Re: Is ontology an information resource?

Hi,

Short answer: A vocabulary (or ontology) *is* an information resource.

Longer answer: You correctly point out that the documents cited below  
advocate the use of 303 redirects on vocabulary URIs. However, this  
has nothing to do with the question wether vocabularies are  
information resources or not.

Note that the 303 status code has existed in HTTP since long before  
the Semantic Web. A 303 status code simply means: “The server is  
unable or unwilling to send you a representation of the resource you  
asked for. Instead, it points you to some related information.” There  
are many reasons why the server might be unable to provide a  
representation. One of them is that the resource is not an  
information resource, but this is not the only possible reason for  
setting up 303 redirects.

But in our case, the reason is a different one. It has to do with  
content negotiation on hash URIs. (Note that the recipes you cite  
deal *only* with hash URIs.)

According to the specifications, if <myVocabulary> serves HTML, then  
<myVocabulary#myTerm> identifies a part of that HTML document. If  
<myVocabulary> also servers RDF via content negotiation, then  
<myVocabulary#myTerm> identifies whatever the RDF document says, for  
example a term in my vocabulary.

Some people (including me) see this is a contradiction, a URI cannot  
identify a section of an HTML document and a vocabulary term at the  
same time. Hence the 303 redirect at <myVocabulary>. It's simply a  
trick that lets us avoid serving HTML at <myVocabulary>. Instead, we  
redirect to <myVocabulary.html> and serve the HTML there. This  
removes the contradiction, <myVocabulary#term> identifies now  
whatever the RDF says, while the HTML section is identified by  
<myVocabulary.html#term>.

In summary: A 303 redirect does *not* mean that the URI identifies a  
non-information resource. It just means that the URI *possibly* might  
identify a non-information resource. In this case, however, it does  
not. The 303 redirect here is just a trick to avoid a semantic  
ambiguity arising from content negotiation.

Finally, I should point out that W3C's Dan Connolly has gone on  
record saying that perhaps the ambiguity in this case is not such a  
big deal, and will possibly be fixed in the future by a simple change  
to the specifications. So, there's also some precedence for *not*  
setting up 303 redirects in cases where we have content negotiation  
on hash URIs.

Best,
Richard


On 15 Oct 2007, at 05:39, Yoshio Fukushige wrote:

>
> Hi all,
>
> I (still again?) have some problems in my attempt to publish an  
> ontology on the Web.
>
> So please let me solve them by giving your advices.
> Let me ask questions one by one, for some of them might be  
> automatically solved by
> preceding answers.
>
> My first question is as follows
>
> QUESTION:  Is an ontology an information resource or non- 
> information resource?
>
> Reto Bahmann-Gmür said in his message[1] an ontology can be seen as
> an information resource, and Tim BL said in his message[2]
> [[
> when any term in that namespace S looked up, it is the
> document <http://ont.example.org/myOnt/example> which is dereference.
>
> That resource is the ontology.
> ]].
>
> So it seems there is a consensus that an ontology is an information  
> resource.
>
> In "Best Practice Recipes for Publishing RDF Vocabularies"[3],  
> however,
> Recipe 3, 4, 5 shows
> that a Client will receive a 303 response for a request for  
> 'vocabulary URI',
> where the document says in its Introduction
> [[
> In this document, vocabulary and ontology are used interchangeably.
> ]] ,
> which together with the TAG's httpRange-14 resolution[4] might imply
> that an ontology is a non-information resource...
>
> Chris Bizer's "How to Publish Linked Data in the Web"[5] also shows an
> example for a client to request a 'vocabulary URI' and get 303  
> response,
> though what 'vocabulary' means is not explicitly defined in that  
> document.
>
> And Leo Sauermann's 'Cool URIs for the Semantic Web'[6] shows an
> example where a request for http://www.acme.com/about#alice (and
> http://www.acme.com/about) will get 303 redirect (in section 4.2),
> which makes me assume http://www.acme.com/about is a non-information
> resource. But is it not an ontology?
>
> Many thanks in advance
>
> [1] http://lists.w3.org/Archives/Public/semantic-web/2007Jul/0129.html
> [2] http://lists.w3.org/Archives/Public/semantic-web/2007Jul/0132.html
> [3] http://www.w3.org/TR/swbp-vocab-pub
> [4] http://www.w3.org/2001/tag/issues#httpRange-14
> [5] http://sites.wiwiss.fu-berlin.de/suhl/bizer/pub/LinkedDataTutorial
> [6] http://www.dfki.uni-kl.de/~sauermann/2006/11/cooluris
> Best,
> Yoshio Fukushige
>
>
> -- 
> Yoshio Fukushige <fukushige.yoshio@jp.panasonic.com>
> Network Development Center,
> Matsushita Electric Industrial Co., Ltd.
>
>
>
>

Received on Monday, 15 October 2007 06:07:07 UTC