Breaking news is currently in beta. We do not recommend using it in production
as it may undergo changes.
Overview
Breaking news represents important news stories rapidly gaining traction across multiple sources. Our system identifies these stories through automated analysis of news coverage patterns, allowing you to discover high-impact events as they emerge. A breaking news event is characterized by:- Coverage from multiple different sources and authors
- Rapid publication within a short time frame
- Significant importance within a specific topic area
How it works
Our breaking news detection works through several interconnected stages:1
Clustering
Recently published articles are automatically grouped together based on
their content similarity.
2
Validation
Each cluster undergoes analysis for breaking news signals, including:
- Publication frequency (sudden spikes in coverage)
- Source diversity (coverage across multiple publishers)
- Coverage quality (presence of high-ranking news sources)
3
Deduplication
The system filters out duplicate content within clusters while preserving
source diversity.
4
Historical Analysis
New candidates get compared against recent breaking news to track continuing
stories.
5
AI Evaluation
An AI analysis step confirms whether potential breaking news clusters
represent significant events.
6
Classification
Clusters that meet the breaking news criteria receive a special flag in the
system.
/breaking_news
endpoint, you receive the most
representative article from each breaking news cluster. Results are ordered by
cluster size, so the most widely covered breaking news appears first.
Breaking news endpoint
The/breaking_news
endpoint provides access to breaking news stories through
simple API calls.
Base URL
HTTP Methods
GET
: Query parameters in the URLPOST
: Parameters in request body (JSON)
Request examples
Response format
The API returns a JSON object with the following structure:breaking_news_events
: Array of breaking news events, each representing a significant news clusterevent_id
: Unique identifier for the breaking news event/clusterarticles_count
: Total number of articles in this breaking news clusterarticles
: Array of articles related to this breaking news event (limited by thetop_n_articles
parameter)
Comparison with other endpoints
Endpoint | Breaking news | Search | Latest headlines |
---|---|---|---|
Purpose | Discover emerging stories with significant traction | Find specific content matching search criteria | Retrieve recent headlines for the specified time period |
Time scope | Fixed at 24 hours | Configurable with from_ and to_ | Configurable with when |
Result grouping | Clustered by event with configurable articles per event | Individual articles | Individual articles |
Primary sorting | By cluster size (coverage volume) | By relevance, date, or rank | By publication date |
Best for | Discovering important new stories | Finding specific content | Monitoring regular updates |