Zone 01
Source
Ownership
🌐
Discovery
open datasets
domains, not articles
🔎
Verification
probe → detect
verify or flag
📡
Channel discovery
feeds · news sitemaps
layered probing
🗃️
Registry
candidate → verified → active
sightings · authority
publishers are discovered
and verified, not hand-listed
Zone 02
Crawl Lane
(own queue)
🔁
Pollers
conditional GET
health ladder · budgets
⬇️
Fetcher
robots · politeness gate
egress through a proxy
🧬
Extract & dedup
one publisher, same text → duplicate
across publishers → wire copy, kept
isolated worker - a slow publisher
cannot starve the pipeline
Zone 03
Understanding
🧭
Embeddings
multilingual vectors
approximate NN index
🧩
Three-way clustering
a higher bar to create an event
than to join one
📐
Scoring
originality · evidence · depth
new facts · sensationalism
📝
Synthesis
headline · summary · timeline
earlier entries preserved
an event is never declared -
it is seeded by sources that agree
event +
identity
re-synthesis
reopen
Zone 04
Identity
& Relations
🎯
Candidate retrieval
identity sentence · centroid
rare-tag blocking
⚖️
Adjudication
duplicate · follow-up
reaction · unrelated
🔗
Merge · relate · reopen
the loser redirects forever
late coverage wakes an event
✦ shadow mode first: check and log
before it may change data
Zone 05
Ranking
& Delivery
📊
Importance & geo
weighted signals under decay
tags · regions · tiers
🍽️
Value-first feed
must know · for you · horizons
sensationalism subtracted
🗞️
Public surface
our own words + a link out
never the publisher's text
ranked by knowledge value,
never by clickability
Breaking news
A story gaining sources fast jumps to a fast lane and shows as Developing - confirmed only once independent sources agree, never by one loud outlet.
Perspectives
How each country covers the same event: who reported first, who lagged, who stayed silent, and which facts each one left out.
Any language
Articles in every language merge into one event. Its story is written once in the original language, and every translation derives from it.
Two geographies
Where a publisher sits and what an event concerns are separate axes: US media writing about Japan still make it a Japanese story.