Fake News Detection using BERT

This is a repository for a university class project about fake news detection using basic data science methods.
It explores NLP Processing in 4 Python Notebooks and trains multiple BERT Models as well as some basic binary classification models on the dataset.
The repo hosts also a Streamlit app, which lets you explore the data as well as play around with different concepts of BERT.

Run it locally

You can run it using the "UV" Package Manager.
Just install uv using pip install uv.

Then run uv sync to create a virtual environment with all the required packages.
Use either uv run or source .venv/bin/activate to launch the virtual environment.
Don't forget to select the venv as your interpreter.

You should now see (fakenews-detection) in your terminal in front of every line.
The interpreter should be Python 3.10.

View the Streamlit App

This Project uses a Streamlit application to visual its findings.

The project is hosted on the Streamlit community cloud here.
You can also run it locally by going to the streamlit_app directory and running streamlit run FakeNewsDetection.py.

Note

This Repo will stop active development after the 05/02/2025.