Lists Home |
Date Index |
Idea maps for the Semantic Web
<stuff> http://www.isacat.net </stuff>
Semantic Web Log :
>From: Simon St.Laurent [mailto:email@example.com]
>Sent: 17 November 2002 18:10
>Subject: Re: [xml-dev] RDF for unstructured databases, RDF for axiomatic
>firstname.lastname@example.org (Shelley Powers) writes:
>>I think, in many ways, that's what many of the objectors have against
>>RDF/XML -- you have to have some pre-knowledge of RDF in order to
>>read it, write it, work with it, and truly understand it.
>I think there are two serious questions any time people move from a
>"just XML" perspective to the "new, improved RDF" perspective.
>The first is learning the RDF model. While the core of that model is
>pretty simple (I think Jonathan Borden has said it takes only one
>slide), figuring out how RDF really works is more complicated. Anyone
>who has doubts about the intrinsic crunchy goodness of URIs is liable to
>have an aneurysm during any serious encounter with RDF. The Concept and
>Abstract Syntax document is pretty good for what it does, but the RDF
>model still requires a rather different level of modeling than does XML
>document markup - a level I see as unnecessarily complicating in the
>vast majority of cases where markup is being used.
>(I vastly prefer patterns embedded directly in documents to patterns you
>have to assemble in your head outside of the document by linking chains
>of abstract identifiers.)
Agreed it would be preferable to embed the patterns directly, but
unfortunately I believe that is a topological impossibility in linear
strings of characters. I'm not aware of any explicitly 2+ dimensional
formats in current use. I also agree that this is a complication is
unnecessary in a considerable proportion of cases, hence my headless triples
>The second issue is what I see as a more or less permanent mismatch
>between RDF's graph model and XML's hierarchical model, which produces
>some serious syntactic complications. Jonathan Borden has done very
>well recently with his revised RDDL/RDF, but I think he's mostly
>achieved it by hiding the RDF impact on the XML syntax to the maximum
>extent possible. (It's very reassuring to see that this is possible,
>>So then the question becomes: Is the issue really about the existing
>>RDF/XML? Or is it about the complexity of the RDF model? I think we
>>need to be very sure about this before we run off into alternative
I do rather think the complexity of the model is being overstated somewhat.
Ok, there are aspects that go beyond the usual XML data languages, but not
many, and those that do are way simpler than a lot of the modelling people
use within programming languages every day. Most if not all the constructs
have pretty close parallels in OO and/or functional languages (which is
another source of confusion, but that's another story). I think the problem
may have more to do with complacency instilled by regular exposure to XML
than with the RDF model. I also think that any dumbing-down of the RDF model
would almost certainly be a bad move, baby & bathwatersville. Better hiding
of the features of the baby that worry the neighbours would strike me as a
>For me, it's definitely about both. I can read the RDF model, but have
>no interest at all in using it to model anything more complicated than
>about a FOAF file. For the modeling I need to do, the XML BB gun is
>much more appropriate than the RDF Gatling Gun, and far less likely to
>cause collateral damage.
Does it really get much more complicated than a FOAF file? Ok, I don't think
I've seen reification used much around there, but such techniques can exist
without having to be used, without impacting the FOAF text file at all. Any
increase in computational complexity comes from being able to connect up
nodes easily, not through any particular strangeness in the language.
The BB gun may be all right for mice, but you never know when you might need
to deal with bigger vermin. You can keep adding ammunition (in the form of
new elements) but this will only scale so far. The difference in cost is
likely to be relatively small in most systems, most of those where plain XML
is an obvious candidate and a trivial solution isn't possible. But for a
growing number of applications, particularly any that need to be able to
communicate outside of the local network then IMHO the RDF solution is
likely to be cheaper.
Collateral damage shouldn't be an issue as long as the basic contracts are
kept to - URIs treated with respect, vocabularies used only when the
definition of the terms is a good match for the intended meaning. Which is
just regular good practice.
>The existing RDF/XML looks ridiculous to me, and I find its odd bouncing
>off qualified/unqualified attributes to be a warning sign for the XML
>namespaces specs, but I think a lot of that has to do with basic
>incompatibilities between the models.
There, alas, yoou've probably got a good point.