semantic-relations

To prevent incorrect semantic labeling of knowledge graph edges, apply a typed-relation taxonomy with substitution tests to classify IS-A and PART-OF claims, verifying symmetry, asymmetry, transitivity, and reflexivity expectations explicitly.

IS-A and PART-OF hierarchy claims differ fundamentally: IS-A represents taxonomic subsumption validated by substitution tests, while PART-OF represents mereological part-whole semantics requiring distinct validation checks to prevent conflation.

Audit knowledge graph relations for anti-patterns by applying a verification checklist that detects vague related_to edges, circular IS-A structures, IS-A vs PART-OF conflation, synonym sprawl, and untyped polysemy.

No, this approach to relation typing does not commit to a specific formal ontology or database implementation; it focuses purely on classifying edge types and relation properties without making implementation commitments.

Route semantic relation tasks to adjacent skills like taxonomy-design or ontology-modeling when the work shifts from classifying edge types and relation properties to broader structural design, linguistic analysis, or entity-relationship modeling.

Validate thematic roles and lexical relations like polysemy and homonymy by applying a typed-relation taxonomy that distinguishes agent, patient, instrument, cause, and goal connections alongside synonym, antonym, and metonymy distinctions.