DCAT - schema.org alignment (profile)

[dr-shorthair]

Prepare a formal alignment of DCAT with schema.org

Also see #65

[dr-shorthair]

This is a separate issue to prepare the alignment.

65 concerns guidance on use of DCAT in weakly axiomatized environment, which probably has a dependency on the alignment/mapping, but is a distinct concern.

Relevant commentary copied over from #65

@andrea-perego commented on 20 Jan • Some work has been carried out in the framework of SDW that can turn to be relevant here:

https://www.w3.org/2015/spatial/wiki/ISO_19115_-_DCAT_-_Schema.org_mapping

The specified mappings were based on existing work, including a mapping of DCAT-AP, GeoDCAT-AP, and StatDCAT-AP to Schema.org:

https://ec-jrc.github.io/dcat-ap-to-schema-org/

@makxdekkers commented on 7 Feb This would be very useful for tools that could create the relevant schema.org markup for a landing page for the dataset based on the DCAT description.

@andrea-perego commented on 8 Feb • This would be very useful for tools that could create the relevant schema.org markup for a landing page for the dataset based on the DCAT description.

The DCAT-AP to Schema.org mapping exercise I mentioned earlier in this thread includes also SPARQL queries implementing such mappings:

https://github.com/ec-jrc/dcat-ap-to-schema-org/tree/master/sparql

@dr-shorthair commented on 1 Mar I propose adding an RDF alignment document for DCAT-schema.org to enable this discussion

@dr-shorthair commented on 7 Mar More mappings from SDWWG and Project Open Data added in #146

@dr-shorthair commented on 23 Mar Preliminary DCAT-schema.org alignment here: https://github.com/w3c/dxwg/blob/gh-pages/dcat/rdf/schema.ttl

[dr-shorthair]

RDF file renamed to dcat-schema.ttl Work underway in branch /dcat-alignments-schema/ - see https://github.com/w3c/dxwg/blob/dcat-alignments-schema/dcat/rdf/dcat-schema.ttl

[dr-shorthair]

DCAT element	mappings	to schema.org
dc:description	owl:equivalentProperty	schema:description
dc:title	owl:equivalentProperty	schema:name
dct:description	owl:equivalentProperty	schema:description
dct:format	[ owl:inverseOf rdfs:subPropertyOf ]	schema:encodingFormat
dct:identifier	owl:equivalentProperty	schema:identifier
dct:issued	owl:equivalentProperty	schema:datePublished
dct:language	owl:equivalentProperty	schema:inLanguage
dct:license	owl:equivalentProperty	schema:license
dct:modified	owl:equivalentProperty	schema:dateModified
dct:publisher	owl:equivalentProperty	schema:publisher
dct:spatial	owl:equivalentProperty	schema:spatial
	owl:equivalentProperty	schema:spatialCoverage
dct:temporal	owl:equivalentProperty	schema:datasetTimeInterval
	owl:equivalentProperty	schema:temporal
	owl:equivalentProperty	schema:temporalCoverage
dct:title	owl:equivalentProperty	schema:name
dct:type	owl:equivalentProperty	schema:additionalType
dcat:Catalog	owl:equivalentClass	schema:DataCatalog
dcat:DataService	owl:equivalentClass	schema:DataFeed
		Not quite sure if a DataFeed is a data service, or a data collection. From a REST viewpoint there is no difference. But other APIs support additional queries, slices, etc which make the characterization of a service more efficient than listing all the (potentially infinite) resources available from it.
dcat:Dataset	owl:equivalentClass	schema:Dataset
dcat:Distribution	owl:equivalentClass	schema:DataDownload
dcat:Resource	rdfs:subClassOf	schema:Thing
dcat:accessURL	rdfs:subPropertyOf	schema:contentUrl
	schema:domainIncludes	dcat:Distribution , schema:DataDownload
	schema:rangeIncludes	rdfs:Resource , schema:URL
dcat:byteSize	rdfs:subPropertyOf	schema:contentSize
	schema:domainIncludes	dcat:Distribution , schema:DataDownload
	schema:rangeIncludes	rdfs:Literal , schema:Text
dcat:catalog	schema:domainIncludes	dcat:Catalog , schema:DataCatalog
	schema:rangeIncludes	dcat:Catalog , schema:DataCatalog
dcat:contactPoint	owl:equivalentProperty	schema:contactPoint
	schema:domainIncludes	, dcat:Dataset , dcat:DataService , schema:Dataset
dcat:dataset	owl:equivalentProperty	schema:dataset
	schema:domainIncludes	dcat:Catalog , schema:DataCatalog
	schema:rangeIncludes	dcat:Dataset , schema:Dataset
dcat:distribution	owl:equivalentProperty	schema:distribution
	schema:domainIncludes	dcat:Dataset , schema:Dataset
	schema:rangeIncludes	dcat:Distribution , schema:DataDownload
dcat:downloadURL	rdfs:subPropertyOf	schema:contentUrl
	schema:domainIncludes	dcat:Distribution , schema:DataDownload
	schema:rangeIncludes	rdfs:Resource , schema:Thing
dcat:keyword	rdfs:subPropertyOf	schema:keywords
		dcat:keyword is singular, schema:keywords is plural
	schema:domainIncludes	dcat:Resource , dcat:Dataset , dcat:DataService , schema:Dataset
	schema:rangeIncludes	rdfs:Literal , schema:Text
dcat:landingPage	rdfs:subPropertyOf	schema:url
	schema:domainIncludes	dcat:Resource , dcat:Dataset , dcat:DataService , schema:Dataset
	schema:rangeIncludes	foaf:Document , schema:WebPage
dcat:mediaType	owl:equivalentProperty	schema:encodingFormat
	schema:domainIncludes	dcat:Distribution , schema:DataDownload
	schema:rangeIncludes	dct:MediaTypeOrExtent
dcat:record	schema:domainIncludes	dcat:Catalog , schema:DataCatalog
	schema:rangeIncludes	dcat:CatalogRecord
dcat:service	schema:domainIncludes	dcat:Catalog , schema:DataCatalog
	schema:rangeIncludes	dcat:DataService
dcat:theme	owl:equivalentProperty	schema:about
	schema:domainIncludes	dcat:dcat:Resource , dcat:Dataset , dcat:DataService , schema:Dataset
	schema:rangeIncludes	skos:Concept , schema:Class
dcat:themeTaxonomy	schema:domainIncludes	dcat:Catalog , schema:DataCatalog
	schema:rangeIncludes	skos:ConceptScheme
foaf:Organization	owl:equivalentClass	schema:Organization
foaf:homepage	owl:equivalentProperty	schema:url
foaf:mbox	owl:equivalentProperty	schema:email
		and potentially of interest to be mixed-in for scientific data descriptions
sosa:hasFeatureOfInterest	rdfs:subPropertyOf	schema:object
sosa:hasResult	rdfs:subPropertyOf	schema:result
sosa:madeByActuator	rdfs:subPropertyOf	schema:instrument
sosa:madeBySampler	rdfs:subPropertyOf	schema:instrument
sosa:madeBySensor	rdfs:subPropertyOf	schema:instrument
sosa:observedProperty	owl:equivalentProperty	schema:variableMeasured
sosa:phenomenonTime	rdfs:subPropertyOf	schema:temporalCoverage
sosa:resultTime	rdfs:subPropertyOf	schema:endTime
sosa:usedProcedure	owl:equivalentProperty	schema:measurementTechnique

Some items struck through because they are marked 'superseded' in schema.org.

[dr-shorthair]

Given the interest in Google Dataset Search, should we put this into 2PWD to trigger feedback?

[PeterParslow]

I'm pleased you're looking at this. Could you supersede https://www.w3.org/2015/spatial/wiki/ISO_19115_-_DCAT_-_Schema.org_mapping when this is mature enough? Perhaps just putting something on that wiki draft to point to this issue.

