🌍 Humanitarian News Dashboard

🌐 Live Demo: https://steadfast-light-production-b4f8.up.railway.app

A real-time humanitarian crisis monitoring platform built for analysts and policy professionals. The dashboard ingests articles from leading humanitarian sources, enriches them with NLP, and surfaces AI-generated situation briefs and a RAG-powered chat assistant.

---

✨ Features

• 📡 Live ingestion from OCHA, ReliefWeb, MSF, ICRC, Al Jazeera, BBC, Reuters and more
• 🗺️ Country-filtered dashboard — Sudan, Yemen, Gaza, DRC, Ethiopia, Somalia and 10+ more
• 🏷️ NLP topic tagging — conflict, displacement, famine, disease outbreak, natural disaster, and early warning
• 📰 Topics feed — filterable grid of articles by topic and country with timeline selector
• 🤖 AI Situation Briefs — GPT-4 powered analysis written for humanitarian professionals
• 💬 RAG Chat Assistant — ask questions about recent articles, grounded in your ingested data
• 🗄️ pgvector semantic search — embeddings stored in PostgreSQL for retrieval-augmented generation

---

🛠 Tech Stack

Layer	Technology
Frontend	Next.js 14, TypeScript, Tailwind CSS, SWR
Backend	FastAPI, Python 3.11, SQLAlchemy (async)
Database	PostgreSQL 16 + pgvector extension
AI	OpenAI GPT-4, text-embedding-ada-002
Ingestion	RSS/Atom feeds, httpx, readability-lxml, BeautifulSoup
NLP	langdetect, custom topic classifier
Deployment	Docker, Railway

---

📁 Project Structure

Humanitarian-News-Dashboard/
├── frontend/               # Next.js application
│   ├── app/
│   │   ├── page.tsx        # Main dashboard
│   │   ├── topics/         # Topics feed
│   │   ├── country-briefs/ # Country intelligence
│   │   └── feed-manager/   # Source management
│   └── components/
│       ├── Sidebar.tsx
│       └── ArticleChat.tsx # RAG chat component
└── backend/                # FastAPI application
    └── app/
        ├── api/            # REST endpoints
        ├── db/             # Models & session
        ├── models/         # Pydantic schemas
        └── services/
            ├── ingest/     # RSS ingestion & content extraction
            ├── nlp/        # Topic tagging & language detection
            └── rag/        # Chat & retrieval service

---

🚀 Getting Started

Prerequisites

• Python 3.11+, Poetry
• Node.js 18+
• PostgreSQL 16 with pgvector
• OpenAI API key

Backend

cd backend
poetry install
cp .env.example .env   # add your OPENAI_API_KEY and DATABASE_URL
alembic upgrade head
uvicorn app.main:app --reload

Frontend

cd frontend
npm install
cp .env.local.example .env.local   # set NEXT_PUBLIC_API_URL=http://localhost:8000
npm run dev

Or run everything with Docker:

docker-compose up

API docs available at http://localhost:8000/docs

---

📡 Key API Endpoints

Endpoint	Description
GET /articles	Paginated article list with country/topic/date filters
GET /articles/top-stories	Ranked top stories for a country
GET /briefs/latest	Generate AI situation brief
POST /chat	RAG chat endpoint
GET /stats/topic-breakdown	Crisis type distribution
GET /stats/activity	Article activity over time

---

🌐 Data Sources

OCHA · ReliefWeb · MSF · ICRC · Al Jazeera · BBC News · Reuters · The Guardian · UN News · Human Rights Watch · Amnesty International · France 24 · Deutsche Welle · AP News

---

📄 License

MIT