W3C home > Mailing lists > Public > semantic-web@w3.org > February 2012

Re: [http-range14] how to publish RDF for Information Resources

From: Andrea Splendiani (RRes-Roth) <andrea.splendiani@rothamsted.ac.uk>
Date: Fri, 17 Feb 2012 18:21:43 +0000
To: Yang Squared <yang.square@gmail.com>
CC: "<sergio.fernandez@fundacionctic.org>" <sergio.fernandez@fundacionctic.org>, "<david@dbooth.org>" <david@dbooth.org>, "<leo.sauermann@gnowsis.com>" <leo.sauermann@gnowsis.com>, "<semantic-web@w3.org>" <semantic-web@w3.org>
Message-ID: <F14C7A94-7B64-43A9-9DC4-F982DCE55835@rothamsted.ac.uk>
Hi,

scenario (1): ok. In may also happen to have a third identifier at time, for the resource "per se", independently of its representation.
Note that the *.rdf and *.html resources are distinct information resources which refer to the same information content.

scenario (2):
You have some data, and some metadata (let's say: creator of an html page). If for some specific reason, you need metadata in rdf (and not RDFa), then probably a link is the best solution.
But just to avoid confusion: this is not the general case. In general nothing prevents to put you metadata in the html itself (indeed... "meta" tags remind of some old intensions). In fact, you may often find a copyright statement in the page which is essentially metadata.
So, if what you propose comes from the fact that you RDF is for metadata only, I think that is one possible use of it, not the general case.

Another thing ti note: nothing says that the document http://example.com/data/rdf should contain statements about http://example.com/data/rdf. It could very well contain statements about the html: http://example.com/homepage.html .

In conclusion: if you need this separation of metadata in RDF, I would go for a link (or RDFa-ish).
Otherwise in general I would add metadata in both the html and the rdf representation. If you are talking specifically about an html file, I would add metadata about it (html)in the rdf file. This may be a bit questionable at first, but at second I think it makes sense. You don't want infinite assertion of meta-* statements ;)

ciao,
Andrea


Il giorno 17/feb/2012, alle ore 17.59, Yang Squared ha scritto:

I see your point now. Thx, Andrea.

I just realise that there may be two scenarios:

1) RDF and HTML are just the representation of the same resource

Assume we have a weather report http://example.com/weatherreport which has two representation in HTML (human readable representation) and RDF (machine readable representation)

In this case, the content of HTML and RDF should be consistent.

We can use then using content negotiation to serve these two representations for the same URI.

2) RDF describes a HTML Web Page Resource

For example

I have a resource about my homepage http://example.com/homepage.html
I have a metadata RDF about someone created this homepage http://example.com/data/rdf

In this case, these two URIs are not referring to the same thing.

According to David Booth and Leo Sauermann, we should use <link> tag in HTMl or not widely known HTTP link header (not accepted proposal).

I hope these are all cases. Please correct me if I am wrong.

Yang





On Fri, Feb 17, 2012 at 5:18 PM, Andrea Splendiani (RRes-Roth) <andrea.splendiani@rothamsted.ac.uk<mailto:andrea.splendiani@rothamsted.ac.uk>> wrote:
Wait...

http://www.test.com/test.html <- this is a resource, and has it's own URI (1)
http://www.test.com/test.rdf <- this is another resource, and has it's own (distinct) URI (2)

Now, what I am saying is that in the RDF you can also predicate on the the URI (1).
So if you ask information about (1), requiring a result in rdf, it seems reasonable to return (2) with predicates (eventually) on (1).
This is not really incorrect, though it may be unpractical for some.

But perhaps I don't get your distinction between page and metadata.

Metadata on the page can also be in the html representation itself.
In RDF I'm asking for a machine readable representation of the content of the page, rather than it's metadata only.

Do you mean something else by metadata ?

ciao,
Andrea



Il giorno 17/feb/2012, alle ore 17.03, Yang Squared ha scritto:

Hi Andrea,

I think using content negotiation is incorrect, the homepage is a resource need a URI, and the metadata RDF is another resource need another URI. Otherwise, we are saying the homepage and the metadata of the homepage is the same thing.

