A Review on Machine Learning Techniques for Identification of Abusive and Hateful Speech

Natural language processing has a large number of applications especially for semantic web. This study provides a comprehensive assessment of contemporary machine learning and deep learning algorithms for hate speech identification. Examples of such occurrences include hate speech, abusive language, threats, and derogatory remarks. Hate speech constitutes abuse that is not limited to a single gender; it affects all individuals. In the present context, comprehending the dynamic patterns (incidents, geographical predominance, demography, etc.) is essential for formulating ways to analyse hate speech activities. Social media platforms function as an information system that aggregates and categorises hate speech data from diverse sources, mostly users. This aggregated information is analysed to discern insightful patterns from vast quantities of social media data, which cannot be monitored continuously. Contextual interdependence across diverse lexicons in data will be essential for identifying hate speech. Current research reveals a scarcity of studies focused on hate speech detection concerning user behaviour. This study addresses hate speech as an online exponential issue aimed at damaging targeted individuals. Such incidents exacerbate social disparities and asymmetries by rendering online spaces unfriendly and inaccessible.