Crosswalks & Alignment Between Standards

No single standard covers everything — so you map

On a real project you rarely use one standard in isolation. You hold data described in several (SNOMED↔ICD, GS1↔schema.org, FIBO↔your internal model) and the value is in aligning them faithfully. This is the capstone skill: a good mapping preserves meaning and is honest about where meanings don't quite line up.

• Class & property alignment (OWL) — owl:equivalentClass, rdfs:subClassOf, owl:equivalentProperty. Powerful, but equivalentClass is a strong claim a reasoner will act on — assert it only when the two classes truly coincide.
• SKOS mapping relations — for vocabularies/thesauri, SKOS offers graded honesty: skos:exactMatch, skos:closeMatch, skos:broadMatch / narrowMatch, skos:relatedMatch. closeMatch is the unsung hero — it says 'near enough to follow, not safe to infer through'.
• SSSOM — the Simple Standard for Sharing Ontological Mappings captures each mapping with metadata: who made it, the predicate, a confidence score, the justification. Mappings become data you can review and maintain, not lost tribal knowledge.
• The pitfalls — false exactMatch (granularity mismatch: one standard splits what another lumps), unversioned maps that rot as standards evolve, and no owner for the crosswalk. Treat mappings as governed, versioned artefacts.

Use Case Example: A health analytics team needs ICD-coded billing data to line up with SNOMED-coded clinical records. They use the published SNOMED→ICD map, record each link as SSSOM with skos:closeMatch (not exactMatch, because the granularities differ), and version the crosswalk — so when SNOMED's next release ships, they can re-validate instead of silently drifting.

Aligning standards is translation between languages. A careless translator declares two words 'the same' (exactMatch) and quietly changes the meaning; a good one knows when there's a perfect equivalent and when there's only a close word that needs a footnote (closeMatch). SSSOM is the translator's annotated glossary — every choice recorded with who decided it and how confident they were — so the next person can trust, review and update the work instead of guessing.

A hub of standards joined by graded mappings

Different predicates carry different strengths of claim — and honesty about granularity is the whole game.

Web and mapping field guide

schema.org and crosswalk work reward restraint. You are not always trying to prove strict logical identity.

The practical warning: the most damaging mapping bug is a false exact match. Prefer a weaker, honest relation over a strong, convenient lie. Your future reasoner, report and regulator will thank you.