The representation of the homepage is the HTML page
the representation of the metadata of the homepage is the RDF

Regards,
Yang


On Fri, Feb 17, 2012 at 4:24 PM, Andrea Splendiani (RRes-Roth) <andrea.splendiani@rothamsted.ac.uk<mailto:andrea.splendiani@rothamsted.ac.uk>> wrote:
Hi,

I'm using content negotiation.
In most of the cases, the "information resource" is actually something in a database or stored somewhere. So representing it in html or rdf via content negotiation seems plausible.

In case the information resource in html is what you want to predicate (like, you have a collection of html files that you are serving) the rdf returned via content negotiation could include predicates about the html file uri.
If you are asking for the html file resource, this is always in html by definition. Requesting an rdf implies anyway requesting something else.

ciao,
Andrea


Il giorno 17/feb/2012, alle ore 02.21, Yang Squared ha scritto:

Hi all,

I have a Web architecture question here.

Assume I have a information resource URI  http://example.com/homepage.html

I would like to publish a RDF metadata (http://example.com/data/homepagerdf) about this information resource (e.g. homepage isCreatedBy steve). What publishing mechanism can I use?

since http://example.com/homepage.html is an Information Resource, when dereferencing it, we should get that homepage.html document returned. How can we possible redirect to a RDF?

Content negotiation can use to serve two different representation of the resource, but both representation is for the same resource. So we cannot use it.

303 can redirect one information resources to another information resource, e.g. http://example.com/homepage.html --303--> http://example.com/data/homepagerdf --200-->RDF

but in this way, when I dereferencing the original http://example.com/homepage.html it did not result as a homepage.html itself and got a RDF. So there is a paradox here.

Can anyone please suggest anything? Or the conclusion is that the RDFa (or by using the link element to RDF) is the only way to publish RDF metadata for information resources?

I am writing a paper and I would like to conclude that there will be no case that a hashURI publishing mechanism and 303 redirection can be used for Information Resource to publish RDF metadata. Do you have any object case?

------------------------------
One may recommend me to use RDFa. However, I consider that the RDFa is not ideal solution to publish Linked Data at all.
First of all, embedding metadata together with data prohibits the independent curation of data and metadata. Secondly, following the principles of the Web Architecture, any distinct resource of significance should be given a distinct URI, but in this approach a single URI is used to identify two information resources. In general, the RDFa embedded metadata approach can be replaced by using the <link> element href in XHTML to pointing to an external RDF document, where the rel=”meta” attribute can be used to indicate a relationship between resources.

Thanks a lot,
Yang Yang

-----------------------------------

Web and Internet Science

Room 3027 EEE Building

Electronics and Computer Science

University of Southampton, SO17 1BJ


Tel: +44(0)23 8059 8346<tel:%2B44%280%2923%208059%208346>

twitter: @yang_squared<http://twitter.com/#!/yang_squared>




--

-----------------------------------

Web and Internet Science

Room 3027 EEE Building

Electronics and Computer Science

University of Southampton, SO17 1BJ


Tel: +44(0)23 8059 8346<tel:%2B44%280%2923%208059%208346>

twitter: @yang_squared<http://twitter.com/#!/yang_squared>





--

-----------------------------------

Web and Internet Science

Room 3027 EEE Building

Electronics and Computer Science

University of Southampton, SO17 1BJ


Tel: +44(0)23 8059 8346<tel:%2B44%280%2923%208059%208346>

twitter: @yang_squared<http://twitter.com/#!/yang_squared>





--

-----------------------------------

Web and Internet Science

Room 3027 EEE Building

Electronics and Computer Science

University of Southampton, SO17 1BJ


Tel: +44(0)23 8059 8346

twitter: @yang_squared<http://twitter.com/#!/yang_squared>
Received on Friday, 17 February 2012 18:22:32 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:47 GMT