Inspiration

Inspired to fight misinformation

What it does

Leverage the power of Machine Learning to predict if the title of the news that we are reading are real or may be fake.

How we built it

We used a logistic regression classifier from Sklearn to create our model. We trained it by using the data from the "Fake and real news dataset" hosted in Kaggle.

Challenges we ran into

We tried to save our sklearn model using ONNX and then loading it with onnx.js library in our extension javascript code, but we encountered some data types incompatibilities in the process. To solve this, we created a very simple flask server in which we hosted our classifier and made an endpoint to interact with it.

Accomplishments that we're proud of

We were able to create an end-to-end working product in just 24 hours!

What we learned

In such short period of time to develop an application, you do not always have time to solve all the problems that may arise, therefore thinking a quick alternative is very useful.

What's next for Fake News Extension

We are aware that using just the titles of the news is not enough for detecting if a news is real or fake, because even for humans it is a hard task. In further work, we are planning to get the entire text of the news and update our machine learning classifier to make better predictions.

Built With

Share this project:

Updates