RE: [plcs-dex] Ontologies

["David Price" <david.price@eurostep.com>]

First, interesting reading. I've tried to break a few concerns into separate
sections that follow. Please take these queries/concerns as intended, an
attempt to understand the issue, what people really mean by these terms and
point out what seem to be a few potential problems.

1) I don't think I quite understand what problem the proposal is trying to
solve. The simple way I look at this is:

- the PLCS ARM has been converted into classes forming the upper ontology
for use in defining the PLCS Reference Data

- the language of choice, OWL, supports the definition of a set of
subclasses that are complete (i.e. everything must be one and only one of
the subclasses and classes can be abstract, disjoint, etc.)

- the language of choice, OWL, supports incomplete definitions of subclasses
as well

- OWL allows the same individual to be defined to be "of type" any number of
classes (I realize this isn't to be used for DEX exchange but will be of
interest in the future)

It seems to me that this satisfies all the requirements described in the
paper - both with respect to "classification" and "categorization". 

2) I am concerned with the statement in the paper that categories are
unrelated unless stated explicitly in the RD. That seems to imply that the
relationship between classifications need not be explicitly stated in the
RD. This seems to me to be a certain cause of problems in the long run, not
to mention bad ontology engineering practice. Perhaps I misunderstand?

3) The paper seems to talk about classifying categories. With the OWL
language this would cause the ontology to become OWL Full, rather than OWL
DL (for Description Logics). When choosing OWL, we made the decision to try
and stay with OWL DL because of the availability of free reasoners and
because of statements in the OWL spec such as "When using OWL Full as
compared to OWL DL, reasoning support is less predictable since complete OWL
Full implementations do not currently exist.".

One reason we added the Dublin Core annotations to the RDL was so that such
requirements can be met without affecting the ontology itself. Perhaps I'm
being a purist, but introducing "management" classes into the ontology
doesn't seem necessary at the moment.

4) Earlier I asked, for what purpose are these terms introduced into the
management of PLCS RD? I don't have the OED handy, but the Webster online
dictionary says

classification = systematic arrangement in groups or categories according to
established criteria

category = 1) any of several fundamental and distinct classes to which
entities or concepts belong and 2) a division within a system of
classification

So, there does not seem to be agreement that the two are indeed distinct.
Does the distinction made in the paper result from the terminology being
adopted by in the Defence Industry or elsewhere? If so, is there a
reference? Also, how does this relate to "codification"... is that just
"classification using shorthand"?

As I've not been intimately involved in the recent RD definition work, I may
have missed the core issue here so any help in understanding the problem
before agreeing with the solution is appreciated.

Cheers,
David

> -----Original Message-----
> From: Barker, Sean (UK) [mailto:sean.barker@baesystems.com]
> Sent: 02 July 2004 15:17
> To: Plcs-Dex Sub Group Oasis (E-mail) (E-mail)
> Subject: [plcs-dex] Ontologies
> 
> 
> Attached is a rough sketch on how I think we should use ontologies in
> PLCS. Its a rough draft, working out what should be done. I'm currently in
> the process of testing the ideas by adding in some of the task reference
> data (just uploaded). Once done, I'll revise in the light of experience.
> 
> I'll bring this up during the reference data workshops at the SC4 meeting
> next week. (BTW - I'll not be there before 10.00 Monday, due to urgent
> need to fix daughter's train tracks).
> 
> Sean Barker
> ATC Filton
> 0117 302 8184
> 
> 
> 
> ********************************************************************
> This email and any attachments are confidential to the intended
> recipient and may also be privileged. If you are not the intended
> recipient please delete it from your system and notify the sender.
> You should not copy it or use it for any purpose nor disclose or
> distribute its contents to any other person.
> ********************************************************************