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


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Object Role Modelling (ORM) or UML or ?? for designing Schema s

Apply concept modeling when the thing to be returned 
MUST be the thing asked for and must in every detail 
behave as expected/required without regard to the 
implementation.  The trick being played here is 
that some want to push the definition of "communication" 
down to the network layer of routing and node naming 
proving the precise point of the thread:  interpretive 
means applied to common terms for each local node 
produces conflicts in results.  To get reliable 
results without intensive trust verification, the 
simplify the means of verification and use a priori 
definitions of the results.  In all cases, no?  
But even for the pizza, it isn't predictable outside 
a known context.  Try ordering a customized 500 
passenger airliner.  Context rules.

That is why UML sits above the implementation language. 
Can you do this kind of modeling in XML?  Sure.  
XMLIsAMetaAsYouNeed.  That is why I posted the 
ESML example.   They went all the way back to 
the syntax notation. 

Should you do that?  Hard to say.  The ontologists 
suggest that once you get into heavy ontology 
modeling, a common markup language is awfully handy 
particularly when merging models.  That is why 
PDES had Express.


Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h

-----Original Message-----
From: Danny Ayers [mailto:danny@panlanka.net]
Sent: Wednesday, February 14, 2001 11:34 PM
Subject: RE: Object Role Modelling (ORM) or UML or ?? for designing
Schema s

<- > If I order pizza, I expect pizza.
<- Of course if you order pizza in e.g. New Haven CT, it may not
<- automatically
<- come with red sauce.
<- A major benefit of using a URI to refer to something is that you can
<- dereference it to get a definition.

I reckon that's the key - you don't have to explicitly setup concept
mappings if you don't want to, if there's a reference at the end of the
chain somewhere then you've got something you can make use of mechanically.

The infrastructure doesn't need conceptual models, they are merely an option
at the nodes.

How the information is handled will be node-specific : completely
pass-through, using predicate logic, fuzzy mappings or whatever - there's no
reason not to do conceptual mapping if you want to, to a large extent these
processes would be independent of the communication. The only potential
problem comes when your looking for a pizza and call on NLU mapping x which
in turn calls on ANN mapping y before the order is interpreted as an order
and passed to dog food outlet z. I'm afraid the old trust issue comes into
play. Having said that, it's not that much better in human terms - order a
pizza in Sri Lanka and all you can be sure of is bread with a red-hot chili
on top.


To unsubscribe from this elist send a message with the single word
"unsubscribe" in the body to: xml-dev-request@lists.xml.org