XFakeDetector

Inspiration

Fake news is one of the facts of life on the internet and social media that have bad effects on individuals and communities, health and public safety, ethical behavior, and democracy. So, battelling it is important.

What it does

We wanted to make a model to detect fake news and provide insights about it to the users.

How we built it

We used a pre-trained transformer model (DistilBert) and add a classification layer to it to detect fake news. Moreover, we used SHAP to explainable our model and build visualizations based on this explanation.

Challenges we ran into

The imbalanced dataset was one of the challenges that we handled by Random oversampling, and the other challenge was handling the missing values that we dropped the samples with no text in them.

Accomplishments that we're proud of

We built a high-performance model that can detect fake news and make it explainable to help the user understand the reason for the prediction. So, our model is not a black box!

What we learned

This project raised our awareness about the importance of misinformation and how it can be harmful in many aspects. Moreover, this hackathon was a fantastic opportunity to learn the importance of teamwork in achieving our goals. As a first-time participant, I thoroughly enjoyed the experience.