Here's the result of the sanity check for D1 graph. Btw, if you say
yes for item #12, I will go ahead and make a small addition to the
ontology.
Non-literal value of rdfs:label ==> value of rdfs:label should be
literal, so even though it's a URI, it needs to be given as literal,
preferably explicitly typed.
For example, the following triple was caught by sanity check:
Found usage of glbase:hasIdentifierValue with non literal value ==>
glbase:hasIdentifierValue property is a datatype property so it should
point to a literal value of type xsd:string.
Found usage of glbase:hasLandingPage with non literal value ==>
glbase:hasLandingPage property is a datatype property so it should
point to a literal value of type xsd:anyURI.
Found triples that use Chinese text in the value of description
property ==> I did not attempt to detect the language tag, so found
them coincidentally. It's a good practice to put a language tag if you
haven't put one :-)
==> We don't have any property named
http://schema.geolink.org/1.0/base/main#address in the ontology and
there is no property that can be used to point to organization's
address. We can add one to the ontology, and if we do, I would suggest
that we use <http://schema.geolink.org/1.0/base/main#hasPhysicalAddress
with domain Organization and range xsd:string. Note that though I'm
not sure if other data providers have organizations' physical address
in their triples.
What do you think?
With regards geometry information, I found the following stats:
geolink:hasGeometryAsWktLiteral occurrences: 116257
geolink:Dataset instances: 160675
geolink:Dataset instances that uses geolink:hasGeometryAsWktLiteral: 116256
values for geolink:hasGeometryAsWktLiteral correctly typed: 0
So, there are approx. 44 thousands instances of Dataset class that do
not use hasGeometryAsWktLiteral property. I didn't check whether a
wktLiteral is attached to them via some other property.
Regarding the correct literal typing, the literal should be given as, e.g.,
But since Yingjie said that the map interface silently assumes that
the value for hasGeometryAsWktLiteral property is always a wktLiteral,
then my remark regarding literal typing above is really just a minor
one.