Detecting weak and strong Islamophobic hate speech on social media

TL;DR

This paper develops a multi-class classifier to detect and differentiate between non-Islamophobic, weak, and strong Islamophobic hate speech on social media, using a large dataset of tweets and contextual word embeddings.

Contribution

It introduces a novel multi-class approach for Islamophobic hate speech detection that captures varying severity levels, outperforming traditional binary methods.

Findings

Accuracy of 77.6% and balanced accuracy of 83%

Weak Islamophobia is more prevalent than strong Islamophobia in the dataset

A multi-class SVM outperforms deep learning models in this task

Abstract

Islamophobic hate speech on social media inflicts considerable harm on both targeted individuals and wider society, and also risks reputational damage for the host platforms. Accordingly, there is a pressing need for robust tools to detect and classify Islamophobic hate speech at scale. Previous research has largely approached the detection of Islamophobic hate speech on social media as a binary task. However, the varied nature of Islamophobia means that this is often inappropriate for both theoretically-informed social science and effectively monitoring social media. Drawing on in-depth conceptual work we build a multi-class classifier which distinguishes between non-Islamophobic, weak Islamophobic and strong Islamophobic content. Accuracy is 77.6% and balanced accuracy is 83%. We apply the classifier to a dataset of 109,488 tweets produced by far right Twitter accounts during 2017. Whilst most tweets are not Islamophobic, weak Islamophobia is considerably more prevalent (36,963 tweets) than strong (14,895 tweets). Our main input feature is a gloVe word embeddings model trained on a newly collected corpus of 140 million tweets. It outperforms a generic word embeddings model by 5.9 percentage points, demonstrating the importan4ce of context. Unexpectedly, we also find that a one-against-one multi class SVM outperforms a deep learning algorithm.