Canonical intelligence pipeline

Publishing Workflow

Operational flow describing how EvidTrace transforms candidate news extractions into canonical public claim records linked to claimants, sources, evidence trails and timelines.

01
Ingest
Curated RSS feeds and monitored publishers generate candidate article records.
INPUT → candidate article
02
Extract
Candidate claims are extracted manually or semi-assisted from high-signal content.
OUTPUT → candidate claim
03
Review
Editorial review validates specificity, attribution, evidence relevance and significance.
STATE → editor_reviewed
04
Canonicalise
The claim receives a permanent canonical record linked to claimant and provenance entities.
OUTPUT → canonical claim
05
Publish
The claim appears in the public ledger with evidence trails, timelines and confidence scoring.
DESTINATION → public intelligence layer
Canonical transformation rules
Specificity requirement
Claims should be materially specific and attributable to a named claimant or institution.
Evidence relevance
Claims should support meaningful evidence analysis rather than vague commentary.
Timeline capability
Claims should support future updates, revisions or measurable developments over time.
Canonical identity
Each published claim receives a stable permanent claim identifier.
Future operational architecture
curated RSS → ingestion queue → extraction review → claimant matching → source matching → evidence enrichment → confidence scoring → canonical publication → timeline updates → relationship graphing