What it does


After Hurricane Harvey left so many people, families, and pets stranded on rooftops and trapped in their homes, the coast guard and police were overwhelmed by rescue requests on social networks like Twitter. They urged the people still stranded to not post their information online, but to instead call 911 until the police became available to take down their information. The problem is that since the hurricane left the majority of people its wake without power, very few people had the battery power to make that emergency call after waiting days. With hurricane Irma barreling towards Florida, we decided to try to help mitigate this problem to help emergency response and good Samaritans reach their trapped neighbors faster.

We built a system to help people affected by the hurricane reach out for help and also A few examples: https://twitter.com/sparklingLily/status/903722507982839808

Beacon provides three ways for people affected by natural disasters to beam out their locations. The first way is through the native app, where users can fill out a form. Additionally, Beacon also provides a text-sms service that users can report locations to. This platform may be best for people who are affected and live in areas of low connectivity. Finally, Beacon mines Twitter, searching for tweets of distress and extracting locations from them. Using natural language processing, we look through each tweet to find locations of people who still need help. All three of Beacon’s data gathering methods feed locations to a heat map in the cloud, accessible through our app, so that emergency responders and volunteers can see where people still need help.

How we built it

We used java to build an android app, and the google-maps-api to help display the live map. We used nexmo to set up our text-sms service, and flask to set up our online server. We wrote some of our own natural language processing algorithms as well as google-cloud-computing's natural language processing service.

Challenges we ran into

Filtering natural language processing data was difficult because of how much of it there was. Stitching every part together was also a challenge because of how many moving parts to the project there were.

Accomplishments that we're proud of

We had to learn how to use web hooks to receive SMS messages.

Share this project: