Plaza

Inspiration

All three of us often find ourselves debating with others on various issues related to current events. We all want to be well-informed from a variety of sources, but also lead busy lives and don't have the time to read the necessary amount of articles. We decided that we wanted to aggregate this information so that we can stay informed on current events while mitigating the amount of bias.

What it does

Plaza turns news coverage from major outlets into a structured, chat-style thread. Our core experience is a group chat where outlets speak as personas, so users compare perspectives, see alignment vs. conflict, and read bias signals at a glance, with citations on hover. We cluster related articles, run stance and sentiment analysis per source, and compute a transparent bias score. We treat digital media as data using topic clustering, stance/bias scoring, and sentiment trends. Plaza helps people form evidence-backed opinions and build healthier news habits.

How we built it

Refer to the architecture outlined in our slideshow here: Presentation!

What's next for Plaza

Our next steps will focus on improving performance, search, and accessibility. On the performance side, we’ll implement multi-tier caching with an in-memory LRU system for requests, as well as caching conversation responses and article results to reduce latency. For search and filtering, we’ll work on creating stronger article clustering, adding a search bar for exploring relevant news topics, and enabling filtering by country, popularity, and more. Finally, to make the platform more engaging and digestible, we’ll integrate a more natural text-to-speech API, allowing group chats to be experienced more like a podcast where different sources can be heard speaking to each other.

Built With

• claude
• javascript
• news-api
• nltk
• python
• react