April 5, 2017 | Written by: Dan Cunnington and Dave Braines
Categorized: External Events | Innovation
Share this post:
How can technology help in the humanitarian response to natural disasters? At CrisisHack, our intrepid Emerging Technology team – the “Hursley Hackers” (Dan Cunnington, Adrian Lee, Sanaz Yeganefard, Rosie Lickorish, Perry Harwood, Richard Tomsett, and Dave Braines) alongside John Archer from Gather – set out to investigate this question. We had access to real datasets and scenarios and the opportunity to get insight from some really interesting people with practical humanitarian experiences in the field. The response to a humanitarian crisis can be complex and chaotic, and technology is just one small piece of the puzzle. We quickly realized that there’s a great deal of valuable knowledge in the communities affected by the crisis and access to this knowledge could be incredibly valuable.
So, in just over 48 hours we developed CrisiSMS, a crowd-sourced knowledge gathering application that enables NGOs to rapidly take advantage of existing open data sources, and to better understand the crisis from the perspective of the local community in real-time. By combining open data with information gathered via simple SMS text message, CrisiSMS allows aid workers to gather essential, crisis-specific knowledge even in regions with relatively low-tech infrastructure. The solution also enables guidance and knowledge to be shared out to the community using the same simple technology too.
In the scenarios given to us by CrisisHack, “Emerging Technology” has a very different meaning from our day-to-day experiences in our comfortable Emerging Technology lab in the peaceful Hampshire countryside. From conversations with John Archer, an expert in the humanitarian aid sector and co-founder of sanitation charity Gather, we learned that while mobile data communication can be very expensive in low-income countries, many people do have access to mobile phones and frequently communicate via SMS.
John also explained that poor communication between aid workers and communities in these areas can lead to a lack of understanding between the providers and receivers of aid. This inspired us to focus our solution on earning the trust of these communities so that the humanitarian workers could be better understood and trusted, and therefore be more useful to the local communities when disaster strikes.
Solution & Storyboard
Inspired by our discussions with John, we developed a solution that provides both long-term benefits to the local communities, and crucial support during a crisis. During day-to-day life, people can send us text messages to report infrastructure problems – anything from potholes to a broken toilet. These messages are simple SMS texts, with no additional capabilities required for the end user. We propose that service providers could provide a value-added service through which the identity (e.g. phone number) or the sender is anonymized and replaced instead with a location. This would be known to the users of the service and can be achieved by the provides through the cellphone mast infrastructure. This would be a service that the humanitarian organisations would pay the network providers for. We created a prototype that showed these SMS message conversations presented to aid workers through a web interface alongside a map with key data relating to the crisis. The humanitarian worker can use this interface to see incoming real-time information about what help is needed where through SMS messages from the local community. They can also use the interface to reply to messages – for example, to request more details, or ask for an update, and engage in conversation with the community members when appropriate.
Using the system in this way builds a mutually beneficial, trusting relationship with the local community to make rapid information gathering possible. When a crisis occurs, the aid worker uses our simple drag-and-drop interface to send a message to everyone who has subscribed to the service and is located within the affected area. This SMS message could be a request for specific information about the potential spread of a disease outbreak, or contain advice such as the location of their nearest treatment centre.
CrisiSMS tags incoming messages based on their content, allowing the aid workers to filter messages about specific topics. These topics can both be built into the system – for example, tailored to the remit of relevant Non-Governmental Organisation (NGO )– or learned as they emerge from the incoming texts. The learned insights could also be packaged and sold as useful knowledge models or services to third parties, providing an additional business model for the humanitarian sector in the future.
Technologies Behind the Scenes
We could achieve this in such a short time thanks to a handful of clever technologies and nifty APIs. On the front-end we used Leaflet.js, OpenStreetMap and Mapbox. We added data from the Humanitarian Data Exchange and provided a drag-and-drop interface to allow aid workers to add any data they required to the map.
In the back-end we used our own Controlled English (CE) asset to quickly create complex data models for storing all the information about, and correspondence with, the users. This model linked SMS conversations to geographical locations and classified their messages’ topics based on keywords built into the model. Emerging (and potentially unexpected) topics are learned by passing the messages to the IBM Watson Natural Language Understanding API which uses machine learning techniques and is ideal for this task. We used Twilio to manage our SMS communication, and connected all the different parts together using another Emerging Technology asset from our team: Node-RED. Our app runs in the cloud on IBM Bluemix and we were able to build and deploy within minutes, and use continuous deployment techniques to keep the team as productive as possible during the hack.
CrisisHack was a great experience for us and we loved the opportunity to meet and interact with such a wide range of people. The other teams all had great, and very varied, ideas and the whole event was a real buzz. From our perspective we learned a lot about the complex environments that humanitarian workers fine themselves in, and we managed to learn a whole load of new tech as well as using some of our existing assets very productively. We’d love to think that a system like CrisiSMS would add real value for community engagement before, during and after crises and applaud John and others in their ongoing efforts to help communities around the world.