duleab/Africa-Crisis-Intelligence

# Africa Crisis Intelligence **AI-Powered Crisis Monitoring, Verification & Early Warning for East Africa** Africa Crisis Intelligence (ACI) is an early-stage crisis-intelligence prototype developed for the **STP'26 Hackathon by Ejo Labs**. It combines multi-source crisis monitoring, AI-assisted analysis, explainable confidence scoring, geospatial visualization, duplicate-event control, and Telegram alerts in one automated workflow. > **Demo notice:** The dashboard includes a curated East Africa > demonstration dataset to show how the platform can visualize and > analyze regional crisis signals. Operational decisions require human > verification. ## Problem During disasters and emerging crises, information is fragmented across news outlets, disaster feeds, government sources, and other channels. Reports may be duplicated, incomplete, difficult to verify, or geographically unclear. This makes rapid situational awareness difficult for communities and response organizations. ## Solution ACI converts fragmented reports into structured crisis intelligence: **Sources → Normalize → Merge/Deduplicate → Confidence Scoring → AI Analysis → Geolocation → Storage → Dashboard → Telegram Alerts** The prototype currently focuses on East Africa while supporting a longer-term vision for Sub-Saharan Africa. ## Core Features - Multi-source crisis ingestion, including news and GDACS-style disaster feeds - n8n-based workflow orchestration - AI-assisted crisis classification and summarization - Explainable confidence scoring - 50% analyst-attention threshold for the hackathon prototype - Duplicate-event filtering to reduce repeated alerts - Geographic coordinates and location precision - Interactive Leaflet/OpenStreetMap dashboard - Search and filters by country, crisis type, source, and confidence - Telegram early-warning notifications - Structured event storage - Human-verification safeguards ## Repository Structure ``` text Africa-Crisis-Intelligenc …