Re: Formalism to generate a URI as an object?

Ah, o.k., this[1] has indeed undergone significant changes and my comments have become indeed moot.

Thanks!

Ivan

[1] http://www.w3.org/2001/sw/rdb2rdf/r2rml/


On Nov 5, 2010, at 03:27 , Souripriya Das wrote:

> Ivan,
> 
> In the latest Editor's draft, we have removed the <> from IRI values and _: from the blank node values. Instead, a new property, rr:termtype is used to designate the type of RDF term (IRI, BlankNode, or Literal).
> 
> Thanks,
> - Souri.
> 
> On 11/1/2010 11:48 AM, Ivan Herman wrote:
>> I am not sure whether this is a comment pointing at an issue in the document, or whether it is my own misunderstanding, in which case this is a question rather than a comment...
>> 
>> Looking at the example A.2.2, the mapping to a logical table maps the original table to cells that include a URI using the NTriple '<...>' syntax. E.g., it maps the 'job' column to the jobTypeURI column with the cell becoming <xyz.com/emp/job/CLERK> in the example. Then the mapping specification includes
>> 
>> rr:propertyObjectMap [ rr:RDFTypeProperty emp:jobtype; rr:column "jobTypeURI" ]
>> 
>> which means that the final triple will use  xyz.com/emp/job/CLERK as URI for an RDF class.
>> 
>> What bothers me is that there seems to be a conceptual mixture of a particular syntax of RDF URI Resources (namely NTRiples) that encloses the URI in the '<' and '>' characters, and the fact that rr:RDFTypeProperty automatically maps on an RDF Resource whose URI is xyz.com/emp/job/CLERK. This should not be the case.
>> 
>> Although the example does not have that, what if I wanted to generate the extra triple
>> 
>> <xyz.com/emp/7369> ivan:bla <xyz.com/emp/job/CLERK> . 
>> 
>> My reading of the spec is that what I would have to do is 
>> 
>> rr:propertyObjectMap [
>>     a rr:IRIMap ;
>>     rr:property ivan:bla ;
>>     rr:column "jobTypeURI" 
>> ]
>> 
>> But that is then confusing: the value of the jobTypeURI cell is the string '<xyz.com/emp/job/CLERK>', so the system has to know that it has to strip the '<' and '>' characters to produce the Resource with the URI 'xyz.com/emp/job/CLERK'. Or is it so that the system always and automatically understands the syntax of the form '<xyz.com/emp/job/CLERK>' to be a string for an RDF URI Resource? What if I want to store, in my graph, the <xyz.com/emp/job/CLERK> as a Literal? What is then the usage and the reason of having rr:IRIMap?
>> 
>> I think the correct way is if the logical table mapping does
>> 
>>    ('xyz.com/emp/job/'|| job) AS jobTypeURI
>> 
>> the system has enough information both for the rr:IRIMap class and the rr:RDFTypeProperty provides enough information for the system to know that a URI resource should be created. The '<' and the '>' characters should not be used.
>> 
>> I have the same issue with the way blank node ID's are explicitly created using the '_:' syntax. Again, if one uses the rr:BlankNodeMap then the system knows that the identifier in the cell must be used to generate a blank node. Mixing the NTRiple syntax into the mapping is unnecessary for my taste. 
>> 
>> Ivan
>> 
>> 
>> ----
>> Ivan Herman, W3C Semantic Web Activity Lead
>> Home: 
>> http://www.w3.org/People/Ivan/
>> 
>> mobile: +31-641044153
>> PGP Key: 
>> http://www.ivan-herman.net/pgpkey.html
>> 
>> FOAF: 
>> http://www.ivan-herman.net/foaf.rdf
>> 
>> 
>> 
>> 
>> 
>> 
>>   
>> 
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf