A Text Classification Approach for Detecting Cyberbullying Risk on Twitter Using Support Vector Machine with Naive Bayes and Random Forest Comparison

Abstract

The rapid development of social media as a means of digital interaction also presents serious challenges in the form of the spread of negative content, including cyberbullying. Cyberbullying is a form of verbal violence committed online and has a significant impact on mental health, especially in adolescents. This research aims to develop a text classification model to detect the risk of cyberbullying using the Support Vector Machine (SVM) algorithm. The data used comes from a collection of cyberbullying-themed tweets. The research stages include text preprocessing (normalization, cleaning, tokenization, stopword removal, and stemming), feature extraction using Term Frequency-Inverse Document Frequency (TF-IDF), data division into training and testing sets, and model training using linear kernel of SVM. The model was evaluated using accuracy, precision, recall, and F1-score metrics. The results show that this approach is able to identify risky comments quite accurately, with optimal performance on the linear kernel. This research contributes to the development of automated detection systems to create a safer and healthier digital ecosystem, and supports preventive efforts in mitigating cyberbullying online.