We wanted to pursue this project because we thought it would be interesting to see what differentiates left and right leaning parties and we wanted to help the public learn more about the articles they read whether it be on CNN or Breitbert News.

What it does

Check-It works to scrape, analyze, and display information regarding the political bias for each sentence in the article and the article in general.

How we built it

We worked together to build the overall app. Tuna was involved with the backend where he created the model and the flask server in Python. Sachit was involved with the frontend and the UI for highlighting the information from the backend to the frontend of the Chrome extension.

Challenges we ran into

On the backend, there were issues with speed and deployability of the model. We got around this by using a specifically created model rather than relying on pretrained models and transfer learning. Furthermore, we had troubles connecting both frontend and backend using the chrome Extension options available. On the frontend, we faced difficulty with making a good and adaptable UI to the different bias classifications.

Accomplishments that we're proud of

  • Making a model that gains state of the art accuracy without the deployability issues of using a large pretrained model.
  • Connecting the backend and the frontend
  • Making a good first time UI

What we learned

We learnt a lot about how to function as a team and how to manage different tasks together so that they cohesively work in union as demonstrated by Check-It.

What's next for Check It!

Our plan is to utilize the backend model to deploy as a Twitter API and also to continue developing the chrome Extension both for scalability and usage.

Share this project: