Identifying code-mixed and code-switched hateful remarks on social media using NLP

Abstract

Online bullying has prevailed for years in the vast cesspool that is commonly known as the online social media. Increasing use of social media and online communication has led to a rise in cyberbullying– which is often facilitated by the abundant usage of code-mixing and code-switching. Research has been done to filter out these derogatory remarks. However, little research has been done on code-switched and code-mixed hateful remarks. English has blended into our Bangla language so effectively that people regularly use English letters to convey Bangla due to its convenience. English and Bangla are used interchangeably in regular conversations as well. Our main objective in this research is to detect these code-switched and code-mixed remarks– which we plan to do by taking advantage of the state-of-theart natural language processing technologies.