Sentinel is a language model aimed at detecting fraudulent posts and users across various social media platforms/mediums, including YouTube, Twitter, Instagram, and SMS. It uses a transformer architecture, and the BERT pretrained model for NLP, with data preparation and cleaning facilitated by Apache Spark. The application's interface is built with Python Dash, enabling straightforward user interactions.
- Fraud Detection: Detects potential fraudulent activities in posts and user behaviors.
- Transformer Architecture with BERT: Utilizes BERT for advanced language processing.
- Data Processing with Apache Spark: Streamlines data cleaning and preparation.
- Cross-Platform Compatibility: Analyzes data from multiple sources.
- Python Dash Interface: Offers a user-friendly front end.
- Python 3.6 or newer
- Apache Spark
- Pretrained BERT model
- Google Colab for model training
- Python Dash