"Subverting the Jewtocracy": Online Antisemitism Detection Using Multimodal Deep Learning

Abstract

The exponential rise of online social media has enabled the creation, distribution, and consumption of information at an unprecedented rate. However, it has also led to the burgeoning of various forms of online abuse. Increasing cases of online antisemitism have become one of the major concerns because of its socio-political consequences. Unlike other major forms of online abuse like racism, sexism, etc., online antisemitism has not been studied much from a machine learning perspective. To the best of our knowledge, we present the first work in the direction of automated multimodal detection of online antisemitism. The task poses multiple challenges that include extracting signals across multiple modalities, contextual references, and handling multiple aspects of antisemitism. Unfortunately, there does not exist any publicly available benchmark corpus for this critical task. Hence, we collect and label two datasets with 3,102 and 3,509 social media posts from Twitter and Gab respectively. Further, we present a multimodal deep learning system that detects the presence of antisemitic content and its specific antisemitism category using text and images from posts. We perform an extensive set of experiments on the two datasets to evaluate the efficacy of the proposed system. Finally, we also present a qualitative analysis of our study.

Codes, Patterns and Shapes of Contemporary Online Antisemitism and Conspiracy Narratives – an Annotation Guide and Labeled German-Language Dataset in the Context of COVID-19

“…Antisemitic Online Content Almost all scientific studies known to us addressing the large-scale annotation of antisemitic content in text data rely on the working defini-tion by the International Holocaust Remembrance Alliance (IHRA) as the main basis for their coding schemes Chandra et al 2021;Guhl, Ebner, and Rau 2020;Jikeli, Cavar, and Miehling 2019;Jikeli et al 2022;Schwarz-Friesel 2019). 2 As shown by Jikeli, Cavar, and Miehling (2019), using an English-language corpus containing the word (stems) 'Jew*' or 'Israel', the IHRA definition is well suited in such a setting to generate a gold standard corpus for antisemitic content.…”

“…The model is based on a self-developed lexicon consisting of over 2,000 relevant words and phrases containing Nazi-Germany rhetoric, dehumanizing adjectives and violence-inciting verbs, far-right terminology, alt-right neologisms, coded language, and revived conspiracy theories. In Chandra et al (2021), a multimodal deep learning classification model is trained on text and images, with an F1-score of 0.71 for Twitter and 0.9 on Gab.…”

TeamX@DravidianLangTech-ACL2022: A Comparative Analysis for Troll-Based Meme Classification

“…Prior work on detecting harmful aspects of memes include categorizing hateful memes (Kiela et al, 2020), antisemitism (Chandra et al, 2021) and propaganda detection techniques in memes (Dimitrov et al, 2021a), harmful memes and their target (Pramanick et al, 2021), identifying protected category such as race, sex that has been attacked (Zia et al, 2021), and identifying offensive content (Suryawanshi et al, 2020a). Among the studies most notable effort that streamlined the research work include shared tasks such as "Hateful Memes Challenge" (Kiela et al, 2020), detection of persuasion techniques (Dimitrov et al, 2021b) and troll meme classification (Suryawanshi and Chakravarthi, 2021).…”

“…Among the studies most notable effort that streamlined the research work include shared tasks such as "Hateful Memes Challenge" (Kiela et al, 2020), detection of persuasion techniques (Dimitrov et al, 2021b) and troll meme classification (Suryawanshi and Chakravarthi, 2021). The work by Chandra et al (2021) investigates antisemitism along with its types by addressing the tasks as binary and multi-class classification using pretrained transformers and CNN as modalityspecific encoders along with various multimodal fusion strategies. Dimitrov et al (2021a) developed a dataset with 22 propaganda techniques and investigates the different state-of-the-art pretrained models and demonstrate that joint vision-language models perform best.…”

Detecting Propaganda Techniques in Memes

“…Multimodal Content Previous work has explored the use of multimodal content for detecting misleading information , deception (Glenski et al, 2019), emotions and propaganda (Abd Kadir et al, 2016), hateful memes (Kiela et al, 2020;Lippe et al, 2020;Das et al, 2020), antisemitism (Chandra et al, 2021) and propaganda in images (Seo, 2014). proposed models for detecting misleading information using images and text.…”