Global Events
Understand root events and cross-lingual trend clustering.
Global Events
Global Events show you the big picture. While individual trends capture what is happening in a specific country and language, Global Events group related trends into macro-level stories that span borders and languages. A single event like "EU AI Regulation" might manifest as distinct trends in France, Germany, the US, and Japan -- Global Events tie them together.
How it works
IANews uses a 2-layer architecture to organize trends:
- Root Event (macro layer) -- A cross-lingual, cross-country grouping. Example: "AI Regulation Debate".
- Sub-Clusters (micro layer) -- Individual trends per country and language that belong to the root event. Example: "EU AI Act" (France/fr), "AI Safety Bill" (US/en), "AI規制" (Japan/ja).
The system matches trends across languages automatically, regardless of language. Smart entity disambiguation prevents false matches -- "Apple" the company and "apple" the fruit are never confused.
Temporal windowing
Not all stories have the same shelf life. The system uses category-aware time windows:
| Category | Window | Rationale |
|---|---|---|
| Breaking news | 48 hours | Fast-moving, short-lived stories |
| Markets / Crypto | 24 hours | Rapid price movements and reactions |
| Science / Academic | 14 days | Papers and discoveries circulate slowly |
| Default | 7 days | Standard news cycle |
Trends outside their category's time window are not clustered together, even if they are textually similar. This prevents a 2024 event from merging with a 2026 event about the same topic.
The Events page
Navigate to Events in the sidebar to open the Global Events page. The page shows a filterable list of root events, each displaying:
- Title -- The root event name, representing the macro story.
- Global score -- The maximum score among all sub-clusters. If the US sub-cluster scores 85 and the France sub-cluster scores 72, the root event shows 85.
- Country count -- How many countries have sub-clusters for this event.
- Language count -- How many languages the event spans.
- Propagation speed -- How quickly the event spread geographically, measured as hours from the 1st to the 5th country detecting it.
Event detail panel
Click any root event to open the detail panel. It contains:
Sub-clusters
A list of all sub-clusters grouped by country. Each sub-cluster shows:
- Country flag and name
- Language
- Local trend title
- Score
- Lifecycle stage
- Source count
Propagation timeline
A geographic visualization showing how the event spread over time:
- First detection -- The country and timestamp where the event first appeared.
- Spread sequence -- Subsequent countries in chronological order.
- Current reach -- The total geographic spread as a percentage of the 50 monitored countries.
For example, a tech announcement might show: US (0h) --> UK (1h) --> Germany (2h) --> France (2.5h) --> Japan (6h).
Root events update every 60 seconds. New sub-clusters are matched and merged automatically as the system detects related trends in additional countries or languages.
Use case: tracking global stories
Global Events are particularly valuable for editorial teams covering international news. Instead of monitoring 50 country-specific trend feeds, you see one unified event with all its regional manifestations. This helps you:
- Spot global stories early -- An event appearing in 3+ countries is likely significant.
- Understand geographic spread -- The propagation timeline shows which regions are leading or lagging.
- Write multi-angle articles -- Use sub-clusters as distinct perspectives for a comprehensive piece.
Sort events by propagation speed to find stories that are spreading fastest. A rapidly propagating event (e.g., 5 countries in under 6 hours) is a strong signal for breaking news coverage.
What's next?
- Explore trends -- Dive into individual country-level trends
- Follow and watch trends -- Track specific trends over time
- Trend alerts -- Get notified when trends match your criteria