Fake News Detection System

This project is a complete solution for detecting fake news using machine learning. It includes a trained model, a Flask API, a user interface for predictions, and a MySQL database for logging all requests. The entire project is dockerized for seamless deployment.

---

Features

• Machine Learning Model: Trained using a dataset to classify news as real or fake.
• Flask API: Serves predictions via an endpoint.
• Web Interface: A simple UI for users to input news and view predictions.
• Database Logging: Tracks all predictions in a MySQL database.
• Dockerized: Easy deployment using docker-compose.

---

Requirements

• Docker
• Docker Compose

---

Quick Start

1. Clone this repository:

   git clone <repository-url>
   cd <project-directory>

2. Run the application:

   docker-compose up

3. Access the application:

   • Web Interface: http://localhost:5000
   • API Endpoint: http://localhost:5000/predict

---

Project Structure

• notebooks/: Contains the notebook used to train the model.
• model/: Stores the pickled model (model.pkl).
• api/: Flask API for serving predictions.
• ui/: HTML and CSS for the user interface.
• db/: Database schema and MySQL configuration.
• docker-compose.yml: Orchestrates the services.
• Dockerfile: Configures the Docker image.

---

API Usage

Endpoint: /predict

• Method: POST
• Payload:

  {
    "news": "Enter the news article text here"
  }

• Response:

  {
    "prediction": "real" or "fake"
  }

---

Database

• Tracks Requests: Logs predictions with timestamps.
• Database: MySQL
• Schema:
  • id: Auto-increment primary key
  • news: Text of the news article
  • prediction: Predicted label (real/fake)
  • timestamp: Time of the request

---

Deployment

This project is fully containerized. To deploy:

• Clone the repo.
• Run docker-compose up.

---