XKOS Best Practices: skos:notation - datatype and multiple notations

[ChLaaboudi]

XKOS Best Practices recommends to use the datatype "xmlns:string" for skos:notation.
We use by default "rdf:PlainLiteral". In some cases, we had also to define additional notations and datatypes for some items reused in different context.
As examples,

• CN 2022 (https://showvoc.op.europa.eu/#/datasets/ESTAT_Combined_Nomenclature,_2022_%28CN_2022%29/data), which contains a CN and HS Coding codes.
• Geonomenclature (https://showvoc.op.europa.eu/#/datasets/ESTAT_Geonomenclature_2023_%28GEONOM_2023%29/data), which includes 3 notations (numeric, ISO alpha-2 and code used in dissemination).

What would be the best practice for managing multiple notations and their datatypes?

Christine LAABOUDI-SPOIDEN (Eurostat B.1.)

[tfrancart]

Hello @ChLaaboudi !

Have you noted that the recommendation for using skos:notation with datatype xsd:string is for the ConceptScheme ? not for the Concepts. It is intended to capture "the short name of the classification, with no language tag.".

For example:

<http://example.org/codes/isic4>
   a skos:ConceptScheme ;
   skos:prefLabel "International Standard Industrial Classification of All Economic
   Activities (ISIC), Revision 4"@en ,
   skos:notation "ISIC4" ;

Using skos:notation with multiple datatypes on the Concepts is indeed possible, even encouraged.

BTW I see that some example in the best practices guide are using a language tag on skos:notation, which should not happen, I will file a separate issue.

[ChLaaboudi]

Hi @tfrancart,
Indeed, what I have mentioned applies to skos:Concept as well as skos:Collection.

[tfrancart]

We acknowledge that SKOS proposes to use different datatypes for skos:notation. This is still valid and allowed in XKOS for classification items. We will make that explicit when we add a table of properties for skos:Concept as discussed in #179.