Online Library

Since the beginning of 2020, while societies and economies around the world have struggled to cope with the realities of the COVID-19 pandemic, cyberspace has given governments, businesses, and general end-users the ability to work, play, and connect in new and innovative ways. With everything from workspaces and classrooms to family gatherings and exercise routines forced online, the Internet has enabled people across the globe to carry on and maintain a sense of normalcy during very abnormal times. However, at the same time, while the world has been focused on the health, economic, political, and social ramifications of the pandemic, terrorist organizations, fringe groups, and extremist communities around the world have become emboldened, finding opportunity to exploit the situation, incite hate, (re)mobilize, and promote their ideologies online in novel ways. These groups—which we loosely classify as malicious nonstate actors for the purposes of this research —have been primarily focused on exploiting and contributing to the diffusion of information during the pandemic for their own strategic gain.

Since its early mainstream adoption in the 1990s, the Internet has been leveraged by white supremacist groups to recruit and radicalize individuals. Twenty years later, social media platforms, like YouTube, reddit, and Twitter, continue to further this practice. The attention of researchers has been primarily centered on white supremacist men, and this focus on white men erases white women’s roles as active agents in the spread of white supremacy, skewing our understanding of white supremacy as a whole. This study used digital ethnography and interviews to examine the ways white women are radicalized to organized white supremacy through popular social media platforms YouTube, reddit, and Twitter. The study found white women were radicalized by engaging with posts and joining communities focusing on beauty, anti-feminism or “The Red Pill,” traditionalist gender values or #TradLives, and alt-right politics. White supremacist recruiters leveraged gendered topics and weaponized platform features – likes, sharing, comments, recommendation algorithms, etc. – to cultivate a sense of community. Through involvement with these communities, women were introduced to racialized perspectives on each topic, usually after a catalytic pop culture or newsworthy event, and slowly radicalized to organized white supremacy.

Using a dataset of over 1.9 million messages posted on Twitter by about 25,000 ISIS members, we explore how ISIS makes use of social media to spread its propaganda and to recruit militants from the Arab world and across the globe. By distinguishing between violence-driven, theological, and sectarian content, we trace the connection between online rhetoric and key events on the ground. To the best of our knowledge, ours is one of the first studies to focus on Arabic content, while most literature focuses on English content. Our findings yield new important insights about how social media is used by radical militant groups to target the Arab-speaking world, and reveal important patterns in their propaganda efforts.

VOX-Pol Blog post.

With the ever-increasing cases of hate spread on social media platforms, it is critical to design abuse detection mechanisms to pro-actively avoid and control such incidents. While there exist methods for hate speech detection, they stereotype words and hence suffer from inherently biased training. Bias removal has been traditionally studied for structured datasets, but we aim at bias mitigation from unstructured text data. In this paper, we make two important contributions. First, we systematically design methods to quantify the bias for any model and propose algorithms for identifying the set of words which the model stereotypes. Second, we propose novel methods leveraging knowledge-based generalizations for bias-free learning. Knowledge-based generalization provides an effective way to encode knowledge because the abstraction they provide not only generalizes content but also facilitates retraction of information from the hate speech detection classifier, thereby reducing the imbalance. We experiment with multiple knowledge generalization policies and analyze their effect on general performance and in mitigating bias. Our experiments with two real-world datasets, a Wikipedia Talk Pages dataset (WikiDetox) of size ~ 96k and a Twitter dataset of size ~ 24k, show that the use of knowledge-based generalizations results in better performance by forcing the classifier to learn from generalized content. Our methods utilize existing knowledge-bases and can easily be extended to other tasks.

Hate speech detection on Twitter is critical for applications like controversial event extraction, building AI chatterbots, content recommendation, and sentiment analysis. We define this task as being able to classify a tweet as racist, sexist or neither. The complexity of the natural language constructs makes this task very challenging. We perform extensive experiments with multiple deep learning architectures to learn semantic word embeddings to handle this complexity. Our experiments on a benchmark dataset of 16K annotated tweets show that such deep learning methods outperform state-of-the-art char/word n-gram methods by ~18 F1 points.