Deviant Racist Behaviour

Research on deviant racist behavior focuses on automatically detecting and classifying racist language in digital text, aiming to understand its diverse forms and mitigate its spread. Current efforts employ deep learning models, such as RNNs, LSTMs, and BERT-based architectures, often within ensemble frameworks, to analyze large corpora and identify subtle manifestations of racism including stigmatization, offensiveness, blame, and exclusion. However, a critical challenge highlighted by recent studies is the inherent bias in annotation processes, stemming from annotator beliefs and identities, underscoring the need for more robust and contextually aware detection methods. This research has significant implications for understanding and combating online hate speech and improving the fairness and accuracy of automated content moderation systems.