- RNN model in tensorflow
- LSTM model in tensorflow
- Bert-base model after fine-tuning
- Python script for building web app
The focus of this project is to develop a news authenticity detection system. Models including RNN and LSTM were built using TensorFlow to analyze and classify the authenticity of news articles. Another model was fine-tuned from a pre-trained BERT model in Pythorch. The dataset consists of more than 22,000+ news articles that underwent preprocessing steps such as removing stop words, tokenization, vectorization, and sequence padding/truncation. The models were then deployed as a user-friendly web app using Gradio, allowing users to input news articles and receive real-time authenticity predictions. By combining state-of-the-art modeling, meticulous data processing, and seamless deployment, this project aims to provide an efficient tool for identifying the authenticity of news articles.