Because it's that older wiki draft which comes up top when I search for "DCAT vs Schema.org"

[larsgsvensson]

The aligments dct:format owl:equivalentProperty schema:encodingFormat and dcat:mediaType owl:equivalentProperty schema:encodingFormat imply dct:format owl:equivalentProperty dcat:mediaType if owl:equivalentProperty is transitive.

[danbri]

my notes from yesterday towards a mapping tested by applying it to turning Google Dataset Search documentation into DCAT RDF is in https://twitter.com/danbri/status/1055787656318214144

The python notebook runs a SPARQL CONTEXT query, based on consulting @dr-shorthair 's term mappings:

PREFIX s: <http://schema.org/>
PREFIX dcat: <http://www.w3.org/ns/dcat#>
PREFIX dct: <http://purl.org/dc/terms/>

CONSTRUCT { 
  ?d a dcat:Dataset ; dct:title ?t ; dct:description ?desc .
  ?d dcat:distribution [a dcat:Distribution ; dct:format ?ef ] .
  ?d dct:license ?lic .
  ?d dct:temporal ?tempC .
} WHERE { 
  ?d a s:Dataset; s:name ?t ; s:description ?desc . 
  ?d s:distribution [ s:encodingFormat ?ef ] .
  OPTIONAL { 
  ?d s:license ?lic . 
  }
  OPTIONAL {
  ?d s:temporalCoverage ?tempC .  
  }
}

It would be interesting to get at least one such query blessed by the WG. One motivation for doing so would be to communicate to Google what the WG thinks a DCAT version of a particular example ought to look like.

This is in the opposite direction to https://ec-jrc.github.io/dcat-ap-to-schema-org/#formal-definition-sparql-mapping-properties-dataset but the same basic approach.

[kcoyle]

Following on @larsgsvensson comment (and maybe this should be in its own github issue) all of the one-to-many equivalencies are implying that the "many" have the same semantics, which probably is not the case or schema.org would not have different properties. One could look at using subproperties here if that fits. @dr-shorthair

dct:spatial	owl:equivalentProperty	schema:spatial
	owl:equivalentProperty	schema:spatialCoverage
dct:temporal	owl:equivalentProperty	schema:datasetTimeInterval
	owl:equivalentProperty	schema:temporal
	owl:equivalentProperty	schema:temporalCoverage

p.s. this has given me renewed respect for the library practice of "use for" between terms, as this to me is closer to what is meant with a vocabulary cross-walk. It doesn't mean that the terms are semantically equivalent, it just means that you substitute one for the other even if it isn't exact in meaning. Cross-walks usually are semantically lossy to some degree.

[kcoyle]

I have another question about the 1-to-many entries. Wouldn't they only work in the direction of many-to-one? You can translate schema:datasetTimeInterval and schema:temporal (etc.) to dct:temporal, but if you are going in the direction from DCAT to schema.org, which of the three schema.org properties would you translate dct:temporal to?

It seems that if the two vocabularies in a cross-walk are not symmetrical then you actually need two cross-walks (or some clever use of the table format) to show both directions. You may decide that you don't need both directions, but in that case the one-to-many properties should be reduced to one-to-one from DCAT to schema.org.

[dr-shorthair]

Thanks @kcoyle -

Concerning the one-to-many relations - schema:spatial is supersededBy schema:spatialCoverage, and both schema:temporal and schema:datasetTimeInterval by schema:temporalCoverage so the older forms are not included in the document, only in the issue thread here.

[dr-shorthair]

I've fixed the incorrect equivalent class axioms, and relaxed some of the stronger alignments from equivalent to skos:closeMatch - see https://github.com/w3c/dxwg/pull/609

[davebrowning]

It looks to me that we have completed this, at least for DCAT2 - or is there something more to be done? (It could always be re-visited....) Looks like this would be better in the 'DCAT Alignment' milestone?

[dr-shorthair]

Indeed. Closing.