The most important part in any quality conversation is knowledge. Knowledge is what ignites conversation and drive - knowledge is the spark that gets people on their feet to take the first step to change. While we live in a time where we are spoiled by the abundance of accessible information, trying to keep up and consume information from a multitude of sources can give you information indigestion: it can be confusing to extract the most relevant points of a new story.

What it does

Macaron is a service that allows you to keep track of all the relevant events that happen in the world without combing through a long news feed. When a major event happens in the world, news outlets write articles. Articles are aggregated from multiple sources and uses NLP to condense the information, classify the summary into a topic, extracts some keywords, then presents it to the user in a digestible, bite-sized info page.

How we built it

Macaron also goes through various social media platforms (twitter at the moment) to perform sentiment analysis to see what the public opinion is on the issue: displayed by the sentiment bar on every event card! We used a lot of Google Cloud Platform to help publish our app.

What we learned

Macaron also finds the most relevant charities for an event (if applicable) and makes donating to it a super simple process. We think that by adding an easy call-to-action button on an article informing you about an event itself, we'll lower the barrier to everyday charity for the busy modern person.

Our front end was built on NextJS, with a neumorphism inspired design incorporating usable and contemporary UI/UX design.

We used the Tweepy library to scrape twitter for tweets relating to an event, then used NLTK's vader to perform sentiment analysis on each tweet to build a ratio of positive to negative tweets surrounding an event.

We also used MonkeyLearn's API to summarize text, extract keywords and classify the aggregated articles into a topic (Health, Society, Sports etc..) The scripts were all written in python.

The process was super challenging as the scope of our project was way bigger than we anticipated! Between getting rate limited by twitter and the script not running fast enough, we did hit a lot of roadbumps and had to make quick decisions to cut out the elements of the project we didn't or couldn't implement in time.

Overall, however, the experience was really rewarding and we had a lot of fun moving fast and breaking stuff in our 24 hours!

Share this project: