Shield Cortex

Inspiration

As we looked through the available challenges for Bitcamp, we quickly found that the Bloomberg challenge was the most appealing for our team, as we all wanted to see what difference we could make with the defense data. We quickly found a room and got to work brainstorming across several whiteboards and comparing the different ways we could use this data for the greater good. As we searched for problems that we could address using this data, we realized that clarity with the public was the most pressing issue for this data at the moment.

What it does

Shield Cortex starts by parsing through updated Defense Contract data straight from the DOD website and extracts the most important information straight to our website, where it will be charted on a world map for public viewing. This service helps keep a more informed and safe public by providing people with openly accessible information on DOD initiatives and how they are impacting the world. This will both help suppress the fears of people who formerly lacked exposure to this form of data, and provide opportunities for a more informed and democratic public to voice their opinions on US defense contracting. This can provide for not only a more informed America, but a more ethically progressive society across the globe.

How we built it

Backend development used a webscraper to scrape the DOD data which was then passed through our own Natural Language Processing Algorithm to extract key information from the contracts we needed to pass through our Google Maps interface. The Webscraper used Beautiful Soup and Selenium libraries and the Natural Language Processing model uses spaCy and implements a Named Entity Recognition model to extract key information. Front-end development worked with React to import the Google Maps interface and add interactive popups based on backend-passed data.

Challenges we ran into

Geolocation Data
npm vulnerabilities
Webscraping with Selenium and adjusting to dynamic updating of the website
Connecting the natural language model to Firebase

Accomplishments that we're proud of

Creating our own Natural Language Processing algorithm to extract key information from paragraph data
creating an interactive Google Maps UI to visualize data

What we learned

How to parse through data
Natural Language Processing
Web Scraping
Google Maps frontend development
How to share a repository with branches

What's next for Shield Cortex

We plan to continue growing by finding new ways to connect with activist groups and keep the public informed and happy with the way government is run. We will also continue to develop the interface with potentially other pages, as well as a scroll feature for different dates.