Social media has become an integral part of our lives. It gives us the freedom to express ourselves and to communicate with people around the globe. But currently, the platform is being exploited for cyberbullying and personal harassment. Because of the increasing expansion of social media and its integration into ordinary living, cyberbullying has become extremely common. Being the victim of a cyberbully could have a severe emotional and psychological impact on an individual. It can make anyone feel vulnerable and exploited. One of the main challenges faced by cyberbullying detection is the lack of labeled data. Keeping this in mind, a Semi-supervised learning model is proposed to detect and prevent cyberbullying on social media platforms. This model uses partially labeled training data, with a small amount of labeled data and a larger amount of unlabeled data.

Keywords : Cyberbullying; Label Propagation; Natural Language Toolkit (NLTK); Semi-supervised learning; Social media; Sentiment analysis; Support Vector Machine (SVM