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

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