Decoding Digital Hostility: Examining and Addressing Hate Speech on Social Media Platforms

There is no denying that the widespread usage of social media and the sharing of information have greatly benefited humanity. However, a number of issues have also emerged as a result of this increase in online engagement, most notably the spread of hate speech. Recent research has addressed this growing problem on social media platforms by automatically detecting hate speech in a variety of datasets using a variety of feature engineering approaches and machine/deep learning algorithms. It is interesting, though, that many of these studies—to the best of our knowledge—resort to identifying hate speech messages using traditional feature engineering techniques, which leads to less-than-ideal classification results. This is explained by the shortcomings of the feature engineering techniques now in use, specifically their vulnerability to the word order and word context problems. Therefore, more advanced strategies are desperately needed to address these issues and improve the precision of hate speech identification on social media platforms. To create fundamental lexical benchmarks is the goal. As distinguishing characteristics, our methodology makes use of the power of n-grams, which cover both character and word levels, and skip-grams, which cover both character and word levels. Notably, we successfully identify posts in all three defined categories with an impressive 78% accuracy. The results highlight how difficult it is to discriminate between vulgarities and hate speech. We also conduct a thorough exploration of potential directions for future research projects.