A map view shows the clustered reactions
Our graphs offer more analytical views of the same data
Our news feed shows you the news and the emotional reactions to each article
Scraped Facebook data that is stored in Mongo DB and actually real
Visualizing the world as a better place using sentiment analysis from social media
We are at a point where news outlets aren't trusted and people are tired of the negativity of the news. We wanted to build a system that will show community members what their community is feeling about local events and news. We wanted community leaders to be able to see how their constituents are reacting to local events and to see the overall demeanor of the community so they can better guide their work.
What it does
Something Good takes data from social media regarding emotional responses to local news, it synthesizes that data and organizes it, then makes it available to our front-end visualizer interface real-time.
How we built it
We designed the system to be platform agnostic throughout the stack. Our backend scrapes Facebook pages of local news outlets for the community's emotions regarding specific journalism and gives a sentiment score to each article/post. It then sends that sentiment analyzed data to our middleware server that aggregates the data and organizes it into a Mongo DB, then it makes that available to our front end visualizer. Our visualizer then takes the data and makes it sortable and dynamically viewable.
Challenges we ran into
Facebook's Graph API doesn't allow for more detailed searches based on location like Twitter and Instagram.
Accomplishments that we're proud of
We overcame the Facebook API issue by scraping for local news outlets and finding their pages and corresponding ID's which allows us to use the Facebook API to collect more emotional response details.
We added an easter egg in homage to the old clippy assistant in Windows of old! His name is Earthy!. He's pretty enthusiastic about helping you out!
What we learned
What's next for Something Good
We'd like to expand the data set by location and cover at least the continental US. We would also like to enhance the aesthetic of some of the graphs.