Re: [xml-dev] The triples datamodel -- was Re: [xml-dev] Semantic Web permathread, iteration n+1

[Michael Champion <mc@xegesis.org>]

On Jun 8, 2004, at 4:46 AM, Alaric B Snell wrote:

> This 'flexibility' isn't something new to XML, it's inherent in any 
> format that uses some kind of tagged values; including things like 
> TIFF and PNG image files. And MIME headers, and SMTP email messages. 
> Nothing special about XML in this respect! XML fans seem to have 
> similar marketing ideas to Microsoft, picking up a good idea from 
> elsewhere and claiming to have invented it ;-)

Fair enough.   I agree that it is idea of labeled trees that supplies 
most of the real power in XML, and acknowledge the point that other 
ways of exchanging labeled data work for the use case I'm talking 
about.  XML's biggest real advantage is the network effect -- it's 
supported on every commercially viable language and platform.

> I think agreed schemas will *increase* reliability of systems.

No disputing that!  The problem, as I think Elliotte Harold pointed 
out, is that it is extremely difficult to get humans to agree on much 
of anything when the costs of the agreement are short term and tangible 
and the benefits are long term and hypothetical.

> The objection to this seems to be "Oh, so your system dies the second 
> it sees somebody's private extension?" - which is in no way implied by 
> schemas being agreed between the communicating parties. And for very 
> large scale systems, that 'agreement' can be as simple as saying "This 
> site publishes data in the format documented _here_"

That's not agreement it's assertion.   It works when everyone agrees on 
the semantics implied by the labels.  RSS is a fine example of this 
working in practice -- producers pick which variation they want to 
publish, and consumers sort out the differences among them, handling 
all sorts of ad hoc intermediate and experimental forms along the way 
because at the end of the day everything is being stuffed into the 
well-known semantics of a newsfeed.  Things get interesting in RSS-land 
when the same label (namespaced or not) does not have the same 
semantics, because of different date formats, different assumptions 
about well-formedness or embedded markup within the label, and so on.

This leads to the foundation hope of the Semantic Web -- we don't have 
to agree, I'll just publish what I assert about the syntax and 
semantics of the data are, and you can follow the chains of assertion 
back to something you recognize, and use third-party metadata 
assertions about my own attributes (such as competence and integrity) 
of my assertions...and sometime before the heat death of the universe 
you can infer what to do with the data I publish :-)  As I said at the 
beginning of all this, I hypothesize that this will actually work 
within organizations where the meaning of labels can be quickly mapped 
back to something concrete, and it won't work on the open Web where 
"meaning" fades off into greater and greater abstraction and 
contention.