-
-
A fake news headline example showing a low truthiness score
-
A real news headline example showing a high truthiness score
Inspiration
Our team was inspired to build this tool because we believe in the importance of having an information-literate and educated society. Lack of information literacy has hindered societal progress in areas including public health, fighting climate change, public safety, and technological innovation.
What it does
Real? allows users to input news headlines to see whether the information is true or false.
How we built it
We used NLP to train the model based on article headlines from Onion and less fake from other news organizations, coupled with a React frontend to send the headline text to the backend flask API
Challenges we ran into
We initially wanted to build a Chrome extension but we ran into a number of issues with regard to permissions and limited functions allowed by the extension platform
Accomplishments that we're proud of
Training the NLP model based on 400,000 headlines from various news organizations, and the good UI design
What we learned
Complex Chrome extensions are difficult to create, especially when you can't import files or modules without going through additional hoops
What's next for Real?
The original idea was to create a Chrome extension that allows the user to highlight a certain text and the extension automatically produces a "truthiness" score; while that had some challenges, we believe with more time, we can find a solution. A more reach goal is to create a general API that allows a user to use our fake news predictor NLP model and have that model automatically train itself with more recent news headlines
Log in or sign up for Devpost to join the conversation.