AN INTELLIGENT MULTI-STAGE MODEL COMBINING MACHINE LEARNING AND FUZZY LOGIC FOR CYBER-HATE DETECTION

Abstract

Social media has completely changed how people communicate and exchange information throughout the world. But as these platforms have grown in popularity, cyber-hatred has spread, which is a serious issue that has drawn attention from scholars. Numerous machine learning and deep learning approaches, including Naive Bayes, Logistic Regression, Convolutional Neural Networks, and Recurrent Neural Networks, have been put out to address this problem. To differentiate one class from another, these techniques use a mathematical methodology. However, because sentiment-oriented data offers a more realistic depiction of how individuals read online communications, a more "critical thinking" viewpoint is required for correct categorisation. Two machine learning classifiers, Multinomial Naïve Bayes and Logistic Regression, were used to four online hate datasets in this work, which was based on a survey of the literature to investigate effective categorisation methods. To better comprehend the text in the datasets, the classifiers' output was enhanced by combining fuzzy logic with bio-inspired optimisation methods including particle swarm optimisation and genetic algorithms.