VisiUN

Methodology Tab
Info Tab
Crisis Tab - Crisis Categories
Countries Tab - Sortable
Crisis Tab - Conflict Category
Crisis Tab - Individual Crisis Statistics
Crisis Tab - Individual Crisis Content
Front Page - Global Humanitarian View
Front Page - Top Funding Gaps
Front Page - Top Overlooked Crises
AI Tab
Countries Tab - Individual Country

Inspiration

In an era of digital misinformation and polarization, finding centralized, objective information about global conflicts is incredibly difficult. As a result, critical humanitarian emergencies often get lost in the noise. With the U.N. currently experiencing severe financial cuts, maximizing awareness of these overlooked crises has never been more important. To tackle this, our team took on the U.N. challenge. We built a transparent, data-driven platform that translates massive datasets into accessible, easily understandable graphics, allowing anyone to clearly see the scale of worldwide situations and the stark funding gaps they face. Introducing VisiUN, your centralized hub for humanitarian crises.

What it does

VisiUN serves as an interactive data visualizer across a multitude of statistics, providing a holistic perspective grounded in facts for the user to interpret and analyze.

The Data

Taking data from the U.N.'s Financial Tracking Service, ACAPS' INFORM Severity Index, and the Humanitarian Data Exchange, we use well-informed calculations and graphics to demonstrate disparities and potential holes in funding.

The Globe

Our most eye-catching tool is the interactive globe, plotting spikes to represent a country's Severity Index and Funding. Clicking on a country lets the user learn more about any situations in the area, allowing a geographical analysis of how funds are distributed and crises are happening.

Overview Tab

This is the at-a-glance breakdown of overlooked crises, overlooked countries, absolute funding gaps, and summarizing statistics. By looking at this sorted data, the user can see the most pressing crises and click on any element to learn more. A holistic view presents a quick, insightful look into current underserved emergencies.

Crisis Tab

This tab provides insight into crises grouped by category: Displacement, conflict, natural disaster, drought, economic crisis, political crisis, and climate. By clicking on a category, the user can investigate into the collective severity and funding statistics, as well as a list of every crisis involved.

Clicking into a crisis provides a detailed case study into the related country's funding gap, potential anomalies, and support effectiveness. It also provides a brief summary of each crisis with a digestible timeline and external links for more in-depth understanding.

Countries Tab

This tab gives users the power to sort and filter between countries and their funding or crisis statistics. Clicking on a country informs them about where funds are coming from, what crises are receiving these funds, how many funds are needed, and how people are being affected. These statistics are filtered into our transparent Neglect Index, which calculates how overlooked a country is.

Crisis AI

Need more in-depth analysis? Engage with our specialized LLM chatbot with direct access to an array of presented datasets. It performs high-quality and efficient data analysis on any related query, deriving personalized insights with conversational feedback.

Info Dropdown

Curious about any of our data sources, processing, terminology, calculations, or anomaly detection? Click the (i) button in the top right to learn more.

How we built it

Frontend - Our user interface was built with Next.js, TailwindCSS, and ShadCN for seamless animations and customizable components, using Three.js to design our globe map with detailed and efficient 3D rendering.
Backend - We use Node.js and OpenRouter to handle server-side operations and LLM connections.
Hosting - We use Vercel for the development process and for hosting our production environment, connecting with the GitHub Repo for streamlined development.
Machine Learning - From vector embeddings to LLMs, we use Langchain, Grok, and Huggingface to analyze our data and derive custom insights instantaneously based on user input.

Challenges and accomplishments

We spent a lot of time devising effective strategies to visualize and represent data comprehensively, since current crises contain many factors and layers that can easily overwhelm users. Portraying these intricate realities through clean graphics was a major hurdle.
We also placed heavy emphasis on unbiased reporting and proper statistics, ensuring our platform remains an objective source of truth. Navigating raw data without feeding into the very digital polarization we sought to fight required meticulous attention to detail and careful dataset parsing. Creating reputable and objective summaries, timelines, and supporting links.
Our 3D globe visualization, colorful charts, and personalized chatbot diversified the application's utility, allowing different users to interact for a variety of purposes.

What we learned

We saw the potential of Retrieval-Augmented Generation in real-world analysis on a global scale, conducting well-informed insights that could improve lives.
We recognized the importance of organization and compartmentalization; when scaling our data operations and expanding our codebase, we realized that organization was vital to make efficient improvements. We refactored our codebase in the middle of the hackathon to maintain readability.

What's next for VisiUN

Throughout the course of the hackathon, we had to prioritize foundational features, but we had many visions to expand.

Including time-series with a scrollable timeline, providing an extra layer of analysis that data tables do not provide.
Including a wider variety of datasets across topics like hunger, mortality, happiness, and infrastructure destruction to consider greater factors.