Ensuring safety in digital spaces: Detecting code-mixed hate speech in social media posts (Data & Knowledge Engineering)

With a performance of 75% accuracy, the results show that the transliteration strategy outperforms both raw and translated data by 1% and 3%, respectively. By successfully reducing hate speech and inflammatory information on social media sites, the suggested solution improves user experience. Numerous advantages are presented by this research from a managerial standpoint, including better content filtering, more efficient use of resources, data-driven decision-making, increased user happiness, improved reputation management, and increased scalability. These developments highlight how cutting-edge tools may be used to tackle difficult social media management problems.

https://www.sciencedirect.com/science/article/abs/pii/S0169023X25000047