DD3 R7. Where possible, add Schema.org “corresponding element” entries to GEMINI elements

[PeterParslow]

We have used W3C’s recommendations for mapping from ISO 19115 to Schema.org. This table summarises the Schema.org equivalence statements given for each element below.
Whilst there is no specific DD2 recommendation concerning DCAT, we believe a DCAT2 “equivalent element” for each GEMINI element would be useful, by supporting those whose web publication of GEMINI records uses DCAT as opposed to Schema.org. Where this is easily available from the same W3C source, we have included this below. You will see that the two vocabularies are very similar, but note that:
• some of the DCAT elements sit in the DCAT “distribution” section, not their “dataset”;
• many DCAT properties have structured content, so this is not a complete list of how to implement it; and
• there are many other DCAT properties that should also be used, beyond those that exist in Schema.org (e.g. conformsTo, creator, spatialResolutionInMeters, format).

GEMINI element	Condition	Schema.org	DCAT/DCAT2[1]	Notes
Title		name	dct:title
Dataset language		inLanguage	dct:language
Abstract		description	dct:description
Topic category		keywords	dct:subject
Keyword	INSPIRE theme	keywords	dcat:theme / dct:subject
Keyword	free text	keywords	dcat:keyword	Schema.org puts all the ‘free text’ keywords in one value
Keyword	Controlled list, URL	Keywords.DefinedTerm.name Use .description for the textual content of the Anchor or CodeList Use .url for the target of the Anchor	dcat:keyword.DefinedTerm
Temporal extent		temporalCoverage[2]	dct:temporal
Dataset reference date	19115 dateType = publication	datePublished	dct:issued release date / issued
Dataset reference date	19115 dateType = revised	dateModified	update date / dct:modified
Lineage			dct:provenance
Extent		spatialCoverage.Place.name	dct:spatial
Resource locator.linkage	19115 function = download	contentURL (inside “distribution”)	dcat:downloadURL
Resource locator.linkage	19115 function = “information” Where the page links on to download		dcat:accessURL
Resource locator.linkage	19115 function = “information”	url	dcat:landingPage
Data format		encodingFormat	dct:format,	Possibly also dcat:mediaType
Responsible organisation	19115 role = publisher	publisher.Organization (with at least name, email, url)	dct:publisher
Responsible organisation	19115 role = pointOfContact	contactPoint (probably Organisation, with at least name, email, url)	dcat:contactPoint
Use constraints	Use constraints is being used to indicate a license	license	dct:license
	Where GEMINI has an Anchor URL to the licence	licence.CreativeWork .abstract (with the free text) and .url (with the Anchor target URL)
Use constraints	Other circumstances		dct:accessRights
Bounding box		spatialCoverage.geo.GeoShape.box	dct:spatial	Note: needs translating from four edges to two corners
Resource identifier		identifier	dct:identifier
Resource type			rdf:type	Note: DCAT-AP does not distinguish between datasets and dataset series

[PeterParslow]

The W3C mapping, on which this is largely based, is at https://www.w3.org/2015/spatial/wiki/ISO_19115_-_DCAT_-_Schema.org_mapping

[PeterParslow]

Andrea Perego’s ISO 19139 - DCAT mapping in GitHub (James’ link) provides more detail e.g. the range of each element, and also maps somethings outside the DCAT namespace(s).

https://github.com/GeoCat/iso-19139-to-dcat-ap/blob/master/documentation/Mappings.md

(Thanks to James Reid)

[PeterParslow]

Just been contacted by the CDDO data standards team looking to state how to describe "where" in DCAT metadata to be used in the UK government data marketplace. This will include updating the mapping above for DCAT v3.

See co-cddo/ukgov-metadata-exchange-model#1

[nmtoken]

Should we also adapt the GEMINI mapping to DCAT 3 as this now includes better description of dataset series?

[PeterParslow]

That will be a necessary part of the CDDO work; I'll make sure it is available as an update to this GEMINI change request. It's also being discussed (& likely to happen) in the OGC GeoDCAT SWG.

[PeterParslow]

Need to annotate this to show how it aligns (or not!) with the UK Cross-Government Metadata Exchange Model which may be re-branded as a UK Application Profile of DCAT

[archaeogeek]

@PeterParslow to update table, then @archaeogeek to update elements with equivalent mappings, also publish this table as guidance

[PeterParslow]

We'll also need to include guidance or at least comment on converting GEMINI to DCAT covering how many dcat distributions to create (depending on e.g. GEMINI Use constraints & Resource locators).

Revised table, with extra columns for DCAT v3 & UK government metadata exchange model. Note, the UK Gov work is supposed to consider adding spatial & some other things; they also plan to convert it to a full AP of DCAT v3.

