Lexicon · Tools and roadmap
Human review
Fully specified name: Human review (concept)
A recorded evaluation by an accountable person before a proposal is accepted or published.
- Scope note
- The NucLex principle is that AI proposes and humans approve.
- Synonyms
- None recorded as true synonyms in this context.
- Monograph
- Human review of machine proposals (planned; brief only)
- External mappings
- None recorded. In W0 no mapping to SNOMED CT® or any external authority is asserted for this entry. A future mapping record will carry source system, identifier, version, relation type, evidence, author, confidence where meaningful, and review state.
- Formal status
- Editorial entry only. It is not part of any released ontology and a prose edit here changes no formal definition.
Draft. This entry is an unreviewed draft. Its sources have not been checked by a named person and no domain reviewer has approved it. Treat every claim as provisional.
Notes
Human review is the step at which a named, accountable person examines a proposal and decides whether it is accepted, revised, or rejected, and records that decision with a reason. The proposal may be a manuscript, a concept, a synonym, a terminology mapping, an interaction record, a parameter value, or a citation. The proposer may be a colleague, a script, or a language model. What makes the step a review is the record: who looked, at what, when, and what they decided.
The term has established meanings elsewhere. In publishing, peer review evaluates a manuscript against disciplinary standards. In terminology maintenance, editorial review decides whether a proposed change enters a release. In software and regulated industries, "human in the loop" names a point where a person can intervene before an automated action takes effect. NucLex draws on all three but asks for the lightest workable record.
The project's working principle: artificial intelligence may propose text, terms, mappings, evidence packets, and outlines, and humans decide what is accepted. A machine-generated citation or reasoning chain is checked against its source before it is trusted. A rejected proposal may remain in the editorial record but is never displayed as accepted fact. An empty reviewer field means no review has happened; it is never filled because someone requested the work. The monograph examines why the principle is harder to keep than to state.