International Research Journal of Education and Technology

The rapid expansion of the internet has turned social media into a dominant platform for communication. However, the rise in online interactions has also led to an increase in cyber bullying incidents, which can result in severe mental distress, physical health issues, and, in extreme cases, suicide attempts, particularly among women and children. The main goal is to develop and deploy an efficient machine learning model for detecting cyberbullying content. The proposed system utilizes a dataset comprising text messages labelled as either "normal" or "cyberbullying." This dataset, sourced from the Kaggle repository, includes structured information divided into training and testing subsets. The methodology begins with the data upload process, followed by essential pre-processing steps such as data cleaning, dimensionality reduction, and normalization. A decision tree algorithm is applied to develop the classification model by training it on the given dataset. The effectiveness and accuracy of the model are subsequently evaluated using the test dataset. The system successfully classifies text as either normal or indicative of cyber bullying. Early detection of such harmful content can significantly enhance user safety on social media platforms, fostering a more positive and engaging online environment.

Keywords—Cyberbullying detection, Machine learning, Social media, Decision tree algorithm.