Hate-Speech Detection in News Articles: In the Context of West Bengal Assembly Election 2021

Abstract

A hate speech represents an expression or phrase of offensive language. The intention behind the usage of hate speech is to abuse, dehumanize, disparage, or harass a person or a group of persons based on their race, color, gender, religion, caste, ethnicity, disability, language, belief, nationality, or other factors. Hate speech is also used to express violence, harm, or hatred against the targeted people. Nowadays, the amount of hate speech is overgrowing in social media, online newspapers, etc. It becomes very difficult to moderate the data containing hate speeches manually, as it is a tedious and time-consuming task. Therefore, an automated hate-speech detection technique is very essential. Numerous works have been done to develop the technique or tool for automatic detection of hate speeches in Twitter, Facebook, and other social media data. This paper studies hate speeches on political news articles in the context of the West Bengal Legislative Assembly Election 2021. In the computational aspect, this task is carried out in three phases. First, a political news corpus has been created, and next from that corpus a key word/phrase-based hate-speech identifier is developed. A semi-automated approach has been used to find out the set of hate-speech-related key words/phrases. Finally, the system’s performance is evaluated, and the results are investigated.