Namespaced semantic identifiers — v11
Generic name: semantic identifier. For entities/concepts, the precise expression is semantic referent identifier (French: identifiant de référent sémantique, IRS). A bare Q is not a generic name: Q… is a Wikidata item ID.
| Internal family | Meaning | Typical canonical location |
|---|---|---|
KQ… |
Entity/concept/referent | Subject and entity-valued object |
KP… |
Property/relation | Predicate and qualifier |
KA… |
Assertion | Independently identified statement |
KS… |
Source/evidence anchor | Source reference |
The actual identity key is (namespace_id, KQ123) or (namespace_id, KP456), not the number 123 or 456. KQ123 in one project and KQ123 in a different namespace do not automatically designate the same meaning.
Sample (illustrative, not production allocations)
{
"namespace_id": "example.org/kristal-demo",
"entities": ["KQ1001", "KQ1002"],
"properties": ["KP2001"],
"sources": ["KS4001"],
"assertions": [{
"id": "KA3001",
"subject": "KQ1001", "predicate": "KP2001",
"object": {"kind": "entity", "id": "KQ1002"},
"sources": ["KS4001"]
}]
}
A real kristal_semantic_graph additionally includes schema_version, artifact_type, graph_id and is validated together with a matching kristal_identity_registry. The excerpt above is illustrative, not a complete schema-valid artifact.
Typed literals such as {"kind":"literal","datatype":"string","value":"A historical quotation"} may appear as values; the structural identity fields remain opaque IDs. The issue is never whether text exists anywhere; it is whether a word is being used to identify semantic meaning.