DCAT - ISO 19115 alignment (profile)

[dr-shorthair]

Prepare a formal alignment of DCAT to ISO 19115 (Geospatial metadata)

[andrea-perego]

@dr-shorthair , I wonder which formal language we could use for aligning with DCAT non-RDF vocabularies.

For GeoDCAT-AP, this has been done by using XSLT. Should we go this way?

[nicholascar]

There is a semi-official OWL interpretation of ISO19115 (https://github.com/ISO-TC211/GOM/tree/master/isotc211_GOM_harmonizedOntology/19115-1/2014) that we could use to show alignment however it's not very good in that the OWL was auto-created from UML XML files and doesn't follow best practice OWL modelling. We could still use that version to describe alignment though precisely because it is a direct UML (the standard) to OWL conversion.

[dr-shorthair]

ISO 19150-2 defines the UML->OWL conversion rule. As part of the rule it prescribes URIs for all classes and properties, which can be considered 'official' as far as ISO/TC 211 is concerned.

So in the SSN vocabulary we used the ISO 19156 O&M to support an axiomatization of the alignment or crosswalk - see https://www.w3.org/TR/vocab-ssn/#OM_Alignment

(I also have my doubts about the quality of the OWL encoding, but that should not prevent us from using these URIs as a convenient shorthand for the elements of the UML model, which is the canonical representation.)

[dr-shorthair]

Nevertheless, since the XML encoding is used as the reference by many installations it would make sense to provide a pathway. However, if heading that direction you'll probably need to consider 10139, 19139-2, 19115-3 ... I haven't kept up with the progress there so I don't know if they vary much.

[nicholascar]

Why? Do we really think any XML-only 19115 users will want to see DCAT in XML?

I suggest that 19115 users interested in the future and DCAT and Semantic Web generally will be happy with a DCAT to OWL 19115 mapping as they will be mentally or technically mapping their ISO UML to OWL.

[smrgeoinfo]

Actually, having a reviewed set of xpaths for mapping from ISO19139 and ISO19115-3 XML to DCAT would be very useful. I haven't gotten around to building the xslt for the transform yet, but have done ISO19139 to Schema.org, DataCitev4 to Schema.org, and qualifiedDC to ISO19139. Like @dr-shorthair I have reservations about the automated (ISO19150-2) UML to OWL generated ISO and don't see much uptake for that. I think the collections of vocabularies being developed by dxwg are a better bet to replace ISO19139 in the rdf world; that's why I'd like to keep things conceptually in sync.

[andrea-perego]

@smrgeoinfo , the GeoDCAT-AP XSLT is here:

https://github.com/SEMICeu/iso-19139-to-dcat-ap

[nicholascar]

I will soon (by March) need to develop a DCAT-to-ISO19115-3 mapping for clients. I expect to use a Python script to query an rdflib module, in-memory graph of a DCAT document and then to render ISO XML using Jinja2 templating. This is due to the client already using lots of rdflib and Jinja2 and not wanting to use XSLT.

This work might be of wider interest, for instance it would be able to be placed on top of - an extension to - the CKAN DCAT module (updated for DCAT revised of course) and export ISO19115-3 from there.

[nicholascar]

I can confirm that I have a requirement for a March delivery here: one govt agency will use CKAN to store DCAT rev records and will then need to transform those to (a profile of) ISO19115-3 for consumption by another agency.

[dr-shorthair]

@andrea-perego is the update of GeoDCAT complete?

[andrea-perego]

@dr-shorthair asked:

@andrea-perego is the update of GeoDCAT complete?

Yep. The new version of GeoDCAT-AP is aligned with DCAT 2:

https://semiceu.github.io/GeoDCAT-AP/releases/2.0.0/

[dr-shorthair]

So I think that means that this ticket is now superseded, and can be closed, right?

[andrea-perego]

+1 from me.

[andrea-perego]

Noting no objections, I am closing this issue.