RE: Multiple RDF documents

Hi Richard,

Thanks for your reply. I understand what you mean. A fully qualified URI
based on inscheme propery as mentioned by Ceri would solve the issue

<skos:Concept rdf:about="
http://hiltm2m.cdlr.strath.ac.uk/hiltm2m/schemes/GCMD.rdf#944">

  <skos:prefLabel xml:lang="en">Soils</skos:prefLabel> 

  <skos:broader rdf:resource="
http://hiltm2m.cdlr.strath.ac.uk/hiltm2m/schemes/GCMD.rdf#888" /> 

  <skos:narrower rdf:resource="
http://hiltm2m.cdlr.strath.ac.uk/hiltm2m/schemes/GCMD.rdf#945" /> 

With regard to the second question, it is entirely upto what is retrieved as
a result of search. I have seen SKOS records with only reference to related
concepts, but I am not sure about the performace with 100 extra queries to
retrieve 100 related concepts if client want to display the name of these
related concepts. Any way, I am getting more ideas about the whole thing and
thanks once again for your time

Anu

> -----Original Message-----
> From: Richard Light [mailto:richard@light.demon.co.uk] 
> Sent: 26 February 2008 09:16
> To: Anu Joseph
> Cc: public-esw-thes@w3.org
> Subject: Re: Multiple RDF documents
> 
> In message <00df01c877cb$3259c6f0$09bb9f82@pcanuj>, Anu 
> Joseph <anu.joseph@strath.ac.uk> writes
> >
> >Hi all,
> >
> >We have been using SKOS to model our data within our HILT 
> project and 
> >there are potential issues with the current model. I am not 
> an expert 
> >in SKOS and I wonder if someone could advice me regarding this.
> >
> >A snippet of existing model is like this
> >
> >http://hiltm2m.cdlr.strath.ac.uk/hilt4/get_filtered.xml
> >
> >There are two potential problems with this model
> >
> >1. A possible id clash within different schemes. As the 
> concepts from 
> >different vocabularies are listed in the same rdf document, 
> referring 
> >to a concept #888 might not be necessarily unique, 
> especially when we 
> >search across multiple vocabularies managed by different 
> organizations.
> 
> I'm an expert neither in SKOS nor in RDF, but surely, in 
> general, you shouldn't be using URI fragment identifiers 
> alone to refer to concepts which come from more than one 
> scheme?  I would expect each scheme to have a base URI which 
> is declared in the scheme itself, and which could be implied 
> for internal rdf:about attributes.  However, as soon as any 
> of this data is exported from its "home" scheme, all of its 
> URIs should be expressed in full, by prepending the base URI 
> to each fragment.
> 
> To put the point another way, a URI fragment identifier 
> without an associated base URI is meaningless.
> 
> >2. Parsing issue - When these concepts are retrieved as a result of 
> >search, there is no way to identify whether a particular concept has 
> >been pulled out as a result of real search or it is listed 
> because it 
> >is a related concept of a search result. (In the above 
> example, there 
> >is no way to identify if the concept #888 has been retrieved 
> because of 
> >search or because it is the broader term for concept #944)
> 
> Again, pardon my ignorance, but doesn't this depend on what 
> is retrieved as the result of a search?  Where is this defined?
> 
> Richard Light
> --
> Richard Light
> SGML/XML and Museum Information Consultancy richard@light.demon.co.uk
> 
>

Received on Wednesday, 27 February 2008 12:05:10 UTC