Analysis and Design of a Robust Decision Support System for Hate Speech Detection in Social Media

Abstract

Social media has revolutionized global interaction and communication, providing a platform for users to share a wide range of content, from excellent to abusive and spammy. Unfortunately, this has also led to a proliferation of online hate speech, which aims to incite hatred based on protected characteristics like race, religion, ethnicity, or national origin. Numerous studies have highlighted a surge in violence associated with online hate speech on a global scale. Consequently, societies worldwide are balancing free speech and censorship on popular social media platforms. As the volume of such content increases, the need to identify and address hate speech and racist language on social media sites becomes increasingly crucial. newlineThis thesis introduces some self-solution-based deep-learning approaches to design a robust decision support system for hate speech detection in social media. Four research objectives have been examined to investigate a generalized data model for multiple social media platforms, design a Hate Detector a recursive system for alert generation for preventive measure, implement and analyze the obtained results of Hate Detector with various experiment, and optimize the alert of Hate Detector for robust Decision Support System. The significant contributions of the thesis are as follows. The thesis addresses the challenges associated with hate speech detection, specifically classifying English Tweets into two categories: 1) Hate Speech and 2) Neither. Detecting hate speech manually, especially within social media data, is complex. There is a need for a robust mechanism capable of automatically identifying hate speech on social networks. Many existing methods need to achieve accurate detection due to challenges such as the massive volume of data, data dependencies, excessive parameters, and reliance on homogeneous social media data. The thesis examines a cross-platform hate speech recognition mechanism for social media interactions to overcome these issues.