RE: Arrays in SKOS: proposal

> From: Miles, AJ (Alistair) [mailto:A.J.Miles@rl.ac.uk] 
> Sent: Thursday, May 13, 2004 11:41 AM
> Subject: Arrays in SKOS: proposal
> 
> By default, without using arrays, the ordering of narrower 
> concepts is not preserved by RDF applications.  Therefore, in 
> any situation where the ordering of a set of concepts is 
> significant, an array should be used to express that 
> ordering.  The array may be given a label, but this is not necessary.
> 
> The skos:Array construct is recommended for encoding the 
> grouping of a set of concepts according to some 
> 'characteristic of division'.  In the majority of cases of 
> the use of characteristics of division, the ordering is 
> signficant.  Applications handling a skos:Array should treat 
> the ordering as significant by default.  
> 
> However, in some cases the ordering of concepts under a 
> characteristic of division is not significant.  It is 
> recommended that a SKOS encoding of a thesaurus does not 
> introduce ordering information where such information was not 
> a part of the original thesaurus.  In these cases, the fact 
> that the ordering is irrelevant should be expressed by adding a
> (skos:ordered,'false') property to the array.
> 

I'm not clear as to why RDF collections were not used vs. the
skos:ordered property.  It seems to me that whether there is
ordering or not maps onto rdf:Bag and rdf:Seq which are both
collections.  If you didn't want to use rdf:Bag and rdf:Seq
you could develop the appropriate SKOS representation and
mark them in the RDFS as collections.  Can you clarify why
RDF collections are not being used?


Thanks, Andy.

Andrew Houghton, OCLC Online Computer Library Center, Inc.
http://www.oclc.org/about/
http://www.oclc.org/research/staff/houghton.htm

Received on Thursday, 13 May 2004 12:10:43 UTC