Enhancing Online Safety: Cyberbullying Detection with Random Forest Classification
M. Purnachandra Rao, Nikitha Kota, Devamsakhi Nidumukkala, Meghana Madoori, Danish Ali · 2024
With regard to the growing issue of cyberbullying in virtual communities, this study introduces Forest Shield, a strong detection system based on Random Forest categorization. Forest Shield is made to detect instances of cyberbullying in a variety of social media settings. Several traditional machine learning techniques have been employed in the past to automatically identify cyberbullying on social media. These models, however, do not account for every possible demand. Machine learning can be used to develop a model that will automatically recognize cases of cyberbullying by identifying the linguistic patterns used by the bullies. The goal of this project is to detect and prevent cyberbullying through the use of supervised machine learning. Using a heterogeneous dataset collected from online sources, we perform preprocessing and extract pertinent features from the text, including linguistic clues and contextual data suggestive of cyberbullying behavior. We test Safeguard's ability to discriminate between cyberbullying and non-cyberbullying cases using a rigorous experimentation process, and we obtain excellent accuracy, precision, recall, and F1-score metrics. Additionally, we run in-depth investigations to investigate how various feature sets and parameter setups affect detection performance. With this work, we make a substantial contribution to the field of cyberbullying detection research by providing methods and insights that are essential for preventing harassment in online communities.