What it does

This google chrome extension checks if an online article has traces that might indicate it is a fake news.

How we built it

We coded the majority of our logic in python, using a large dataset and cosine similarity to find intersections among different articles. We then coded the extension using javascript that connects the article to the python code that runs locally. The little notification window was written in Java.

Challenges we ran into

The first challenge was finding the right calculation method to use. We decided to use cosine similarity, thanks to Hyunjoon's great idea, and adapted it to fit our purpose.

Another difficulty of coding this was using a language that we have never used before: Javascript. Xinyue and Bu Geun did this part: using Javascript to make an extension.

Accomplishments that we're proud of

We ran different test cases, using actual articles that we have found on various websites, such as New York Times, National Geographic, Fox News, and Politico. Although our test script achieved only 60% accuracy, we were able to distinguish most of the fake news when field testing with the extension.

What we learned

We learned how to make a Chrome extension. We learned that combining different programming languages is difficult, but possible and sometimes necessary. We've also learned the limits of computers in the fields of text analysis. Lastly, we learned the amazing things we can do by combining software engineering with data science.

What's next for FakeNewsDetector

We have a few bugs to fix. We also have to increase the accuracy of our detector. We can also improve GUI.

Built With

Share this project:
×

Updates