Define "schema mapping"

[matentzn]

Preferred name: schema mapping

Alternative names: schema crosswalk (narrow), data mapping

Definition:

A category of mapping that specifies how individual data elements or values in a source dataset correspond to elements in a target dataset, typically used in data integration, migration, or transformation tasks.

[matentzn]

Would be great if we could cite a more official, canonical definition.

[maweber-bia]

see the google doc for clarification between "mapping/matching" and "alignment": FAIR Data Mapping Taxonomy (2025)

[jkesanie]

In reference to the question posed by @sveinugu in #47:

"Question: Does this assume that the schemas are described as linked data entities/relationships? In reality, most schemas are not. Example: none of the schemas from the standards by the Global Alliance for Genomics and Health (GA4GH) are defined in terms of linked data entities and relationships."

If look at the "Semantic artefact" definition given in [1], there is requirement for them to be represented using linked data entities/relationships that would be identified with URIs and would resolve to vocabulary/ontology. From that perspective it could appropriate to combine ontology and schema mapping, but I do think that we need a way to distinguish between the two.

1. Corcho, O., Ekaputra, F.J., Heibi, I. et al. A maturity model for catalogues of semantic artefacts. Sci Data 11, 479 (2024). https://doi.org/10.1038/s41597-024-03185-4

[sveinugu]

Sorry for discussing in issues, continuing here. Context is this comment: #47 (comment)

For discussion, here is an example of a schema defined as part of a GA4GH standard, the Data Repository Service (DRS): https://ga4gh.github.io/data-repository-service-schemas/preview/release/drs-1.2.0/docs/#tag/DrsObjectModel

Fields in this schema are not defined with IRIs. So if I understand you correctly, the DRS schema in question is not within the definition of a "semantic artifact"?

[nichtich]

I'd say no: schemas can have many forms and linked data is just one. It could be an XML Schema, an SQL Schema or any data format. Many schema mappings are not defined in declarative form (RDF, CSV, formal diagram...) but in imperative form given as executable transformation to convert from source schema to target schema.