analyze a series of common features of repurposed accounts that give us insight into the mechanics of repurposing and into how accounts can come to be repurposed. We also analyze three online markets that facilitate selling and buying Twitter accounts to better understand the effect that repurposing has on the Twitter account market. We provide the dataset of popular accounts that are flagged as repurposed by our framework.

Characterizing Retweet Bots: The Case of Black Market Accounts

“…Collecting the data from the 5,332 suspended accounts was more challenging. Internet archive's Twitter Stream Grab provides 1% of all tweets since 2011 (Archive 2020) and has been used extensively by past research (Tekumalla, Asl, and Banda 2020;Elmas et al 2020). By mining this dataset, we collected roughly 1% of all tweets from these accounts and their profile information.…”

A Dataset of State-Censored Tweets

“…Although the dataset is publicly available, only a few studies (Tekumalla, Asl, and Banda 2020;Elmas et al 2020a;Elmas et al 2020b) have mined the entire dataset due to the cumbersome process of storing and efficiently processing the data. We mine all the tweets in this dataset to find tweets and users with the "withheld in countries" field and, thus, find the censored tweets and users.…”