W3C home > Mailing lists > Public > public-esw-thes@w3.org > May 2012

Re: UDEF Representation in RDF

From: Chris Harding <c.harding@opengroup.org>
Date: Wed, 23 May 2012 14:21:38 +0100
Cc: All UDEF Interested Parties <udef@opengroup.org>, Simon Spero <sesuncedu@gmail.com>, public-esw-thes@w3.org, "richard.parent" <richard.parent@servicesquebec.gouv.qc.ca>, Alan Doniger <alan.doniger@gmail.com>
Message-Id: <55262CC2-B994-43FA-B213-B14A55C41211@opengroup.org>
To: rreck@rrecktek.com
Hi, Ronald -

IRIs would certainly be more general than URIs. Isn't there a problem, though, that we would want to use the IRI to locate resources in HTTP requests, but this is not allowed for in the HTTP specification?

> 6. An HTTP GET request to the base URI shall return a set of RDF statements about the vocabulary. (The precise nature of these statements is to be determined.)

I guess that a way round this is to allow URIs as identifiers, but have the resources retrieved by HTTP requests that use URIs that encode the IRIs. I'm not familiar with the detailed issues here. Do you know whether this is normal practice?

I don't think that either UDEF or SKOS distinguishes between different kinds of narrowing. (I stand subject to correction on this.)

Regards,
Chris
++++

Chris Harding
c.harding@opengroup.org



On 23 May 2012, at 13:37, Ronald P. Reck wrote:

> 
> 
> On 05/23/2012 05:42 AM, Chris Harding wrote:
>> Hi -
>> 
>> There hasn't been anything new on this discussion for a while, so it's time to make some conclusions. Unless anyone objects, here's what I propose we do.
>> 
>> 1. The core UDEF, and each associated UDEF vocabulary, shall have a base representation in RDF, plus other representations that will be mechanically derivable from the base representation and available for the convenience of applications that use particular representation formats. The derived representations shall include RDFS Class, SKOS, XML, and HTML representations.
>> 
>> 2. Each node of each UDEF vocabulary shall have its own URI. This shall consist of a base URI that identifies the vocabulary, the # character, and a fragment that identifies the node. For an object class node, the fragment shall be the string "UDEF-" followed by the UDEF object class id. For a property node, the fragment shall be the _ character followed by the UDEF property id.
>> 
>> 
> 
> Some may consider it a small point but I have found it better to use IRI's instead of URI's for SPARQL compatibility.
> 
> 
> 
> 
>> 8. The SKOS representation of a vocabulary shall consist of the core representation plus a set of statements that define each node as a SKOS concept and relate each node to its parent by the skos:broader property. The representation shall also include statements defining the vocabulary itself as a SKOS concept scheme, the nodes as concepts of that scheme, the root nodes as top concepts of that scheme, and the labels of each node as SKOS Preferred Lexical Labels in their respective languages. Where a node in the core UDEF has a connected node in another vocabulary, the representations of the core UDEF and the other vocabulary shall include SKOS broader and narrower statements relating the two nodes, with the core UDEF node as the narrower concept..
> 
> It might be useful to consider the distinction between different types of narrowings such as ISA or partitive. If UDEF already has this distinction there can conceivably be a loss of information.
> 
Received on Wednesday, 23 May 2012 13:22:11 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 2 March 2016 13:32:15 UTC