Finding a concept¶
Three ways to discover whether CORA already covers the field your pipeline needs. Pick whichever fits the question you're holding.
By known name¶
If you already know the canonical name — postal_code, street_address, email_address — go straight to crosswalks/concepts/<concept>.yaml. One file per concept; the filename is the concept name.
By alias¶
If you're carrying a vendor's field name (zip_code, Zip_Postal_Code, postcode), the crosswalks declare aliases inside the YAML. A repository grep against the aliases: block is the fastest path:
Or, programmatically:
import yaml
from pathlib import Path
target = "zip_code"
for path in Path("crosswalks/concepts").glob("*.yaml"):
cw = yaml.safe_load(path.read_text())
if target in cw.get("aliases", []) or cw["concept"] == target:
print(path, "->", cw["concept"])
Every concept's primary name plus its aliases are searchable this way.
By coverage matrix¶
If the question is which fields can my pipeline rely on across the sources I receive, open the coverage matrix. Concepts down the left, standards across the top, confidence indicators in the cells. The matrix answers cross-source feasibility questions faster than reading individual crosswalks.
A not_present cell tells you a source genuinely can't answer the question that concept asks; build your pipeline accordingly.
By browsing the generated concept pages¶
For visual inspection, the auto-generated concept pages at docs/generated/concepts/ render each crosswalk as a Markdown table plus a Mermaid graph. Useful when you want to evaluate a concept's mapping without opening the YAML.
These are mechanical projections — the YAML is authoritative. Pipelines should always read the YAML.
When the concept isn't covered¶
CORA today publishes thirty-three concept crosswalks, and the corpus grows with the standards bodies' work and the analyzer-driven editorial passes. Before requesting a crosswalk, check the suggestions report and the semantic suggestions report — the concept may already be a known candidate cluster awaiting an editorial pass. If the field your organization needs is neither covered nor surfaced, file the request. The mapping work is openly maintained; most additions are merged within one review cycle.
What to read next¶
- Reading a crosswalk
- The full YAML shape — confidence labels, version tracking, narrative notes.
- Integrating CORA
- Wiring the lookup pattern above into a production pipeline.