Emergency services and resource management for disaster management. Disaster Response is a nuanced and complex one, where multiple resources must be carefully coordinated and managed. Humanitarian Initiative on giving human aid are working towards establishing standard protocols and procedures and accumulating them in the Sphere guideline. However despite having such thorough and detailed guidelines, the resources are not managed to the best of our capabilities to minimize the impact of natural disasters on humanity. The idea struck us when we were pondering over the facebook utility of marking one as safe in a crisis.
What it does
We provide a platform to facilitate for better management of resources for medical, surgical, infrastructure and basic necessity care. This platform utilizes the response of users response to crisis management on social media (facebook and twitter) to find and prioritize the most affected buildings or micro-locations in a disaster hit area using geolocation tags on map. Also, members who have marked themselves as safe in the affected area can be the first point of contact to get any feedback(if needed) and the nearest people to help(if they can) to provide speediest help.
How we built it
- We built the platform using angular, HTML, CSS and ES6 to display the current disasters being faced with the type of disaster and geographical area.
- Selecting one value displays the geo tags of members who have marked themselves safe in/ near that area and a list of locations to be visited in decreasing priority. This is displayed using angular maps which uses google maps api.
- Monitor locations using socket connections to get Real time data from social media.
- For the purpose of hackathon, (since we were not aware of any disaster hitting during the course of hackathon) we used Existing Data of users who has marked themselves safe during the natural disasters hit in detroit and pheonix.
- The dataset is cleaned and mined using python and inserted in firebase database.
- Live data from point 3 are filtered using Google Cloud Natural Language Processing API to determine the Sentiment and inserted into the database.
- The values from database are mapped as tags on the maps using angular maps which inturn uses google maps api.
Challenges we ran into
Finding relevant data was difficult. facebook does not expose the crisis management api or dataset of people marked safe. Thus we found a relavant dataset and assumed we are able to retrieve this using facebook/twitter. We monitored live feeds for real-time data.
Accomplishments that we're proud of
Socket programming to get real time data from twitter,
What we learned
Disaster Management procedures and responses. Integrating with google cloud platform (Firebase, OAuth, NLP and Node.js) Sentiment Analysis Crisis Management Service of facebook and twitter.
What's next for Safely
The algorithm can be improved using gaussian mixture models for identifying both safe and unsafe areas. Heatmaps to be used to better depict the mixture model and impact.