Exploring Bioinspired Feature Engineering Technique for Online Hate Speech Detection

Abstract

The spreading of hate speech and toxicity on social media and other online platforms has increased severely in the past decade. In the current scenario also, when the whole world is suffering with outspread of COVID-19 online hate speech spreading more than before. The spread of such hate can jeopardize the mental and physical health of many people and is thus necessary to stop its spread on online social media. This paper aims to explore bioinspired algorithms like PSO and GA to detect online hate speech on social media and other online platforms. We explore the hybrid feature selection approach to select valuable and meaningful features from the hate speech dataset to classify between hate and not hate posts efficiently. Our experiments indicate the random behavior of Particle Swarm Optimization and Genetic Algorithm and the decrease in accuracy when applied individually to the experiments. The proposed hybrid approach gives the comparative results as TF-IDF when applied with the baseline machine learning models.