Bots and online hate during the COVID-19 pandemic: case studies in the United States and the Philippines

Abstract

Online hate speech represents a serious problem exacerbated by the ongoing COVID-19 pandemic. Although often anchored in real-world social divisions, hate speech in cyberspace may also be fueled inorganically by inauthentic actors like social bots. This work presents and employs a methodological pipeline for assessing the links between hate speech and bot-driven activity through the lens of social cybersecurity. Using a combination of machine learning and network science tools, we empirically characterize Twitter conversations about the pandemic in the United States and the Philippines. Our integrated analysis reveals idiosyncratic relationships between bots and hate speech across datasets, highlighting different network dynamics of racially charged toxicity in the US and political conflicts in the Philippines. Most crucially, we discover that bot activity is linked to higher hate in both countries, especially in communities which are denser and more isolated from others. We discuss several insights for probing issues of online hate speech and coordinated disinformation, especially through a global approach to computational social science.

What social media told us in the time of COVID-19: a scoping review

“…One study analysed 12 million tweets from the USA and 15 million tweets from the Philippines from March 5 to March 19, 2020, and both countries showed a positive relation between bot activities and rate of hate speech in communities that are denser and more isolated than others. 43 Brennen and colleagues qualitatively analysed 96 samples of visuals (ie, image or video) from January to March, 2020, and categorised misinfor mation into six trends, noting that, fortunately, there has been no involvement of artificial intelligence deepfake techniques (ie, techniques used to make synthetic videos that closely resemble real videos) so far. 34…”

Characterizing network dynamics of online hate communities around the COVID-19 pandemic

“…In other words, at time points featuring high levels of hate community homogeneity but low assortativity, hateful accounts may themselves not interact with each other. But it is possible, for instance, that multiple hateful accounts may be targeting the same non-hateful account, as in coordinated harassment campaigns; or a single hateful account may be targeting multiple non-hateful accounts, as in the actions of a hateful opinion leader or an influential troll (Uyheng and Carley 2020a ).…”

“…As previously mentioned, the difference between hate speech and offensive speech is crucial as it recognizes that some tweets may use expletives and similarly profane language but not hatefully target any group in particular. Our model achieved over 83% in terms of both accuracy and F1 score (Uyheng and Carley 2020a ). Other experiments using alternative algorithms (e.g., logistic regression, support vector machines) consistently yielded results bested by random forest models.…”

“…Here, public health concerns were compounded with additional political challenges of a polarized society and fragmented government response to the pandemic (Montiel and Uyheng 2020 ; Uyheng and Montiel 2020b ). Hate, in this case, does not ‘strike down’ toward any racial minority, but ‘strikes up’ at political leaders deemed ineffective at addressing the public health crisis (Lasco 2020 ; Uyheng and Montiel 2020a ). While both forms of hate thus operate with similar community structures, it is pertinent to recognize their distinct functions, and contributions to wider societal strain.…”

ReOpen demands as public health threat: a sociotechnical framework for understanding the stickiness of misinformation

“…The ability to access and analyze large datasets has transformed social scientific research. Data science helps researchers understand actions, behaviors, and choices within networked publics over short periods of time and provides the opportunity to understand misinformation ecosystems, detect bot-driven activities, and track social movements and activism within and across platforms (Benkler et al 2018 ; Freelon et al 2016 ; Freelon et al 2020 ; Jackson et al 2020 ; Lazer et al 2020 ; Singh et al 2020 ; Tacchini et al 2017 ; Uyheng and Carley 2020 ).…”