Leveraging Data-Driven Innovation for Humanitarian Impact in Refugee Communities
As a member of the Patrick J. McGovern Foundation’s (PJMF) 2023 Data to Safeguard Human Rights Accelerator cohort, Voices of Venezuela, a family-driven endeavor to assist the Venezuelan migrant community by connecting them with the organizations and institutions that support them, is partnering with PJMF’s Data Practice team to develop a cloud-based system with comprehensive data analysis capabilities and a nuanced and location-based understanding of migrant needs.
by Jason Rovig and Nery Santaella
In the complex landscape of global humanitarian efforts, the Voices Of Venezuela (VOV) initiative effectively demonstrates how advanced technology and data analytics can profoundly transform support systems for refugees and migrants. VOV is unique in its success and desire to mutually benefit stakeholders and support community groups through a markedly enhanced reciprocal flow of information. More than just improved data transfer, this approach is about crafting an ecosystem where information is a powerful tool for mutual empowerment and informed decision-making, essential for impactful humanitarian interventions.
Starting in media, VOV reached over four million people. With the addition of our helpline, VOV has exchanged more than a million messages, benefiting over 640,000 migrants. Overcoming the technical challenges of this fast growth has been a constant learning curve. We’ve strategically shifted our focus towards tech and data and began integrating open-source software with our artificial intelligence (AI) models, maintaining a flexible approach.
Art For Impact SAS and Tech Equalized LLC, the driving forces behind VOV, are creating synergistic collaborations across various sectors, including community groups, CSOs, NGOs, and government bodies. Their work has been propelled by participation in the Patrick J. McGovern Foundation’s Data to Safeguard Human Rights Accelerator cohort, which provided essential resources for developing AI-driven models and information management infrastructure. Those AI models are crucial for accurately identifying needs. They are integral to VOV’s data-driven decision-making approach as we build our CSO network and network our helplines for more up-to-date information on changing needs. The goal for this will be to build an early warning system with the network in the future. The service architecture used RapidPro to manage conversational flows connecting multiple social media and messaging platforms. This regularizes data collection and then segments data into meaningful categories through NLP models to identify intents, securely stored in Google Cloud Storage before additional data analysis is done after removing personal information before analysis.
The PJMF Accelerator has been a milestone for VOV, enabling the automated classification of 34 distinct intents within its helpline communications. This process, enhanced by OpenAI and HuggingFace tools and refined with Wit.ai, has led to the development of a natural language processing model that not only achieves accuracy but also resonates with the real-life situations of migrants. Using BigQuery and displayed in a Looker Studio dashboard, these intents enable practical data analysis and utilization. Over 80,000 WhatsApp messages from users, selected for relevance and content richness, were analyzed to develop natural language processing models. These models, built on real migrant experiences and communication for intent classification, have been developed by bulk processing messages before importing models into Wit.ai and integrated with RapidPro. These were then tested and further refined. We aim to make an adaptive system that enables a real-time response as emergent needs surface.
As VOV transitions into a more collaborative and integrative model, we are much more robust with a solid infrastructure and data foundation established through PJMF’s support. This strategic shift is more than an expansion of scope; it represents a deliberate effort to create a cohesive framework for collective action. VOV’s new initiative to connect 50 CSOs with nine mayoral offices of Colombian cities exemplifies this new direction. Our experience in the Accelerator revealed that not all of our tech needs need to be kept in-house, and with that, we saw that if our models and verified information sources were opened up to other organizations, local communities and organizations would be more robust and better able to make data-informed decisions and counter misinformation and disinformation campaigns. Using our work with the Accelerator, we will help other organizations build their helplines, and we hope to set new standards for collaborative efficacy in humanitarian efforts while measuring needs and humanitarian impact across the broader humanitarian ecosystem.
To scale the volume of data and enhance the benefits to migrants, VOV is venturing into Generative AI. This strategic pivot builds on the groundwork laid with PJMF’s support. Embracing these advanced technologies, VOV aims to redefine the standard operational paradigm, shifting from stakeholder competition to a model focused on collective effort. This shift involves creating a system to build and maintain comprehensive, localized knowledge bases. These knowledge bases, crucial for providing up-to-date, locally relevant information for helplines, will be augmented with Large Language Models (LLMs) to ensure that VOV maintains the contextual relevance that its users value. This innovative approach extends beyond traditional data analysis; it offers specific yet broadly applicable insights, providing our services resonate deeply within the unique cultural and situational contexts of our served communities.
Developing these knowledge bases begins with extensive surveys of CSOs, aiming to create a repository of local resources from the broader humanitarian community. When the repository’s scope isn’t sufficient for user needs, an ‘Ask the Community’ service will link the VOV helpline with the CSO network to crowdsource additional information. This model enhances data exchange while allowing user anonymity and giving due recognition to contributing CSOs. It exemplifies a symbiotic environment where information exchange leads to mutual growth and enhancement.
Looking ahead, VOV is embarking on two major initiatives that serve as testing grounds for its evolving system. The first initiative aims to strengthen the connection between CSOs and government bodies, enhancing collaborative frameworks and collective action. The second is a significant financial inclusion project to connect over 100,000 refugees to banking services. These initiatives offer practical opportunities to refine our service and embody the broader vision of VOV’s founders, which is to create a platform designed to support not only Venezuelan migrants but all vulnerable groups in collaboration with the organizations that serve them.
As VOV progresses with new ventures, our commitment to ethical data use, reinforced through education provided by PJMF, remains steadfast. Our focus is clear: to provide innovative support deeply intertwined with the needs and realities of the communities we serve, building on the solid foundation to manage data established through the Foundation’s support.
To learn more, read our full insights report.