GEMINI element	Condition	Schema.org	DCAT/DCAT2[1]	Notes	DCAT3	UK Gov MXM
Title		name	dct:title		Y	Y
Dataset language		inLanguage	dct:language		Y	N
Abstract		description	dct:description		Y	Y
Topic category		keywords	dct:subject		Y	N
Keyword	INSPIRE theme	keywords	dcat:theme / dct:subject	DCAT3 expects theme to be used when the target is a SKOS concept; subject in the more general case, whether or not the term is from a controlled vocab	Y	dcat:theme
Keyword	free text	keywords	dcat:keyword	Schema.org puts all the ‘free text’ keywords in one value; DCAT / MXM keyword are 'uncontrolled' literals	Y	Y
Keyword	Controlled list, URL	Keywords.DefinedTerm.name Use .description for the textual content of the Anchor or CodeList Use .url for the target of the Anchor	dcat:keyword.DefinedTerm		dcat:theme?	dcat:theme
Temporal extent		temporalCoverage[2]	dct:temporal		Y	N proposed
Dataset reference date	19115 dateType = publication	datePublished	dct:issued release date / issued		Y	Y
Dataset reference date	19115 dateType = revised	dateModified	update date / dct:modified		Y	Y
Lineage			dct:provenance		Uses PROV	N
Extent		spatialCoverage.Place.name	dct:spatial if available as a link		Y	N proposed
Resource locator.linkage	19115 function = download	contentURL (inside “distribution”)	dcat:downloadURL		Y	Y
Resource locator.linkage	19115 function = “information” Where the page links on to download		dcat:accessURL of a dcat:Distribution?		Y	N
Resource locator.linkage	19115 function = “information”	url	dcat:landingPage		Y	N
Data format		encodingFormat	dct:format of a dcat:Distribution	Possibly also dcat:mediaType	Y	N
Responsible organisation	19115 role = publisher	publisher.Organization (with at least name, email, url)	dct:publisher		Y	Y
Responsible organisation	19115 role = pointOfContact	contactPoint (probably Organisation, with at least name, email, url)	dcat:contactPoint	dcat:contactPoint is a vCard	Y	must contain email & contactName (organisation)
Use constraints	Use constraints is being used to indicate a licence	license	dct:license	license is a property of a distribution	Y	Y licence
Use constraints	Where GEMINI has an Anchor URL to the licence	licence.CreativeWork .abstract (with the free text) and .url (with the Anchor target URL)			Y	Y
Use constraints	Other circumstances		dct:accessRights	accessRights is a property of the dataset	Y	Y
Bounding box		spatialCoverage.geo.GeoShape.box	dct:spatial.dct:Location.dct:bbox	Note: needs translating from four edges to two corners	Y	N
Resource identifier		identifier	dct:identifier		Y	Y
Resource type			rdf:type	cataloguedResource is either Dataset or DataService; Note: DCAT-AP does not distinguish between datasets and dataset series; DCATv3 does	The CataloguedResource can be either Dataset, DatasetSeries, or DataService	Y

[archaeogeek]

@PeterParslow what do I need to do next? I can't remember...

[PeterParslow]

@PeterParslow what do I need to do next? I can't remember...

See if what I've come up with in a desk exercise matches what you'd expect from the GeoNetwork implementation of DCAT?

[nmtoken]

@archaeogeek do you have a link to where this transformation is mapped in GeoNetwork 4. It is available (in theory) though the OGC API - Records interface, though links aren't working for us

[archaeogeek]

@nmtoken it's not the mapping. We have it working here: https://spatialdata.gov.scot/geonetwork/api/collections/main/items/fa510351-8e30-4147-b984-862be84a6f90. You need to check the log files- I suspect you're missing the relevant xsl files in https://github.com/geonetwork/geonetwork-microservices/tree/main/modules/services/ogc-api-records/src/main/resources/xslt/ogcapir/formats/copy (which is completely undocumented). Basically you need a gemini one that matches the iso19139 one

[nmtoken]

Not the headers then (geonetwork/geonetwork-microservices#114) ?

[archaeogeek]

@nmtoken the above is all I had to do to get it working, YMMV.

[nmtoken]

@archaeogeek Just checking we are not talking at cross purposes, you seem to be saying that in your Tree Preservation Orders - Argyll and Bute example the fact that the schema.org, dcat, dcat_turtle, and geojson tabs link to content is becuase you have a gemini XSL file and we don't.

For us (for example https://metadata.bgs.ac.uk/geonetwork/api/collections/main/items/a2b1143b-5c5d-23d6-e054-002128a47908) and the EEA geospatial data catalogue (for example https://sdi.eea.europa.eu/catalogue/api/collections/main/items/71c47f78-27b6-4080-acd5-47b306b273d8) these tabs don't give any content (only errors).

[archaeogeek]

@nmtoken to be precise, what I'm saying is that the only change we ever needed to make to get a full set of working links in the ogc-api service is to add the gemini xsl record, eg adding a iso19139.gemini23 equivalent of the files here: https://github.com/geonetwork/geonetwork-microservices/tree/main/modules/services/ogc-api-records/src/main/resources/xslt/ogcapir/formats/copy, which is identical to the ones already included. This does trigger an error in the ogc-api service logs if you dig deep enough.

[PeterParslow]

I no longer know what I meant by dcat:keyword.DefinedTerm! It's not something that seems to exist in DCAT (v2 or v3). Conveniently, that makes it clearer (to me) that dcat:keyword if for uncontrolled ones and dcat:theme is for controlled lists (on the assumption they are published as SKOS)