Glossary
- Attestation
- A signed statement by an attestor about one evidence item: the level awarded, the role, a conflict-of-interest declaration, a blind-marking commitment and its reveal.
- Attestor
- A person or organisation, identified by a decentralised identifier, who signs attestations. Nobody appoints attestors for the protocol; verifiers choose which to accept.
- Claim
- A statement that a subject has a skill at a level on a date.
- Commit and reveal
- A method of blind marking: each attestor first publishes a hash of the mark and a secret value, and reveals the mark only after all commitments are in. Marks are therefore independent.
- Decentralised assessment
- Assessment by independent assessors whose results a verifier may accept or not, with no central body that accredits. Sanad does not accredit.
- Evidence
- A typed record that backs a claim: practice runs that can be replayed, a live sitting, reviewed work, or peer review.
- Prerequisite skills inferred from a demonstrated skill by following the skill graph. They are always labelled inferred, carry their derivation, and stay private until the holder publishes them.
- Level
- An integer from 0 to 5: none, aware, guided, independent, fluent and transferring to new problems, expert.
- Policy
- The verifier's own rules: how many attestors, whether a live sitting is required, how fast evidence ages, which attestors are trusted.
- Recognition function
- The published, deterministic function that takes a policy, a claim, evidence and attestations and returns a level, a confidence, a status and an ordered explanation.
- Skill graph
- An open, versioned, directed acyclic graph of skills with prerequisite edges and crosswalks to other vocabularies.
- Transparency log
- An append-only Merkle log of attestation hashes with signed checkpoints and inclusion and consistency proofs.
- Verifier
- Anyone who runs the recognition function on a bundle under their own policy.