Abusive Language Detection in Online User Content (ACL Digital Library)

In recent years, the topic of identifying abusive language in user-generated internet content has gained significant attention. Blacklists and regular expressions are used in the majority of contemporary commercial techniques, but they are insufficient to combat more nuanced, less crude forms of hate speech. In this study, we create a machine learning-based technique that outperforms a cutting-edge deep learning method for identifying hate speech in online user comments from two domains. Additionally, we create the first-ever corpus of user comments with abusive language annotations. Lastly, in order to improve our understanding of abusive language, we employ our detection technology to examine it over time and in various contexts.

https://dl.acm.org/doi/10.1145/2872427.2883062