Cyberbullying Detection Based on Sinhala Language Context on Social Media

Abstract

"Cyberbullying has become a pervasive issue on social networking sites, leading to severe negative impacts on victims. Despite its prevalence, a comprehensive technological solution to address this global problem remains elusive. In response, this study proposes an automated approach to identify cyberbullying in Sinhala language text, including emojis, using a deep learning approach. The primary objective is to provide a means for relevant parties to take preemptive action against cyberbullying instances before they escalate. Utilizing a Convolutional Neural Network (CNN), the technical solution involves training the model on a dataset consisting of Sinhala language text extracted from social media platforms. The CNN architecture comprises multiple layers, including convolutional and pooling layers, designed to extract relevant features from the text data and classify it based on hate level. Evaluation of the proposed approach yielded promising results, with the CNN achieving an accuracy rate of 79%. This performance metric indicates the model's effectiveness in accurately identifying instances of cyberbullying in Sinhala language text, thereby offering a potential solution to mitigate its harmful effects in society."