Re: The Steven King Example

Funny thing about intuition: yours is different than mine.

I am not interested in defending this example if it does not ring true
for someone with more RDF and Semantic Web experience than I possess.
On the other hand, I have top admit to being a bit puzzled.

I came into this with the hope of becoming better versed in RDF and the
Semantic Web. Back in March I explained to TimBL, Ralph, Eric Miller,
and others that I did not GET the Semantic Web, or at least I did not get
the claims made by Semantic Web proponents that using RDF was in some
way superior to using structured markup.

So it was, a few weeks ago, that I worked with Dan to develop a simple example
that could be understood by mere mortals like me and yet still demonstrate 
the power
and simplicity of RDF. I wrote straightforward text to describe a scenario 
that is simple
enough for me to understand, and I solicited the help of Norm Walsh to 
produce the
example inputs. Dan wrote a GRDDL result that is much more complex than I 
had hoped,
but he said that it was an accurate RDF representation of the triple:
         "Stephen King" "is author of" "The Stand".

I am totally willing to accept the proposition that Dan and I got it wrong. 
I would love
to see the example that gets it right, and I only hope that I will be able 
to follow it.

However, I think that there has got to be something fundamentally wrong 
with this
Semantic Web thing if a technical writer with my experience working with DanC
and Norm Walsh couldn't put together a simple example and have it be 
interpreted
identically by all.

This exercise has reinforced my intuition that the Semantic Web is even 
more prone
to misinterpretation than common English prose embedded within XHTML. My 
intuition
tells me that
         http://www.stephenking.com/pages/works/stand/
is a web page which conveys the fact that "Stephen King" "is author of" 
"The Stand"
But, apparently, that is only because I don't know any better. My 
confidence has been
successfully undermined.

Regards,

Murray



At 10:22 AM 11/24/2006 +0000, you wrote:

>At the last meeting I accepted an action to have a closer look at the
>Stephen King example in the spec
>
>   http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action05
>
>The example expresses in RDF, the fact that the resource identified by
>
>   http://www.stephenking.com/pages/works/stand/
>
>Has a title "The Stand" and has a foaf:maker (an agent that made this
>thing) who is the person that is the foaf:isPrimaryTopicOf the resource
>identified by
>
>   http://en.wikipedia.org/wiki/Stephen_King/
>
>Or more simply that
>
>   The resource identified by
>http://www.stephenking.com/pages/works/stand/ has a title "The Stand"
>and was made by Steven King.
>
>When I point my browser at this url I get back an html page describing a
>novel entitled "The Stand" written by Steven King.  This web page has
>considerably fewer words than a novel.  It has a title element in the
>header whose content is "StephenKing.com". It also has "The Stand
>(1978)" as text on the page formatted in a way that might cause it to be
>interpretted as a title.  As far as I can tell there is no redirect when
>I do the GET on this URL.
>
>I think there are murky waters here, related to a tag issue
>(http-range-14) and complex modelling issues concerning the distinction
>between abstract works, expressions of those works and manifestations of
>those expressions.  Now it may be that the strictly speaking the example
>in the spec can be successfully defended as being correct, however I
>think it is confusing because folks reading it might think that
>
>   http://www.stephenking.com/pages/works/stand/
>
>Identifies a resource that is a description of a novel, that the title
>of this description is not "The Stand" and that this description was not
>written by Stephen King and therefore that the RDF in the example is
>false, undermining the confidence the reader in the specification.
>
>I suggest that in this example in the spec it would better to avoid such
>potential confusion and to draft a different example that will be more
>intuitively correct to the reader.
>
>There are several ways the current example could be modified to be
>correct.  One would be to use a URL that clearly does identify the
>novel; another would be to introduce another node into the graph so that
>both the novel and its description are represented there.
>
>A question for the WG is whether it is worth the effort of drafting such
>an alternative?
>
>Brian
>
>
>
>

Received on Friday, 24 November 2006 16:21:09 UTC