The problem Caustic News solves

This webapp gives the theorhetical probability of a news article being fake. This can help people make an educated descision for whether the news should or shouldn't be trusted.

Challenges we ran into

The dataset we were working with was quite small, which is why the application of this app is limited to articles similar to which it was trained. The model fails to capture the essence of all the articles on the web. We overcame this by using some NLP techniques to clean the data and reducing the variance.

We weren't able to properly deploy the backend on herokuapp.com so we also tried it on netlify.