Shaping the future: Our strategy for research and innovation in humanitarian response.
Christophe Billen, Founder of People’s Intelligence, talks about the successes, challenges and next steps for their HIF-supported project to automate the collection and verification of crowdsourced information to improve humanitarian interventions.
The project ‘People’s Intelligence’ aims at building technology to improve affected populations’ ability to report relevant information in a format that will increase humanitarian organisations’ response efficiency and where possible provide affected populations access to potentially life-saving information.
We believe PI’s added value compared to existing solutions is the automation of a series of processes to:
(1) Increase affected populations’ ability to report structured information – quality data for example their location, type of incident suffered, number of affected persons and source of knowledge – relevant to humanitarian organisations;
(2) Improve organisations’ capacity to separate credible information from rumours and disinformation and identify reliable sources of information over time;
(3) Verify information that could not be triangulated through networks of trusted sources and where permissible, the crowd;
(4) Provide, when possible and available, potentially life-saving information to those who reported information, thereby closing the feedback loop;
(5) Send early warning messages to partnering organisations.
Current information collection, evaluation and verification practices of crowdsourced information are human resource intensive. PI builds upon existing crowdsourcing technology and automates, to the extent possible, these processes in order to free organisations’ resources currently used to input, tag, geo-locate and evaluate information.
The first phase of the project has produced a series of users’ stories that form the basis of a functional requirements analysis necessary to the design, development, testing and piloting of PI’s technology. Lessons learned will continue to be collected and shared during the entire life-cycle of the project.
An NGO/humanitarian agency will run the platform and set it up to suit their programme objectives, including by defining surveys and taxonomies to sort responses. Staff will be trained to use the platform; set up the surveys; and push them out on the ground. PI will be accessible through several communication channels to take into account users’ varying literacy levels and access to a reliable internet connection. For instance, information gathered can be entered into the system through sms; a form on the website; and in the future via voice recognition technology. With varying ways to upload information there is the issue of uniformity of data. The PI system mitigates this by asking a predefined set of questions and asks clarification questions if the information cannot be interpreted (e.g. a location cannot be found) or the expected format is not met. And since information received is structured, it can more easily be triangulated, sorted, mapped and visualised to help provide targeted services. Also, feedback can be provided automatically depending on what the information is.
For example, a family fleeing a conflict zone could send an SMS to PI asking for help. This will initiate a dialogue composed of a series of questions and answers. The first message received will inform the person that the SMS channel is not secure and ask for his or her consent to continue. If the person agrees, she or he will be presented with a menu of options devised with the help of PI’s humanitarian partners. For example: reply 1 to report a security incident, 2 to report a missing relative, 3 to report lack of water/food. After replying to one of the options, the person will be guided through a series of questions designed by the relevant humanitarian partner. If the answer to a question cannot be interpreted correctly by the PI system the person will be asked clarification questions until PI is satisfied that the information provided can be interpreted. PI will then provide an automated response that could include potentially lifesaving information, e.g. where is the nearest potable water point.
In the back-end, PI will automatically cluster the information received along our humanitarian partners’ taxonomies by means of keyword recognition and semantic analysis. PI will also attempt to automatically triangulate new information with other information in its database and attribute credibility scores. As for the reliability of the information source, PI will attribute a score that will reflect if a source is known to PI and if over time that source has reported information that was positively triangulated with other information. PI will then automatically map the report and visually present the information in ways that are meaningful to the humanitarian partners running the platform to allow them to make appropriate decisions.
At this invention stage, the idea was to define the main features of the platform and identify real-life applications through user stories. So far we’ve achieved our objectives, however there is a lot more to be done to develop the platform. In order to take this further, we are working with an IT partner in the Netherlands, humanitarian partners in Geneva and a prospective pilot partner in Liberia.
One problem we realised early on is getting to know the right people in the big humanitarian agencies, which at times was tricky. Also, we were expecting more people attending our workshop in Geneva. It’s also difficult to keep people captivated on the project over a long time period when you don’t have anything to show except ideas. This being said, with the publication of the requirements analysis and with a first demo of some of PI’s underlying concept accessible on PI’s website, it should make it easier to retain the engagement from partners such as IOM and UNHCR and hopefully to get the next round of funding to develop and pilot the technology.
At the workshops in Geneva participants explained that there are many old legacy systems already in place that aren’t talking to each other. So a big issue we realised is how to get our system to interact with multiple platforms that involves sensitive information? We realised more than we previously anticipated the need to keep abreast of new developments to bring existing applications in the system so we’re not replicating what already exists. People’s practical experience at these workshops was invaluable.
We have made a ‘development’ application to the Humanitarian Innovation Fund. We are also talking to Humanity United – a big funder of this type of work and the Global Innovation Fund. We need to secure further funding in order to hire developers to build the platform. The plan would be to test the platform in Liberia where we have built great connections with the Liberia Peace Building Office. Through a series of discussions with them, we already have the endorsement of 32 civil society organisations on the ground who are willing to partake in a pilot.
The progression of the innovation is an incremental and iterative process. Scalability is dependent on working with partners who have the necessary networks and the trust of the people on the ground who could make use of the PI technology. First, we need one or two pilots ideally operating within a network, where several agencies are using the information to establish interoperability. One such pilot could take place in Liberia together with LPO and the Early Warning and Early Response (EWER) network. The plan is that once early adopters succeed and provide testimonials the platform will build kudos, e.g. among IOM cluster leads on camp management and OCHA coordinators who could push it through the system.
We conducted a series of workshops with high level humanitarian actors such as ICRC, IOM and UNHCR in Geneva, as well as actors on the ground, including the Liberia Peacebuilding Office and representative of 18 civil society organisations part of the Early Warning and Early Response (EWER)network.
You are seeing this because you are using a browser that is not supported. The Elrha website is built using modern technology and standards. We recommend upgrading your browser with one of the following to properly view our website:
Windows MacPlease note that this is not an exhaustive list of browsers. We also do not intend to recommend a particular manufacturer's browser over another's; only to suggest upgrading to a browser version that is compliant with current standards to give you the best and most secure browsing experience.