Inspiration
Fake news is everywhere, and providers of content oftentimes don't even realize that they might be hosting fake news. We wanted to help!
What it does
This awesome flask project allows users to plug in the link to a news article and recieve info on how credible it is. It also has the functionality for content providers to access the API and make calls to see how their content holds up against our Multinomial Naive Bayes Classifiers!
How I built it
I slept < 3 hours, I barely ate, and it's completely hacked together. (As is the spirit of the event!) It's got a absurdly long dependencies list and was built on dreams and caffeine.
Challenges I ran into
There was what looked like an amazing dataset with over 9 million labeled data points, but the "clean" data was extremely messy and misclassified. It was also difficult reading things in in batches and the dataset was hosted on S3 which made downloading it a pain.
Accomplishments that I'm proud of
It works! It gives a credibility rating and the rating seems pretty dang accurate. Well, unless aliens exist and fake news sites are secretly real news sites.
What I learned
I learned that caffeine is more powerful than I expected, try to have a strong team, and how to work with some new libraries
What's next for TheBeet1
The goal is to create subscription plans that allow users to make API calls to give their articles a credibility rating. I want to add bias, hate speech, political speech, and a lot more! Another goal is to increase the accuracy of the dataset because the one being used was scraped last minute.
Log in or sign up for Devpost to join the conversation.