PhD Candidate
Cornell University
Areas of Interest
- Information, Data, Network, and Communication Sciences
- Security
Poster
Understanding adversarial interactions against politicians on social media
Abstract
Adversarial interactions against politicians on social media such as Twitter have significant impact on society, and in particular both discourage people from seeking office and disrupt substantive political discussions online. In this study, we collected a dataset of 400 thousand users' 1.2 million replies to the 756 candidates for the U.S. House of Representatives in the two months leading up to the 2018 midterm election on Twitter. First we develop new techniques to quantify and categorize adversarial interactions against political candidates at scale. We show that some adversarial interactions against political figures may not be considered toxic in most other contexts hence bring challenge to automatic detection. The breadth of adversarial interactions we observed in the election includes offensive name-calling, threats of violence, posting discrediting information, attacks on identity, and adversarial message repetition. Secondly we characterize users who adversarially interact with political figures. We show that among moderately active users, adversarial activity is associated with decreased centrality in the social graph and increased attention to candidates from opponent party. When compared to users who are similarly active, highly adversarial users tend to engage in fewer supportive interactions with their own party's candidates and express negativity in their public user profile.
Bio
Yiqing is pursuing her PhD in Computer Science at Cornell University under Prof. Tom Ristenpart and Prof. Mor Naaman. Her research interest focuses on measuring and mitigating online threats. In particular, she has worked on understanding adversarial interactions against political candidates and politicians, and designing privacy-preserving similarity lookup service to support better image moderation for privacy-aware clients.
