World Chaos Map

World Chaos Map ingests global news and structured data feeds, extracts incidents (conflicts, disasters, protests, cyber attacks, etc.), and computes a pressure score for each country on a 0-100 scale. Scores reflect recent observable pressure based on reported incidents. They are not predictions or risk forecasts.

• Local NLP classifier — event classification, casualty extraction, content kind detection, and country role attribution all run locally with zero API cost
• Source credibility — dynamic health tracking per source (success rate, yield, dedup survival) weights severity automatically
• Semantic dedup — event-signature clustering detects same event across different sources, boosting confidence for multi-source reports
• Multilingual detection — 12 Unicode script families identified, Latin sub-language detection for 6 languages
• NLP-enriched scoring — content kind (hard incident vs analysis vs context), casualties, and source diversity feed directly into the scoring pipeline
• Data-driven scoring — hardcoded country floors/caps replaced by dynamic algorithms computed from conflict patterns, casualty data, and headline intensity
• 158 tests across 18 test files