Politeracy

Politeracy promotes political literacy. It finds and summarizes relevant new articles from different political leanings for you and summarizes them for a quick read so you can get the full picture.

Comment

Inspiration

We realized how difficult it is to be politically informed on both ends of the spectrum, so we wanted to increase the ease of accessibility of that for other people.

What it does

Given a user input on any topic, Politeracy scrapes three news websites -- CNN, Associated Press, and Forbes -- and, using Cohere's API, summarizes the most relevant articles pertaining to the topic, giving a clear idea on how the left, center, and right view the issue.

How we built it

  • Problem: People are often too busy, or have too short of attention spans to pay attention to the news, and are quick to accept bias.
  • Solution: Highlight bias, summarize news, and allow people to form opinions off such summaries.
  • Design: We decided to make a website and split up responsibilities, with Cyrus in charge of making the website and finetuning Cohere, and Alyssa aggregating news sources and building a web scraper.

Challenges we ran into

Alyssa: This was my first time building a project outside of my courses, so my setup was not prepared in the slightest. I ran into trouble even installing packages and eventually had to reinstall Python itself! Aside from that, this was my first time working with anything pertaining to the backend -- APIs and web scraping, while sounding cool, were completely new to me! I kept having to go through trial and error with web scraping, scraping myself from a myriad of websites only to finally stumble upon a way I could understand.

Cyrus: I didn't have experience with NLP, so I wasn't sure how to tune the hyperparameters or how it works. That took some time to learn through both tutorials and trail and error. I also ran into a lot of problems with getting the backend to integrate well with the frontend, as I have very little experience with the requests library.

Accomplishments that we're proud of

Alyssa: As this is one of the first apps I've made, I'm especially proud of the way it turned out! Going through the design process of building an app from scratch, and learning all the tools necessary to bring it together was definitely a rewarding experience.

Cyrus: I'm very proud that I've made my first forays into NLP, a field I've always been interested in but was too intimidated to start. It was immensely satisfying to learn about it, try it out, and use it to make a product that's more than just the sum of its parts.

What we learned

Alyssa: I definitely learned a lot about working with APIs and web scraping. I also learned a lot about how to integrate all the different parts of an app together!

Cyrus: Learned how to tune NLP models, and more experience on developing a good frontend interface and how to request data from our backend.

What's next for Politeracy

We hope to expand to more news sources, offer more complex stances, and use CoHere's classification endpoints to classify left-right bias in news sources so that users can understand the different political standpoints better. We'd also expand the UI, and allow users to include/exclude specific news sources, choose how many articles they'd like to see, and adjust the length of the summaries.

Share this project:

Updates