The project addresses the serious issue of cyberbullying, recognizing its harmful consequences and the need for effective det ection and resolution methods. Cyberbullying [2], a form of online aggression, poses challenges that require specialized techniques to identify and mitigate. The primary goal of the project is to propose advanced cyberbullying detection models. These models aim to go beyond traditional approaches by incorp orating contextual, emotion, and sentiment features, recognizing the multi-dimensional nature of cyberbullying instances. This project is all about building an EDM using data from Twitter. These datasets undergo enhancements in terms of annotations. The EDM, along with lexicons, is utilized to extract emotions and sentiments from cyberbullying datasets, contributing to a more nuanced understanding of the emotional aspects of online interactions. Cyberbullying can be detected in part by appealing to people's emotions. This project shows how important emotions are for cyberbullying detection by using emotional cues to enhance detection algorithms. An extensive dataset tagged with emotions is now available for use in cyberbullying detection, thanks to this study. Academics can utilize this dataset to create a system for identifying cyberbullying based on emotions.Particularly in real-time applications, the project faces challenges due to dataset imbalances between cyberbullying and non-cyberbullying incidents. The goal is to develop detection models that effectively handle these imbalances, ensuring reliable performance across different scenarios. We aim to further enhance the performance of our model by exploring ensemble techniques, specific ally utilizing LSTM and LSTM + GRU models, which have demonstrated an impressive 99% accuracy.