Check out this new article by Geetanjali and Mohit Kumar, which provides an in-depth review of current research in hate speech detection. The study analyses the latest methods, available datasets, and ongoing challenges, offering insights for researchers and policymakers working to combat online hate.
The aims of the study were to review existing hate speech detection methods and their limitations, to analyse published research over the last ten years to identify key trends and gaps, and to provide guidance for future research and policy development.
The authors conducted a systematic review of existing literature, evaluating various approaches to hate speech detection, including machine learning and deep learning models. They examined multiple datasets, highlighting their strengths and weaknesses. The study also categorised different types of hate speech and explored the role of linguistic nuances, sarcasm, and cultural context in classification models.
The main findings include:
The article’s suggestions for future applications are that policymakers should invest in updating and expanding hate speech datasets to include diverse languages and cultural contexts. Effective hate speech detection requires cooperation between data scientists, linguists, and legal experts to refine classification models. Regulators must establish guidelines to improve the interpretability of machine learning models, ensuring their decisions are transparent and accountable. Social media companies should integrate adaptive hate speech detection systems that continuously evolve to identify emerging patterns of online abuse.
This study highlights the complexity of hate speech detection and calls for continuous research and innovation to create safer online spaces.