Predicting Cyberbullying Types in Tweets Using Multi-Class and Binary Classification

In the digital era, the rise of social media platforms has brought about numerous benefits, but it has also contributed to the growing issue of cyberbullying. This project, titled "Predicting Cyberbullying Types in Tweets Using Multi-Class and Binary Classification," leverages machine learning techniques to tackle the issue of online harassment by developing automated systems for detecting and classifying cyberbullying in tweets. The project uses a publicly available dataset containing over 47,000 tweets, categorized into classes based on different types of cyberbullying: Age, Ethnicity, Gender, Religion, and Other. Additionally, a binary classification model is developed to flag harmful tweets. The models implemented include traditional machine learning algorithms such as SVM, Random Forest, XGBoost, and Gradient Boosting, alongside advanced deep learning techniques like BiLSTM, to achieve high accuracy in predicting cyberbullying instances. Preprocessing techniques such as TF-IDF are applied to convert textual data into usable features. The project aims to provide a web-based solution that can automatically classify tweets, helping to create a safer digital environment by identifying harmful content. The results of the classification models are evaluated using various performance metrics, including accuracy, precision, recall, and F1-score, ensuring a comprehensive understanding of the model's effectiveness. By offering automated classification of tweets, this project contributes to online safety and aims to mitigate the impact of cyberbullying across social media platforms.

Keywords: cyberbullying, classification, machine learning, tweets, prediction, SVM, Random Forest, XGBoost, BiLSTM, online safety.