ISSUE: owl:quote - 5.4 and 4.3

Structured types and owl:quote:  Arguments in favor of owl:quote.

In Peter's abstract syntax/translation document he noted:

> The document does not have true support for structured data, as
> opposed to the structuring capabilities in individuals, assuming that
> issue 4.3 will be resolved against including structured data in OWL.

> This document does not have a quoting mechanism, assuming that issue
> 5.4 will be resolved against having such a feature in OWL.

I should have proposed to open 5.4 previously.  I think of it as the
ultimate simplification of Jonathan's much more extensive proposal
under 4.3.

> The document assumes that literals can be either typed or untyped. 

If literals can be untyped, then it would be relatively simple to
include arbitrary chunks of XML as literals if we had an OWL:QUOTE
mechanism.  This would impose minimal overhead on the model theory.
(I say this knowing that I won't be the person doing the work.  QED,
minimal overhead.)

Such literals could be translated to RDF triples in a variety of ways,
either as a string representing the printed XML or as a tree using
<list>, <first>, and <rest>.  (Not sure about the semantic impact of
the latter.  We could prove in RDF that the resource "Bill" was the
third element of some list, which we could not do in OWL if quoted
literals where treated as blobs (unstructured values).)

BENEFITS:

A. This provides the ability to make some limited assertions about XML
objects.  The world contains vastly more transactional XML objects
than static Web objects (or soon will). The ability to state
assertions about these literals in our ontologies would increase the
likelihood of this technology spreading into commercial domains.

For example, the banking industry already has an XML standard for all
debit/credit related personal banking operations.  They are working on
an XML standard to capture ATM transactions.  Credit card transactions
are only a matter of time. (This last example is one place where there
are serious problems with transaction volumes and requirements on
processing time that mitigate against the use of XML at the moment.
But we will get there eventually.)

We may not think of these schemas as ontologies but they really are.  
There is strict operational agreement on the meaning of instances 
of these schemas (under appropriate security guarantees).  To the 
extent that money moves around based on them.

Since we would not be able to have quoted objects as subjects (I am
assuming they are treated as any other literal, and thus can only be
the object of a DataProperty), the statements we could make would be
somewhat limited, but at least there would be some way to begin to
address this huge transactional world.

<businessPartner owl:id="Acme Novelties">
 <submittedPurchaseOrder>
  <owl:quote><ebxml:po>...</ebxml:po></owl:quote>
 </submittedPurchaseOrder>
</businessPartner>

The other option is to create a purchase order ontology and some special
purpose stuff to tie it to the ebXML standard.  Which imposes a higher
barrier to entry and is adhoc.

B. Reification:

<person owl:id="Bob">
 <say>
  <owl:quote>
   <person owl:id="Mary">
    <love><person owl:id="Bob"/></love>
   </person>
  </owl:quote>
 </say>
</person>

This would have no more meaning in OWL than

<person owl:id="Bob">
 <say>
  <owl:quote>BLAH</owl:quote>
 </say>
</person>

But it would be potentially useful in systems that extend OWL.

- Mike


Michael K. Smith, Ph.D., P.E.
EDS - Austin Innovation Centre
98 San Jacinto, #500
Austin, TX  78701

* phone: +01-512-404-6683
* mailto:michael.smith@eds.com
www.eds.com

Received on Wednesday, 17 July 2002 21:15:27 UTC