Re: One example

On 5/17/15 4:00 PM, Holger Knublauch wrote:
> Hi Karen,
> walking through the requirements there appear to be some duplicates
> (apologies if I misunderstand them):

Not entirely sure I understand it myself, so if we can't figure it out 
I'll have to catch someone from Europeana who can help us. Note that 
they do all of their work in XML, and currently validate using 
Schematron. I'm sure this colors their view of the data. Note, too, that 
the comments that are in the RDFXML are not carried forward to the TTL, 
so it is probably best to read the requirements in the context of the 
RDFXML file. Why don't we settle on one or two sets of requirements, and 
I'll try to catch someone tomorrow who can clarify the data more 
generally. The first one seems fairly simple:

> Duplicate 1:
>   * #1 For every CulturalHeritage object (CHO) the URI of the
>     ProvidedCHO and the URI described in edm:aggregatedCHO must be the same
>   * #6 Each EDM CHO must have exactly one instance of rdf:type
>     edm:ProvidedCHO and one instance of rdf:type ore:Aggregation
>     (DC:R-225) (W3:R-6) (W3:R-12.2)

The first one does not limit cardinality to "1", the second does. The 
first checks that the subject URI's are the same:


> Duplicate 2 (maybe subtle differences, hard to understand for me):
>   * #2 For every CHO or ore:Proxy dc:coverage or dc:subject or dc:type
>     and dcterms:spatial; dc:language for text,dc:title or
>     dc:description, must be present (DC:R-68) (W3:R-5.2)

#2 isn't clear to me, so let's skip it for now.

>   * #13 if edm:ProvidedCHO or ore:Proxy is present, at least one
>     dc:title or dc:description should be present, one dc:subject or
>     dc:type or dc:coverage or dcterms:spatial and edm:type should be
>     present (DC:R-71) (W3:R-6) (W3:R-6.2) (W3:R-6.7)

#13 seems to be almost clear:

for each (edm:ProvidedCHO, ore:Proxy)
    required: at least one dc:title or one dc:description [can it have 
both? I'll ask]
    required: at least one dc:subject or dc:type or dc:coverage or 
dcterms:spatial [more than one of any? will ask]
    required: edm:type

>   * #14 For each edmProvidedCHO or ore:Proxy (provider proxy)
>     dc:coverage or dc:subject or dc:type or dcterms:spatial,dc:language
>     (for text objects), dc:title or dc:description or edm:type must be
>     present (DC:R-67) (W3:R-6.2) (W3:R-6.7)

For #14, I need to figure out what edm:type modifies, but we do know that:
     valid values for edm:type are: (VIDEO, SOUND, IMAGE, TEXT)

For the rest, this is how far I've gotten:

for each (edm:ProvidedCHO, ore:Proxy)
     required: dc:coverage or dc:subject or dc:type or dcterms:spatial
     if resource type is text, "<edm:type>TEXT</edm:type>") <!-- need to 
discover how "resource type" is defined -->
         required: dc:language, dc:title, dc:description or edm:type 
<!-- which "or"? -->

> Duplicate 3:
>   * #3 We must have only one PrefLabel per language. Ideally the
>     language tag should be present (DC:R 49) (W3:R-6.4)
>   * #5 We must have only one PrefLabel per language. Ideally the
>     language tag should be present (DC:R 49) ((W3:R-6.4)

Those are within different graphs. I don't know if that matters in your 
design, but one is in the graph <edm:Agent> (in the RDFXML), the other 
in the graph <record>. But since we had to remove <record> because it 
isn't defined, I'll have to come back with a corrected data set for this 

In the TTL file, these show up as:

   rdf:type edm:Agent ;
   dc:date "1767" ;
   dc:identifier "#SMS-MM:N234546" ;
   rdaGr2:dateOfBirth """c
1730""" ;
   rdaGr2:dateOfDeath """p
1783""" ;
   skos:prefLabel """Edmond
Saunier""" ;
   foaf:name """Saunier,
       Edmond""" ;
   rdf:type skos:Concept ;
   skos:broader <> ;
   skos:prefLabel "Mandolin" ;
   skos:prefLabel "Mandolin"@en ;
   skos:prefLabel "Mandolin"@sv ;
   skos:prefLabel "Mandoline"@de ;
   skos:prefLabel "Mandoline"@fr ;
   skos:prefLabel "Mandoline"@nl ;
   skos:prefLabel "Mandolino"@it ;

In this case, I'm thinking that the skos:prefLabel rule can be 
universal/general -- that it should apply wherever there is 
skos:prefLabel. That would fit in with our use case 21 [1], and it's a 
good candidate for a template/macro that would be of general use.


> Some more questions:
> - For #8 "Check that the edm:hasView, edm:isShownBy and edm:object
> resolves to a media", does this mean that the values of those properties
> must have a certain rdf:type (e.g. edm:WebResource) or does it mean that
> the system shall do a web look-up to check for a valid HTTP response?
> - The page at "#9 The value of edm:rights must come from the list at
>" only seems to
> contain HTML - is there an RDF version that lists the valid values? How
> else can this be checked?
> Thanks
> Holger
> On 5/17/2015 13:03, Karen Coyle wrote:
>> OK, the example should work now -- at least, it passes validation in
>> the RDF validator and rapper. It appears to have 3 "records" worth of
>> data, so there are multiple "outer" graphs. If that's a problem, I can
>> reduce it down to a single "outer" graph.
>> kc
>> On 5/16/15 7:20 PM, Karen Coyle wrote:
>>> OK, I'll see if I can find anyone who can answer that, since this isn't
>>> my data. I'm thinking there's something basic going wrong - such as the
>>> software not finding the ontology.
>>> Meanwhile, here's the link to the latest version of the ontology:
>>> I'll try to get back to you soon.
>>> kc
>>> On 5/16/15 5:05 PM, Holger Knublauch wrote:
>>>> Also, the RDF/XML file
>>>> doesn't seem to be parse, and produces many errors, see
>>>> Holger
>>>> On 5/17/2015 8:02, Karen Coyle wrote:
>>>>> In case anyone wants a real example to work with, I have created a
>>>>> short page that lists the requirements and the file of data those
>>>>> requirements pertain to. The requirements themselves are also embedded
>>>>> in the XML file so you can see what graph they are associated with.
>>>>> While it may be the case that no one wants to tackle this at this
>>>>> time, it does include some common use cases. The list includes both DC
>>>>> and SHACL requirements numbers, and links to the spreadsheet with DC
>>>>> requirements, that then in turn links to further information.
>>>>> kc

Karen Coyle
m: 1-510-435-8234
skype: kcoylenet/+1-510-984-3600

Received on Monday, 18 May 2015 00:49:12 UTC