Re: URIs may not mean *exactly* what you think

I'll try and be a bit more positive ...

Pete Johnston wrote:

> Quoting Jeremy Carroll <jjc@hplb.hpl.hp.com>:
> 
> 
>>So the concept of URI ownership in the semantic web is based on two 
>>fallacies:
>>
>>1) you can distinguish a definition by the owner from a description by 
>>someone else
>>
>>2) the meaning of one URI is independent of other URIs (possibly not yet 
>>coined)
> 
> 

We have the same issues for natural language: words do not have owners, 
or authorative definitions; the meaning of one word depends intimately 
on the meaning of other words; the nearest thing we have to definitons, 
dictionaries, in fact are lists of descriptions of words.

Despite this, we manage to have shared meaning.

> Having read your example, I find it hard to argue with this... 
> 
> But is it not in contradiction with the principles presented in the W3C TAG Web 
> Architecture document? 
> 
> Specifically
> 
> 
>>URI ambiguity refers to the use of the same URI to refer
>>to more than one distinct resource.
>>
>>Good practice: URI ambiguity
>>
>>Avoid URI ambiguity.
> 
> 
> http://www.w3.org/TR/webarch/#pr-uri-ambiguity
> 

That paragraph is pretty weak.
It doesn't discuss the problems of using the same URI to refer to 
abstract resources, which, through HTTP content negotiation may be 
realised as an english PDF document or a Japanese web page. It's legit, 
and I expect it is discussed elsewhere in that doc, but the suggestion 
to "Avoid URI ambiguity" expects that we can recognise URI ambiguity 
when we see it.
Saying that different users of dc:creator may mean slightly different 
things by it, and may mean different things than the consumers of the 
documents that use dc:creator, is not saying that dc:creator is 
ambiguous, more that meaning is complex. We accept that even a plain 
HTML page will appear different to different users because that have 
different user agents.



> And doesn't it undermine the prospects for the use of shared vocabularies?
> 
> I accept that in the context of my music database I may use the dc:creator 
> property to represent the relation between a Song and a Songwriter, in my e-
> Prints database I may use dc:creator property to represent the relation between 
> a Document and its Author, somewhere else its the relation between a Collection 
> and its Collector, and so on. 
> 
> But in all cases, the URI http://purl.org/dc/elements/1.1/creator denotes the 
> same concept, the same resource, and that is the relation between resource and 
> its creator which the owner of that URI (the Dublin Core Metadata Initiative) 
> says the URI denotes, and for which DCMI serves an authoritative representation 
> in the form of an RDF/XML document.
> 
> If it doesn't, I'm worried ;-)
> 

I don't think DCMI are in charge here - they thought about it, and 
suggested a term, other people use it, in ways that DCMI did not 
imagine. The community gives the term meaning.

> Pete

Jeremy

Received on Thursday, 6 May 2004 16:24:43 UTC