OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: [xml-dev] Symbol Grounding and Running Code: Is XML Really Extensib

[ Lists Home | Date Index | Thread Index ]

Dare Obasanjo wrote:

> Fine here's the concrete scenario that led to this discussion. Certain
> folks have suggested that the new XML syndication format being worked on
> by Sam Ruby and others (aka Pie/Echo/Atom/Whatever) should allow
> extension elements from other namespaces which is what both RSS 1.0 and
> RSS 2.0 currently allow. 
> The question then came up as to how one could describe the semantics of
> these extension elements and some (I think Danny Ayers specifically)
> claimed that RDF and DAML could solve this problem. 

RDF can solve problems for a very narrow meaning of semantics, 
namely Tarskian semantics. That's no more controversial than saying 
CSS can solve the problem of document rendering for very narrow 
meaning of rendering. A CSS engine is just a rules engine. You give 
it some rules (CSS) to apply against an instance document (markup) - 
the output are instructions to software to display a certain 
arrangement of pixels. An RDF engine is just a rules engine, you 
give it some rules (the model theory) to apply against an RDF 
instance document (a graph) - the output is another RDF graph. 
Regexen, XSLT, XQuery, Prolog, SQL, Unification. All pattern 
matching technologies are much alike - what sorts them is the range 
of patterns (or sentences) they can operate over.

RDF semantics by way, would not describe the semantics of an 
extension element, so your question is somewhat pointless. It's just 
like asking "how would a relational model.." without the existence 
of SQL+triggers or even databases. What you are looking for is a 
solver - "is it true that such and such", /not/ a model of data - 
"such is the case". What RDF semantics describes is what you may say 
to be true about about RDF graphs under certain operations. This is 
a common misunderstanding about model theoretic semantics - the 
semantics apply to the data model, not sentences written with the 
the data model in mind (you can say any old nonsense in RDF, but you 
cannot use RDF to induce nonsense that isn't already present). If 
that were the case, we'd call it a sentence theory not a model 
theory. What you can do when you write that sentence in RDF is mix 
that graph with *any* other RDF graph and use the RDF rules of 
inference (pitiful as they are) to form new conclusions (expressed 
as another RDF graph - RDF is closed under legal operations). Think 
of it as a few steps removed from processing Infosets from arbitrary 
sources, applying XPaths to anything that wil fit in its model, or 
applying an XQuery to across arbitrary datasources. If RDF is 
pointless so are they to some extent. However a lot of people think 
the idea of uniformly merging content from arbitary sources a Really 
Good Idea Whose Time Has Come. As far as I can tell, making Fecho 
RDF happy isn't really putting anyone's nose out, so what's the 
fuss? What happens from there is anyone's guess. Might be a dud, 
might be the best thing since slice bread, You won't know until you 
try or someone else does. Plus if it doesn't work out, it'd be a 
great way detractors to stick two fingers up at RDF and the semantic 
web goop, as you put it. Anyway look, if you're going to harp on, 
take some time to understand what you're harping on about. From your 
posts here, you obviously understand XQuery very well, which is much 
more complicated, so RDF won't be much trouble. I'd urge you you to 
at least form a qualified opinion on it instead of this sort of 'it 
sucks, nah nah' silliness.

> Since you seem like another RDF head I'm curious as to how I'd use RDF
> to let a news aggregator [even an RDF-aware one] know that the
> <dare:stylesheet /> child of the <entry /> element contains a text/css
> stylesheet as content which should be used to style the content of the
> <content /> element when displaying it in the embedded browser within
> the aggregator. 

You could... parse the RSS as an RDF graph and pass that through 
some Prolog rules that can read RDF graphs and act on them; there's 
an article on ora.com somewhere about processing RDF with Prolog. 
But this probably is a stupid scenario for RDF - the correct answer 
to such a question is why aren't you using a stylesheet language?

> These are both examples I brought up when this was mentioned on the Atom
> mailing list to which I didn't here any satisfactory answers as to how
> RDF solves this problem in the specific case of XML vocabularies for
> news syndication. 

That's no fun - define satisfactory :)

Bill de hÓra


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS