DEEP - Data Entry Exploration Platform

The humanitarian-response ecosystem is transitioning from information scarcity to information overload. Voluminous quantitative data, secondary data, and media-based sources are generating information on a scale that humanitarian organizations cannot cope with. They are overwhelmed and lack the expertise to utilize the available data to make the right decision during a crisis.

To address this challenge, DEEP was established in 2016 — an artificial intelligence (AI)-driven web application designed to improve data processes and information management to humanitarian crises at national and global levels.

DEEP allows humanitarian stakeholders to compile and collate crucial information from various sources and systematically organize that information through customizable tagging structures so that the information can be easily monitored and analysed to inform decision making.

DEEP aims to become a leading global consolidator and reference for humanitarian data. It has been piloted in more than 1,800 projects supporting humanitarian responses across the globe since 2016. 

By 2021, 235 million people will be in need of humanitarian assistance. This number is currently 1 out of 33 people worldwide - a significant increase from 1 out of 45 when the Global Humanitarian Outlook 2020 was launched, which was already the highest proportion in decades. 

The pandemic has crippled the traditional approach of collecting primary data - live surveys and focus group discussions. Access to quality structured information in humanitarian contexts will be elusive for the foreseeable future. Most humanitarian responses will be made on the basis of remote techniques, secondary data, social media, expert judgement or journalism. The least developed countries — which are most vulnerable to the pandemic — are the most impacted by this lack of access to direct data, and organizing remote data collection is a necessity. 

There have been thousands of academic papers, research articles, news, and social media posts on COVID 19 providing situational analysis. Attempts to quantify the information abundance may be inaccurate, but we can retrieve hundreds of millions of results when looking at “COVID 19” related news from search engines such as Google. 

DEEP is a web application allowing humanitarian responders to systematically organize information through a customizable and user-friendly web application. Data becomes more easily monitored and analysed in DEEP leading to more informed and efficient decision making.

Major humanitarian organizations (United-Nations, IFRC), INGOs, IGOs, Local & National NGOs, governments and Clusters are also users of the platform for projects all along the Humanitarian Programme Cycle, such as:

• Needs assessment and analysis studies by connecting DEEP to any online or offline data source, and managing all information through DEEP’s customizable interface,

• Strategic response planning preparation, by accessing actionable geolocalized data,

• Resource mobilization where organizations can map their projects & activities, resources, budgets and beneficiaries,

• Implementation and monitoring by carrying out multidimensional analysis to gain real time insights on response plan, from project and activity mapping to monitoring. 3W Who does What & Where, 4W & When, 5W & for Whom,

• Exporting data, infographics and maps to present it to funders and donors.

DEEP aims to increase automation in collecting and analyzing secondary data, with the support of Artificial Intelligence and Natural Language Processing (NLP). By using DEEP,  public health experts and policymakers can focustheir time on critical thinking, data analysis and response plan preparation, rather than on data collection and reporting that are low value added part of the process.

To better understand DEEP’s impact, one should distinguish direct impact - through the work we carried out with humanitarian organizations - from indirect impact, where DEEP’s outputs are used for responses on the ground.

Detailed DEEP use cases have been presented during the Humanitarian Networks and Partnerships Weeks, one of the largest global events for humanitarian networks to meet and address key humanitarian issues. The recording of the session - online this year - can be accessed here, with presentation from the Global Information Management, Assessment & Analysis Cell (GIMAC), International Federation of the Red Crescent / Cross (IFRC), Danish Refugee Council (DRC) and others.

Direct impact 

In times of sensitive humanitarian crises, Artificial Intelligence (AI) can support analysts to extract information from charts and maps and extract/classify informative snippets. This improves time and knowledge acquisition accuracy and thus the quality of the responses. AI can use information from past or similar crises to make recommendations regarding impact and unmet needs.

Without AI, multiple secondary data review analysts are required to manually extract relevant pieces of information from secondary data. For instance, during the first weeks of the 2015 Nepal earthquake, 10 secondary data officers were requested for each analyst based in the organizations’ HQ. It took four weeks to gather all relevant pre-crisis data, and a 50 page document took an analyst half a day to process at a general level. 

Streamlining the analysts’ workflow within DEEP would have allowed responders in Nepal to reduce the time required to perform data collection and analysis by a factor 2 . 

With the upcoming NLP system enhancements within DEEP - October 2021 - we expect to reduce data processing time, compared to the traditional “excel” analysis  by a factor 10. The magnitude of the monetary value saved by organizations shall be very much the same.

Our impact is also reflected within our internal HR and Contractor policies, to whom we provide competitive salaries and high quality working conditions. Among other actions, we recently created an organizational wellness fund to provides funds to team members and help to manage their worklife balance. Staff and Contractors self-describe as 68% Female and 52% of our 85 staff and contractors reside in developing/emerging economies. We are actively training and investing in individuals to take project and leadership roles in the organization.

Additional direct impact indicators are presented in the next section “More about your solution”. 

Indirect impact

One of the key outputs produced by organizations using DEEP is Humanitarian Needs Assessment (HNA), which are necessary to better understand DEEP’s potential impact at the global level. To date, DEEP has been piloted throughmore than 1800 projects, supporting humanitarian responses across the globe ranging from sudden onset to protracted and across all humanitarian sectors. 

Key responses where DEEP have been used to inform humanitarian response include:

• Venezuala migration crisis where UNHCR used the platform for media monitoring for (http://bit.ly/r4vunhcr)

• IFRC’s response to the 2019 floods in Mozambique (http://bit.ly/ifrcidai) 

• ACAPS’ analysis of the Rohingya crisis (http://bit.ly/acapsrohingya).

In 2019, IFRC published an article (http://bit.ly/ifrcdeep) describing the use of the tool in needs analysis for responding to the population displacement crisis in the Americas. 

It is difficult to quantify the impact of DEEP on populations because the tool is used at a coordination level often in large scale responses. While attempts to add concrete figures to populations impacted by the platform are inaccurate, we nevertheless track counts of populations targeted and reached by the general humanitarian response in a given country in which DEEP, and subsequently DEEP, have been used.

DEEP has been an important platform for humanitarian efforts impacting 98 million people, with 66 million earning less than 5 USD per day. Details on this social impact estimate, impacted people and the associated sources here are available on demand.

DEEP allow humanitarian to respond more effectively to outbreaks of disease and better predict the impact of that disease on vulnerable communities. COVID related data currently available on the DEEP includes:

• COVID-19 humanitarian condition assessment, to estimate the number of people in need and plan for responses,

• Geolocalized data on COVID-19 cases, population displacement, legal measures adopted by the governments, effects of the confinement on local economy, schools and health systems, etc. 

• Hundreds of monthly analysis / weekly snapshots at sub-national levels analyzing the evolution of ongoing situations, quarantine effects, access to basic goods and services, market prices, etc.
  

“Scale” seemed to be the appropriate stage of development for the following reasons:

• 85 full time employees or contractor are currently working on DEEP, spread within 15 countries DEEP has been used to inform more than 1,800 international humanitarian projects

• Cumulated funding (past + active contracts) reaches 3,000,000 USD

• Upcoming contracts shall reach 3,800,000 USD for the next 24 months.

• DEEP’s board which consists of some of the world’s largest humanitarian organizations, and are detailed in the Team section.

DEEP is the very first AI-powered application to address humanitarian secondary data challenges that allows humanitarian stakeholders to centralize all existing data in a common space. 

Today, DEEP hosts the largest analysis framework repository in the international humanitarian sector, hosting more than 85,000 carefully annotated response documents and connecting more than 3,000 expert users worldwide.

By fostering collaboration within the humanitarian ecosystem and growing our number of users base in addition to the data repositories stored within the application, our Natural Language Processing (NLP) algorithms will only get smarter. That will allow the development of advanced features such as the forecast of humanitarian needs at a granular level within countries, with sectoral level information - which has never been done yet - as well as advanced search engine features, that are part of our roadmap. 

DEEP aims to become a leading global consolidator and reference for humanitarian data. Our long term vision is to provide a global solution for inter-organizational resource optimization with the help of AI. Both Data Friendly Space and DEEP partners are convinced that the technology has a growing role to play in improving humanitarian coordination by contributing to faster and more accurate humanitarian responses. By sharing resources at the global level between humanitarian actors, the time required for humanitarian responses could be greatly decreased, and lives saved.

DEEP is currently directly serving and connecting more than 3000 users worldwide, who are humanitarian experts working within partner organizations. This number of users has been multiplied by more than 4 over the past 2 years, and the main room for growth is now at the country level. With the upcoming renewal of covid-19 analysis work with the Global Information Management, Assessment & Analysis Cell (GIMAC) and other contracts, we expect to raise this number to 4500 users within 2 year, and 20,000 within 5 years. DEEP aims to be opened outside of the humanitarian community and the number of users could be greatly increased. 

We estimate that DEEP has impact more than 98 million people worldwide since its inception. This should be compared to the 235 million people that will be in need of humanitarian assistance in 2021. This number is expected to rise in the coming years, within the increase in conflicts and natural disasters worldwide. 

Additional information on our impact and goal is presented in the next section.

In partnership with the Danish Refugee Council (DRC), we developed the following list of indicators - to mesure DEEP’s direct impact. They are classified in 3 categories, and presented below. These indicators will be reviewed on a yearly basis, with the support of our partner DRC.

Coordinated Assessments Target

• Number of humanitarian organizations actively coordinating in the proposed area of work 
• Number and percent of humanitarian organizations participating in joint assessments 
• Average number of humanitarian assistance sectors involved in the coordinated needs assessment process per assessment 
• Number of country level coordination groups (i.e., ICCGs) utilizing the Joint analysis platform 
• Percentage of Humanitarian Needs Overviews (HNOs) with a significant improvement of needs and risk analysis

Coordination

• Number of other key actors actively participating in humanitarian coordination mechanisms 
• Number of connectors established between data platforms and providers

Information Management

• Number of humanitarian organizations have received joint assessment information 
• Number of humanitarian organizations utilizing information management services 
• Number of products made available by BHA funded information management services that are accessed by stakeholders 
• Percentage of increase of registered active users in DEEP 
• Number of countries with a new active project and / or forecast model within DEEP

Indirect Impact

• Number of people in needs of humanitarian assistance, related to projects within DEEP 
• Number of people concerned by Humanitarian Needs Overview produced with the help of DEEP

We would love to partner with MIT Solve to refine these indicators, and better present DEEP’s impact and potential.

DFS is staffed by more than 85 employees and contractors across six continents.

Full-time Staff: 3

Full-time International Contractors: 82

• France: 12
• Switzerland: 5
• Nepal: 46
• Spain: 4
• Panama: 4
• Italy: 3
• UK: 2
• Australia: 1
• Austria: 1
• India: 1
• Lebanon: 1
• Nigeria: 1
• Turkey: 1

Data Friendly Space (DFS) is a US based non-profit organization dedicated to improving information management and analysis capacity, tools and processes in humanitarian and Intergovernmental Organizations (IGOs) to enable better-informed and more targeted assistance. DFS is a well known name in the humanitarian ecosystem for the services it provides, and for being the technical guardian of DEEP.

DFS staff is composed of experts from both the humanitarian and private information management/analysis fields, who specialize in real-time secondary data review and build humanitarian applications to support the rapid extraction of information from large volumes of structured and unstructured data.

DEEP was developed by and for humanitarian actors to improve data processes and enable more effective and accurate humanitarian responses to humanitarian crises. DFS works hands in hands with DEEP’s Board to define its strategy and evolve the application to meet these objectives.

DEEP’s board which consists of some of the world’s largest humanitarian organizations and includes: United Nations Office for the High Commissioner of Human Rights (OHCHR), iMMAP, United Nations High Commissioner for Refugees (UNHCR), The United Nations Children’s Fund (UNICEF), Office for the Coordination of Humanitarian Affairs (OCHA), Joint IDP Profiling Service (JIPS), Internal Displacement Monitoring Centre (IDMC) and Okular-Analytics. Additionally, DFS maintains a strong partnership and coordinating relationship with the Danish Refugee Council.

Data Friendly Space believes in a strong, diverse workforce and strives to build a culture where social, cultural, gender and identity differences are valued and promoted. We are always learning from our team and partners and we foster environments where shared experiences provide new perspectives and empowerment.

The DFS Board is made up of Board members from Asian, South American, and European origins. Staff and Contractors self-describe as 68% Female and 52% of staff and contractors reside in developing/emerging economies. We are actively training and investing in individuals to take project and leadership roles in the organization.

Although DEEP has become an integral data service for the global humanitarian community, its full potential and growth has yet to be realized. There are still a plethora of organizations and work streams that could benefit from its service. 

This includes multiple public sectors in the international development and climate change community that are prime candidates for DEEP. Through the purposeful mentoring and networking that would be made available through Solve, we would be able to prioritize growth avenues for the platform to best identify how we want to shape DEEP's growth and collaborate with MIT Solve.

Through Solve's connections to innovation partners, we also envision further developing DEEP's Natural Language Processing and Machine Learning capabilities. We ultimately envision DEEP becoming an intelligence engine for the humanitarian community and going beyond Secondary Data Review. 

Humanitarian responders need actionable analysis informed by the kind of Information Retrieval that DEEP can provide. The advice and support given by technical experts will allow DEEP's artificial intelligence and natural language processing systems to reach their full potential and strengthen the response of the humanitarian community. Improvement in these areas would provide tangible improvements to the humanitarian community's situational analysis capacity and ability to respond more efficiently to global crises.

Human Capital
Technical advisors and private sector experts to advise us on how to develop a private sector, revenue generating, version of DEEP would help extend our mission.

Business model
DEEP has been funded by grants and contracts from its advisory board and USAID.  DEEP would benefit from a business  model that is less reliant on funding from governments and the United Nations.

Public Relations
One of DEEP's most significant soft-spots has been the communication of its importance to the humanitarian community through earned or paid media. Better public relations would lead to better visibility and recognition of the platform which would in turn open doors to additional funding, users and champions.

Monitoring & Evaluation
DEEP is used to inform higher level global coordination structures and complex i- country responses but it is difficult to assess the project's direct impact. Enhanced M&E would allow for the project team to get a better understanding of what impact the platform is providing, and communicate the value of DEEP to other sectors as part of a broader scaling strategy.

Technology
DFS needs technological support to enhance its NLP functions. Our NLP roadmap is substantive t, and includes integrated GIS, artificial intelligence, and predictive analysis.

Our work is only fully realized through collaboration with external partners, and we are interested to collaborate on:

• Share DEEP's COVID-19 related data,

• Partner with academic and private sector organizations to improve the platform’s AI and NLP functionalities. DFS and DEEP are willing to share NLP datasets and models created with standard humanitarian taxonomies, which could be opened to new actors or connected to existing systems on demand.

We have been through MIT Solve ecosystem and partner list, and have identified synergies with the following organizations: 

MIT research centers: 

• MIT NLP CSAI We have seen a very beneficial partnership through researchers and students at EPFL in Switzerland and JKU in Austria, and would like to bring this stateside. Partnership with CSAIL would allow for research into more advanced and experimental feature sour NLP system.

• MIT Humanitarian Supply Chain Lab We are currently raising funds among our existing network for a new module dedicated to anticipatory actions. Humanitarian organizations will use these outputs for optimizing resources and supply chain to provide better responses on the ground.

We have also identified the following foundations who may be interested in the above listed elements:

The Bill & Melinda Gates Foundation, Vodafone Americas Foundation, Andan Foundation, Dubai Cares, Elevate Prize Foundation, Firefly Innovations at CUNY Graduate School of Public Health and Health Policy, Intuitive Foundation, Patrick J. McGovern Foundation, and Straubel Foundation.