|
Dear
Bill, Ken, and Peter,
I
don't think this is an issue that is easily banished.
We
struggled with this issue in developing ISO 15926 and came up with two
options:
1.
Have 2 identifiers for each "record". The first for what the record represents,
the second for the record itself. You use the identifier for the object of
interest.
2.
Make it part of the semantics of the relation whether the relationship was
between the data objects or what the data objects
represented.
I
thought that 1. was cleaner, but as it turned out it was not an option in the
EXPRESS environment. This also seems to be the case for Bill's social security
system, so the semantics should make it clear that it is the person the social
security number refers to was born on the day the date of birth field refers to.
Regards
Matthew West
Reference Data Architecture and
Standards Manager
Shell International Petroleum Company Limited
Registered
in England and Wales
Registered number: 621148
Registered office: Shell
Centre, London SE1 7NA, United Kingdom
Tel: +44 20 7934 4490 Mobile: +44
7796 336538
Email: matthew.west@xxxxxxxxx
http://www.shell.com
http://www.matthew-west.org.uk/
Ken,
I don't much like the sound of this. I thought ontologies were
supposed to be explicit (albeit partial) accounts of the meaning of
terms. Traditional DBs have lots of "features" like social security
numbers having birthdates -- which was locally ok since typically they were
built for a limited application purpose and everyone using the system drank
from the same pot of kool-aid. Ontologies are supposed to put an end to
such things, not perpetuate them...
.bill
On Apr 12, 2007, at 16:29 , Ken Laskey wrote:
Peter,
I often questioned when talking about a URI that dereferenced to a Web
page whether I was talking about
- the Web page (say, a collection of information on King Arthur),
- the subject of the Web page (i.e. King Arthur), or
- some particular piece of information (say, King Arthur had a sword
named Excalibur)
A recent answer I got that made sense was that the domain and range of
the properties used to describe the URI (i.e. the arcs in an RDF graph)
provide the context to disambiguate the meaning. While not considering
that in detail, I think it works.
Does this start to address your problem?
Ken
On Apr 12, 2007, at 3:44 PM, Peter F Brown wrote:
Ed:
Thanks for a fascinating overview. My comment on the
webmeeting chat line was slightly facetious when I stated that: “If OWL
added identifiers as you suggest, that would break Tim Berner-Lee's
underlying model for Web architecture”, but I think that you hit the heart
of one of the problems that I have with OWL and with the W3C axioms about
information modelling: that is the failure to distinguish
between:
-
a URL as an identifier of something, (in the terms you refer
to in slide 34 [1]); and
-
a URL that is the “something” (a resource)
and the fact that you can’t actually make any assertion
about a URL being considered as an identifier.
Although not designed as a standard for inferences and
reasoning, the ISO Topic Maps standard does have an explicit mechanism for
capturing and managing identity (using the concept of published subject
identifiers), in the ways you describe are needed. I’m not sure what other
specs out there do likewise, except for the humble URN…
In either case, we still have the problem of what you
called “relative uniqueness” rather than, presumably, absolute
uniqueness.
I look forward to seeing you at the Ontology Summit
next week!
regards,
Peter
[1] http://ontolog.cim3.net/file/work/DatabaseAndOntology/2007-04-12_EdwardBarkmeyer/InfoModels-Ontologies--EdBarkmeyer_20070412.ppt
-------------
Peter F Brown
Founder, Pensive.eu
Chair, CEN eGovernment Focus Group
Co-Editor, OASIS SOA Reference
Model
Lecturer at XML Summer
School
---
Personal:
+43 676 610 0250
http://public.xdi.org/=Peter.Brown
www.XMLbyStealth.net
www.xmlsummerschool.com
_________________________________________________________________
------------------------------------------------------------------------------------------
Ken Laskey
MITRE Corporation, M/S H305
phone:
703-983-7934
7515 Colshire Drive
fax:
703-983-1379
McLean VA 22102-7508
_________________________________________________________________
|