Re: Migrating from slash to fragment

KangHao Lu (Kenny) wrote:
> 
> On 2010/05/18, at 19:52, Nathan wrote:
> 
>> KangHao Lu (Kenny) wrote:
>>> On 2010/05/16, at 5:00, Nathan wrote:
>>>> Toby Inkster wrote:
>>>>> On Fri, 14 May 2010 17:53:50 +0100
>>>>> Nathan <nathan@webr3.org> wrote:
>>>>>> I'm wondering if there are any recommended paths for migrating RDF 
>>>>>> or specifically an ontology from slash to fragment URIs (?)
>>>>> Cool URIs don't change.
>>> Indeed. But several reasons we might want a canonical way to change 
>>> URIs:
>>> - Slash URI requires more HTTP round trips
>>> - For documents we have 301
>>> - I personally don't like this owl:sameAs culture. Try to use only 
>>> one URI for each thing could encourage cross domain links
>>> TimBL uses the term http://www.w3.org/2000/10/swap/pim/contact# and has
>>> """
>>> tim:i con:preferredURI "http://www.w3.org/People/Berners-Lee/card#i".
>>> """
>>> in his FOAF.
>>
>> I thought the use of 
>> http://www.w3.org/2000/10/swap/pim/contact#preferredURI was a rather 
>> nice touch tbh, noted it yesterday and have just implemented on a 
>> client site (because it makes a new foaf for them if they don't yet 
>> have one, but in the future they may get a foaf, thus i want to point 
>> to their preferred webid when they make their own).
>>
>>> I think this can be generalized and it shouldn't be con:preferredURI 
>>> but something like link:preferredURI.
>>
>> @prefix link: <http://www.w3.org/2000/10/swap/pim/contact#> .
>>
>> ?
>>
>> can't see what difference changing the prefix makes when it'd still 
>> point to the same property; and if you mean make another identical 
>> property, then why?
> 
> Yes I meant an "identical" property. The rdfs:comment of 
> con:preferredURI says
> 
> "A string which is the URI a person, organization, etc, prefers that 
> people use for them"
> 
> so it doesn't apply to general situations, such as ontology migration. 
> (or maybe etc. includes everything? Notice this ontology is for contact 
> information)
> 
> 
> The link: namespace (http://www.w3.org/2006/link# ) was something TimBL 
> created. There's link:Document in it, which is probably equivalent to 
> foaf:Document.

Ahh I'd never seen that one before, note when I deref I only find 
link:uri and link:obsoletes - can't see the Document?

link uri looks useful & from what I can tell, looks like it's a much 
better property to use in many situations where owl:sameAs is used atm - 
opinion?

> link:Document is a fundamental principle in Linked Data, so should this 
> link:preferredURI be, I think.
> 
> (I am totally against to having something as fundamental as "Information 
> Resource" in the FOAF ontology)

fwiw I feel the same about Document in foaf, keep thinking of using 
dc:Text but seems a bit ambiguous & can't see anything that's a closer 
match.

>>> For example, foaf:maker and dc:creator are considered equivalent in 
>>> the FOAF spec. So in the RDF ontology of FOAF, maybe there should be
>>
>> they're quite different, dc:creator is typically used with a string 
>> term, dcterms:creator is pushing towards uri's rather than literals 
>> but it's still a grey area, whereas foaf:maker is always a uri of a 
>> Person.
> 
> Notice that in the example below I used the namespace for *DC terms*. To 
> justify what I did, notice that if you look up 'dc' in 
> http://prefix.cc/ the first his is "http://purl.org/dc/terms/" not 
> "http://purl.org/dc/elements/1.1/". You should never identify things 
> without complete URIs :)
> 
> Or is there a third namespace? Then I completely misunderstood all these.
> 
> I don't want to argue whether foaf:maker is equivalent to 
> dcterms:creator or not. This is out of topic. But FYI,
> "foaf:maker owl:equivalentProperty dcterms:creator" is in the FOAF spec 
> already. But again, for an authoring agent or myself, which one to use 
> is still very confusing.

+1 re confusing, I've taken to using both; redundant but I figure not 
everybody parsing the linked data have reasoners so good to put in 
something many understand.

>>> """foaf:maker link:preferredURI "http://purl.org/dc/terms/creator". """
>>> in the ontology.
>>
>> could be something, but ontology wise I get a feeling you'd only do 
>> this if you'd deprecated a feature, and in that scenario possible 
>> dcterms:isReplacedBy would be a more suitable property?
> 
> Well, I think if you are doing this, you are not deprecating the 
> feature. You are simply saying "If you want to use this feature, I 
> prefer using this URI...."
> 
> I thought about your idea. But generally,
> 
> "<A> link:preferredURI B."  implies "<A> owl:sameAs <B>. "
> 
> 
> so similarly if dcterms:isRelacedBy is used for URI migrating
> 
> "<A> dcterms:isReplacedBy <B>" should imply "<A> owl:sameAs <B>".
> 
> So we can infer that "<A> dcterms:isRelacedBy <A> . "
> This sounds weird in human language. But I am not totally against this. 
> We shouldn't rely on human language too much anyway.

after your previous pointer, link:obsoletes looks quite good..

Best,

Nathan

Received on Tuesday, 18 May 2010 12:02:06 UTC