Analysis of COVID-19 5G Conspiracy Theory Tweets Using SentenceBERT Embedding

Elroy, Or; Yosipof, Abraham

doi:10.1007/978-3-031-15931-2_16

Cited by 8 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One of the most popular models (and architectures) employed is BERT, first published in 2018, which has achieved the state of the art for a range of NLP applications, especially classification-oriented tasks. BERT-like pre-trained language models are typically used in recent research to build text classifiers for various text classification tasks, including hate speech (Basile et al 2019, Aluru et al 2020, Mathew et al 2022, offensive language (Wiegand, Siegel and Ruppenhofer 2018, Zampieri et al 2019, Mandl et al 2021 or (pre-specified) conspiracy theories (Pogorelov et al 2020, Moffitt, King and Carley 2021, Elroy and Yosipof 2022, Phillips, Ng and Carley 2022. The majority of these benchmark datasets are in the English language and were drawn primarily from Twitter (cf.…”

Section: Training Of Custom Models To Detect Antisemitic Commentsmentioning

confidence: 99%

8. Algorithms Against Antisemitism?

Steffen,

Pustet,

Mihaljević

2024

Antisemitism in Online Communication

View full text Add to dashboard Cite

The proliferation of hateful and violent speech in online media underscores the need for technological support to combat such discourse, create safer and more inclusive online environments, support content moderation and study political-discourse dynamics online. Automated detection of antisemitic content has been little explored compared to other forms of hate-speech. This chapter examines the automated detection of antisemitic speech in online and social media using a corpus of online comments sourced from various online and social media platforms. The corpus spans a three-year period and encompasses diverse discourse events that were deemed likely to provoke antisemitic reactions. We adopt two approaches. First, we explore the efficacy of Perspective API, a popular content- moderation tool that rates texts in terms of, e.g., toxicity or identity-related attacks, in scoring antisemitic content as toxic. We find that the tool rates a high proportion of antisemitic texts with very low toxicity scores, indicating a potential blind spot for such content. Additionally, Perspective API demonstrates a keyword bias towards words related to Jewish identities, which could result in texts being falsely flagged and removed from platforms. Second, we fine-tune deep learning models to detect antisemitic texts. We show that OpenAI’s GPT-3.5 can be fine-tuned to effectively detect antisemitic speech in our corpus and beyond, with F1 scores above 0.7. We discuss current achievements in this area and point out directions for future work, such as the utilisation of prompt-based models.

show abstract

Section: Training Of Custom Models To Detect Antisemitic Commentsmentioning

confidence: 99%

8. Algorithms Against Antisemitism?

Steffen,

Pustet,

Mihaljević

2024

Antisemitism in Online Communication

View full text Add to dashboard Cite

show abstract

Section: Training Of Custom Models To Detect Antisemitic Commentsmentioning

confidence: 99%

Antisemitism in Online Communication

2024

View full text Add to dashboard Cite

The normalisation of hate speech, including antisemitic rhetoric, poses a significant threat to social cohesion and democracy. While global efforts have been made to counter contemporary antisemitism, there is an urgent need to understand its online manifestations. Hate speech spreads easily across the internet, facilitated by anonymity and reinforced by algorithms that favour engaging--even if offensive--content. It often takes coded forms, making detection challenging. Antisemitism in Online Communication addresses these issues by analysing explicit and implicit antisemitic statements in mainstream online discourse. Drawing from disciplines such as corpus linguistics, computational linguistics, semiotics, history, and philosophy, this edited collection examines over 100,000 user comments from three language communities. Contributors explore various facets of online antisemitism, including its intersectionality with misogyny and its dissemination through memes and social networks. Through case studies, they examine the reproduction, support, and rejection of antisemitic tropes, alongside quantitative assessments of comment structures in online discussions. Additionally, the volume delves into the capabilities of content moderation tools and deep-learning models for automated hate speech detection. This multidisciplinary approach provides a comprehensive understanding of contemporary antisemitism in digital spaces, recognising the importance of addressing its insidious spread from multiple angles.

show abstract

“…However, with the passage of time, even the most publicised and newsworthy events disappear from public memory or even become a subject of misinformation (Elroy, Erokhin, Komendantova, & Yosipof, 2023;Elroy & Yosipof, 2022;Erokhin, Yosipof, & Komendantova, 2022). Features of human cognition lead to reliance on shortcuts, to cognitive biases which may lead to the creation of a subjective reality about the event, its causes and consequences.…”

Section: Introductionmentioning

confidence: 99%