Inspiration

Humanitarian response relies on a complex system of multifaceted interactions and collaborations between organizations and governmental hierarchies. Effectively coordinating information can be challenging in the best of time. In the chaos of a crisis, mistakes can hold even more weight!
Having inaccurate or old data can waste resources and time, putting unserved communities at risk of further harm and fatalities. Organizations require a clear path to relevant actionable data, the geographic spread of collective needs, and evolving insights from field regarding supplies that are most needed by communities in a disaster zone.

What it does

Eirene--named after the greek goddess of peace--is a platform that uses a SMS based chat bot to allow users to collectively generate information of the highest priority need of resources in the aftermath of a disaster. These requests are then translated to an interactive and dynamic map, pinning the location of the request and allowing humanitarian organizations and government to analyze the demand for aid using various filters and statistics. Organizations can triage what supplies are needed in which areas to build a robust and efficient humanitarian response.

How we built it

Frontend:

We used React and Bootstrap to build our front end. This depends on our Node server that serves the webpage through the browser. We used Proto.io to prototype our UI. We added these UI elements to our React app using Bootstrap.

Backend:

The backend of this application is comprised of three servers running on our local and cloud machines. This was built using a combination of Node, Express and Vanilla JS. We used the Twillio API through a network tunneling server to direct the SMS notifications to our Express server. The Express server serves this data to our node server that hosts the React app through a REST API.

Cloud:

The node server is hosted on Heroku. This allows our webpage to be accessed through the Internet rather than just being a local application. It has git integration so we were able to push our changes to the cloud as we were building it.

Natural Language Processing:

Implementing the sklearn library, we built a Support Vector Classifier model to identify sentiment in user comments, evaluating the incoming text on 6 features. The SVC was trained using a pre-classified dataset of tweets with specific sentiments, yielding an accuracy score of 60 percent.

Challenges we ran into

The most challenging part of the building Eirene was the integration of all the different technologies we used. Integrating the backend to pipe data from our twillio SMS chatbot to the the user facing react web app provided challenges that required innovative hacking. Additionally, creating the servers to host our web app as well as the twillio chatbot was a feat of collaboration and triumphant ‘ah ha!’ moments. Since many of us were new to developing front end creating our user interface was a learning curve.

Accomplishments that we're proud of

We were able to build a dynamic application that runs on the cloud and train a Natural Language Processing model. The SMS notifications are able to update the webpage in real-time through our backend processing.

What we learned

We learned about the stunning complexities of intergovernmental and organizational communications and effort collaborations during a disaster, and the inefficiencies directly borne of said interactions.

What's next for Eirene

Moving forward, the platform will translate between languages, allowing various multinational organizations to work cohesively and collaborate to tackle large problems with increasing ease.

The advent of a task management feature will allow organizations and humanitarian workers to balance workflows and their allocate efforts to high priority targets - ensuring a coherent focus is maintained as the resources and responses are scaled up depending on situational need.

Ultimately integration with other disaster information platforms will stabilize and standardize interorganizational data - providing foundation for sophisticated filtering and statistics, bringing humanitarians a powerful tool to leverage the most accurate and efficient aid response.

Markkula Ethical Analysis Prize Response:

https://docs.google.com/document/d/1svdcD2YhcM81Hde-w9xwm3DJYD4VujCWtW2I0UZAyII/edit?usp=sharing

Link to Presentation:

https://www.canva.com/design/DAD0lub-xdc/hgWs3VNaMAT4esp8BEyiug/view?utm_content=DAD0lub-xdc&utm_campaign=designshare&utm_medium=link&utm_source=publishsharelink

Share this project:

Updates