Fake News Detection Project
Welcome to our Fake News Detection Project, where we embark on a journey to build a robust system that can determine the authenticity of news articles. In a world inundated with information, the ability to distinguish between real and fake news is crucial.
Real Project from the ground up
We'll begin by creating a solid development environment, leverage cutting-edge Natural Language Processing (NLP) techniques to convert news articles into meaningful embeddings, and train a machine learning model capable of making accurate predictions.
Throughout this project, we will embrace best practices in software development, version control, and containerization to ensure reproducibility and ease of deployment. Our goal is to not only develop a powerful tool but also to learn and grow as we build a project from the ground up. Let's get started on this exciting journey towards a more informed and discerning world!
Project Overview
- Create a virtual environment
- Start a new project using Poetry
- Manage the repository using GIT
- Download dataset from Kaggle
- Docker-Compose: Create environment with MLFlow
- Use a sentence transformers model to transform news to vectors.
- Retrieve the embeddings and Train a single model on those embeddings and register to MLFlow
- Docker-Compose: Add a service that takes the latest model from MLFlow and crates an API
- Create a Streamlit App that calls the API and returns a response
- Dockerize Streamlit App and add it to the docker compose