STANDARD A

Verification Standard

Reference: FIDES-RFC-001
Status: Active Standard
Target Audience: Verifiers, Data Aggregators, AI Model Trainers

1. Objective

The Verification Standard (Standard A) outlines the mandatory criteria for an information asset to receive a verified Trust Score. This standard is designed to be executed by both human auditors and autonomous AI agents.

2. Provenance Verification (Layer 1)

Before content analysis begins, the originating entity must be validated against the Global Source Index. An asset is automatically rejected if its source provenance cannot be established.

  • Domain Integrity:The source domain must have active DNS records for >12 months and valid WHOIS data.
  • Identity Match:SSL certificates must cryptographically match the claimed organization identity.
  • Ads.txt Validation:If the site serves programmatic ads, ownership must be verified via a valid ads.txt file to prevent domain spoofing.

3. Semantic Integrity (Layer 2)

The content body is scanned for linguistic markers of fabrication, emotional manipulation, or AI hallucination.

  • Emotional Density Cap:The ratio of emotive adjectives/adverbs to total words must be below 15%. High emotional density triggers a "Bias Warning" flag.
  • Logical Consistency:The text must not contain internal contradictions (e.g., claiming "X increased" and "X decreased" in the same context without temporal distinction).
  • Clickbait Detection:Headlines must possess a Semantic Match Score of >90% with the body content.

4. Consensus Mechanism (Layer 3)

A factual claim is only fully verified if it can be cross-referenced with other trusted nodes in the network.

  • Tier-1 Match:Validated immediately if matched by at least 1 Tier-1 source (e.g., Reuters, Bloomberg).
  • Tier-2 Consensus:Validated if matched by at least 3 independent Tier-2 sources.
  • Primary Source Link:The asset must contain at least one outbound link to a primary data source (PDF, Government URL, or Official Statement).

5. Trust Score Calculation

Assets are assigned a final score (S) based on the weighted sum of the three layers.

S = (Source_Rating * 0.4) + (Semantic_Score * 0.3) + (Consensus_Factor * 0.3)

*A score of < 60 results in a "Unverified" status. A score of > 90 results in "AAA" status.

← Return to Standard Overview