Transformation: Distribution, LandingPage and Documentation

[hallinpihlatie]

Would it be possible to simplify a bit how the XSLT decides whether an online resource goes to Distribution, Landing Page or Documentation? Now the XSLT is doing it in all cases based on both on the protocol and function, if I've understood it correctly. Conterra have altered the XSLT so at least the protocols WMS and WFS go to the Distribution element directly. Anyhow, the need to add the function:Download exist in many cases and this is challenging as it is not widely implemented as far as I know.

Could it be simplified somehow, for example so that function is only needed if one wants to indicate that it the link should go to Documentation (protocol:Web Address + funtion: Information). If a Web Address exist alone it would go to landing page and if some of the other protocols (see figure) are used they would go to Distribution.

Is it too late to bring this up on Monday @jescriu. And please do correct me if I'm wrong and function isn't needed,

[laers]

I have a related comment: In XSLT, this logic is used in the mapping from ISO's online resource to DCAT-AP's dcat:Distribution:

It is via ISO's OnlineFunctionCode that the different types that determine the mapping to dcat:accessURL, foaf:page and dcat:landingPage:

<cit:CI_OnLineFunctionCode codeList="http://standards.iso.org/iso/19115/resources/Codelists/cat/codelists.xml#CI_OnLineFunctionCode" codeListValue="download"/>

Notice that dcat:landingPage is identified by missing OnlineFunctionCode. This is by no means optimal. In any case, it is not sufficiently supported by data catalogue platforms today. The option to select an empty code was not intended by ISO 19115, which is understandable.

The mapping to dcat:landingPage needs a revision - a suggestion could be that OnlineFunctionCode = Information is mapped to dcat:landingPage and OnlineFunctionCode = Search is kept mapped to foaf:page?

[cnlarsen]

Also a somewhat related issue: If an ISO dataset doesn't contain any resources that map to dcat:distribution, then the metadata on, e.g., license and format is lost, as this metadata is represented at the level of dcat:distribution. This of course speaks to a more fundamental incompability between the data models, but is also a case with practical implications which I think deserves special mention.

[jescriu]

I new mapping proposal - involving the elements mentioned in this issue - is going to be presented in the pilot meeting on 7 July.
https://github.com/INSPIRE-MIF/GeoDCAT-AP-pilot/tree/main/meetings/2026-07-07/T-1B-data-services-apis

Open for comments / feedback until 15 September 2026