W3C home > Mailing lists > Public > public-lod@w3.org > January 2013

Re: Linked Data & RDFa

From: Kevin Ford <kefo@3windmills.com>
Date: Sat, 19 Jan 2013 18:49:50 -0500
Message-ID: <50FB311E.4080305@3windmills.com>
To: public-lod@w3.org
 > So - questions. How much of the linked data cloud is expressed in some
 > variant HTML+RDFa alongside RDF/XML, Turtle etc.?
-- All of the data at ID.LOC.GOV is embedded as RDFa in the HTML and via 
content-negotiation.  It's merely a molecule in the cloud these days.

 > When/if you do so,
 > are you holding some data back and keeping it only in the
 > machine-oriented dumps, or including it in the RDFa?
-- Trying not to hold anything back in the RDFa.  We've endeavoured to 
make the RDFa in the HTML as rich as what you would get accessing the 
RDF directly.  That said, I cannot state unequivocally that the RDFa is 
as complete as what you get when accessing the RDF/XML, n-triples, etc 
directly.  If not, I imagine it is close, very close.

 > Are you finding
 > it hard to generate RDFa from triple datasets because it's 'supposed'
 > to be intermingled with human text?
-- It certainly adds to development time, more so if the data uses 
complex structures.  It's quite manageable, but those not steeped in 
this stuff will have a learning curve.  Simple helps.  When trying to 
imagine your average web designer, however, I could see him understand 
the basic idea and only include RDFa (or microdata) up to a point (that 
is, until the means outstrip the end for the developer with respect to 
his work contract).  It's not that I expect the average developer to 
find the RDFa or microdata a problem as much as the concept of adding 
additional elements to HTML that are not seen.  [ Let me be clear - I'm 
speaking of the average developer.  The developer that works 9-5, isn't 
subscribed to any W3C lists, and probably does not follow W3C types of 
activities closely at all. ]

 > Did/does Microdata confuse the
 > picture?
-- I would imagine so, but I am merely guessing based on my 
understanding of human nature.  It certainly leads to the question, 
asked by many new to this, "Which one is better?"  (The reply to which, 
incidentally, probably favours the respondent's preference and possibly 
his limited knowledge of the differences.)

 > I have acquired the impression somehow that in the Linked Data scene,
 > people lean more towards the classic 'a doc for the humans, another
 > for the machines' partitioning model.
-- I think what you are detecting is that which is easiest for data 
consumers.  Namely, accessing the data directly, without HTML as a 
carrier.  Personally, as a data consumer, if I have the option to access 
the RDF via RDF/XML, n3, or n-triples directly without having to spend 
time worrying about the accuracy and validity of the embedded RDFa (or 
microdata), I'm going to use the RDF/XML, n3, etc.  I don't need 
something designed primarily for humans, nor the flotsam that comes 
along with it.  (Now, if the world were to stop providing direct access 
to data and it *always* had to be embedded in HTML, I suspect we would 
see all the available data in the HTML, accurately described, and 
near-perfect validity. :) )

Just one person's view.

Yours,
Kevin


On 01/18/2013 05:23 AM, Dan Brickley wrote:
> With RDFa maturing (RDFa 1.1, particularly Lite), I wanted to ask here
> about attitudes to RDFa.
>
> I have acquired the impression somehow that in the Linked Data scene,
> people lean more towards the classic 'a doc for the humans, another
> for the machines' partitioning model. Perhaps this is just a
> consequence of history; digging around some old rdfweb/foaf
> discussions[1] I realise just how far we've come. RDFa wasn't an
> option for a long time; but it is now.
>
> So - questions. How much of the linked data cloud is expressed in some
> variant HTML+RDFa alongside RDF/XML, Turtle etc.? When/if you do so,
> are you holding some data back and keeping it only in the
> machine-oriented dumps, or including it in the RDFa? Are you finding
> it hard to generate RDFa from triple datasets because it's 'supposed'
> to be intermingled with human text? What identifiers (if any) are you
> assigning to real-world entities? Dataset maintainers ... as you look
> to the future is RDFa in your planning? Did/does Microdata confuse the
> picture?
>
> I'm curious where we are with this...
>
> Dan
>
>
>
>
>
>
> [1] http://lists.foaf-project.org/pipermail/foaf-dev/2000-September/004222.html
> http://web.archive.org/web/20011123075822/http://rdfwebring.org/2000/09/rdfweblog/example.html
>
Received on Saturday, 19 January 2013 23:50:17 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:45 UTC