Safe Chat is a mobile chat application designed to provide a safer online environment by reducing instances of cyberbullying. It leverages a machine learning backend to filter out toxic messages, making online communication more secure and respectful.
- Toxicity Filtering: Uses a large language model-based filter to detect and reduce cyberbullying in real-time.
- Machine Learning Backend: The chat application is powered by an LSTM-based Recurrent Neural Network (RNN) model trained on binary classification tasks.
- Data Augmentation: Applied techniques to rebalance the training dataset from an imbalanced ratio of 1:4 to a more optimal ratio of 2:3, improving the model's performance.
- Mobile Application & API: The application is complemented by a robust API that facilitates seamless integration with other platforms.
- 92% Accuracy: The LSTM-based RNN model achieved an accuracy of 92% in detecting toxic messages.
- Research Publication: Published a research paper titled “Text-Based Cyberbullying Prevention using Toxicity Filtering Mobile Chat Application and API” at the 2022 IEEE International Conference on Data Science and Information System (ICDSIS).
- Research Paper Link - https://ieeexplore.ieee.org/document/9915958
