Glossary methodology
release@1.3.0 · glossary@1.3.0
Topic areas
D1–D6 identify topic areas, not difficulty levels or a required learning order.
- D1 · Foundations, models and data
- D2 · Generative AI, capabilities and limitations
- D3 · Prompts, context, RAG, tools and agents
- D4 · Evaluation, reliability and evidence
- D5 · Security, privacy, governance and intellectual property
- D6 · Human judgment, workflows and adoption
Source hierarchy
Authoritative public standards, public primary research, and official documentation take precedence. Supplementary sources add context but do not replace the source needed for a claim.
Proof and verification
Each atomic claim is paired with a source role, exact locator, verification date, review date, and its own scope limitation. Independent review checks whether the cited material supports that claim; it does not turn a source into a guarantee.
Rejection rules
A term or claim is withheld when its public fields, locators, provenance, or supporting evidence are incomplete, ambiguous, or outside the stated scope.
Hashing and versioning
The site imports a public release only after its release version, glossary version, and SHA-256 digest match the accepted manifest. The runtime does not read the source workspace.
Freshness and corrections
Review dates are calculated from public citations. Entries remain visible with a localized warning when review is due. To report a correction, use the site contact page and include the term, claimed issue, and a public source or locator.