Re: Wikidata export in RDF

On 2012-08-06, at 13:01, Peter F. Patel-Schneider wrote:

> Hi:
> 
> The general approach appears to be acceptable.
> 
> I would like to have some use of the base relationship in the encoding.   As it stands right now, there is no relationship between p:Population and anything in the encoding, meaning that users cannot reliably infer the relationship between the encoding and the base relationship.  Of course, human users can make guesses (perhaps by looking inside the IRIs), but this does not work for systems.
> 
> It looks as if rdf:label is a tyop for rdfs:label.  However, I don't think that rdfs:label should be used here.  I think that it would be even better to use some other property here.  Note, in particular, that adding provenance should change an rdfs:label property constructed in this way.
> 
> I would suggest using a <=0 restriction instead of an all restriction to the empty class for the no value SNAK, even though they have the same semantics.
> 
> I suggest not using the RDF encoding for OWL restrictions.
> 
> peter
> 
> PS:  I believe that the xsd:date format is YYYY-MM-DD, not DD-MM-YYY.

Yes, it's YYYY-MM-DD.

Also xsd:integer might be a better choice than xsd:int, to make comparisons with other data easier. xsd:integer is a supported type in SPARQL, xsd:int is not. If you're publishing data in Turtle, then you can just write bareword integers, e.g. 3499879 and they will be xsd:integers.

I find the use of predicates with a capital after the namespace a bit jarring (e.g. q:Method), but it's perfectly legal of course.

- Steve

-- 
Steve Harris, CTO
Garlik, a part of Experian
+44 7854 417 874  http://www.garlik.com/
Registered in England and Wales 653331 VAT # 887 1335 93
Registered office: Landmark House, Experian Way, Nottingham, Notts, NG80 1ZZ

Received on Monday, 6 August 2012 12:17:52 UTC