Fake News Detection

Team members

• Yuanchu Dang - Bert model
• Yanmin Ji - LSTM model
• Wei Luo - Demo using multithreaded Tweepy streaming and Flask

Previous work & Acknowledgement

• We forked and built on this fake news challenge repo.
• This NLP repo, particularly the Bert wrapper it provides.
• Our demo is based on this twitter-flask multithreading repo.

Problem Formulation

Given a headline and a body concerning a subject, determine if their views agree or disagree with each other, or, are simply unrelated.

Approaches

Running instruction

To run the demo, you need to first fill in the dict that contains API keys with your actual keys. Also, in the main function, you can specify which port you want Flask to use. Then to run the application, you can simply run

python3 demo.py

Then you can go to

http://[Your_External_IP]:[Port]

to access the demo webpage.