Skip to content

Generalize the shared concept space with an exact-match reuse discipline #57

@konard

Description

@konard

Context

Issue #47 requires "advanced shared concepts space between languages, which
should be reused when only matched exactly, so it will help to simplify
automated translation between any languages". The 351-concept lexicon
(seed_common_concept_ontology) and reference_for_atom exact-find_term
reuse exist, but concepts are not interned across languages under a stated
exact-match discipline, and concept-mediated translation is demo-grade. See
requirements.md R-6 and
solution-plans.md S-9.

Research (competitors-natural-language.md):
Wikidata's two-layer design (language-bound lexeme ↔ language-free Q-item,
all CC0) is the friction-free model; WordNet CILI institutionalizes exactly
the issue's rule - reuse an interlingual id only on exact match, otherwise
mint a new one. BabelNet is rejected (non-commercial license, API-gated).

Scope

  • Two-layer concept links: language-bound expression links connected to
    language-free concept links; reuse a concept only on exact-id match,
    otherwise mint a new concept link.
  • Alias links on concepts for external ids (WordNet CILI ILI ids, Wikidata
    Q-ids) so external vocabularies can be attached without becoming load-bearing.
  • Generalize seed_common_concept_ontology into an import surface that loads
    concept sets from LiNo files, keeping the 351-concept seed as the default.
  • Tests proving the exact-match rule: same id → same link reused; near-miss
    (case, diacritics, sense) → new concept minted.

Acceptance criteria

  • Concept interning API with the exact-match rule, gated by tests
    including near-miss cases.
  • External-id alias links queryable via LinkQuery.
  • Concept import from LiNo round-trips through the ontology.
  • Changelog fragment added (bump: minor).

References


Filed from docs/case-studies/issue-47/proposed-issues/09-shared-concept-space-exact-match.md. Part of the implementation plan for #47.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions