I agree with Ian on this. I think it is more useful to have a simple
clear definition of ontology, and separately from that a definition of
the interesting and cool things that can be done with an ontology, and
stuff about the kinds of formats and languages than an ontology could be
in. I see no benefit in overloading the term "ontology" with all that
stuff. (01)
I thought the term "ontology" was reasonably well defined, so it's been
interesting reading the ideas here. An ontology is the view of the world
as seen by a given application. This specialises the original term
"ontology" in the singular, which was a view of the world as we see it
as humans. Ontologies (plural) are the different views of the world from
the perspective of different computer systems. (02)
In this sense, every system has an ontology, whether anyone has
documented it or not. It might be in the brain of the developer who has
now left to do something more interesting. Or it might be in some ad hoc
spreadsheet that was thrown "over the wall" at the start of the
development project, and as such that spreadsheet may well no longer
represent the complete and up to date ontology because a lot of stuff
had to be fixed during integration and delivery of the system, by on
site engineers, because the original semantics were not all correctly
understood. (03)
An ontology which formalises this business view and can be maintained as
a project deliverable, is a better kind of ontology than the above,
default options. This is before you even think about applying automated
reasoning. It puts the business in control of its system developments. (04)
Then, an ontology which is in a form that supports automated reasoning,
adds another level of benefit - if this kind of functionality is needed
in a given business application. Lots of good reasons for doing it, but
not doing does not mean you don't have an ontology. (05)
All IMHO of course. (06)
Mike (07)
Ian Bailey wrote:
> Hi Chris,
>
> Sorry, yes typo on non-well-founded. By which I mean (at least from my
> limited understanding) that there is no requirement to build everything up
> from the empty set. All the IDEAS Group work has been based on this
> principle, and the ontologies seem to be perfectly implementable to me. I
> doubt they'd support reasoning though.
>
> There's a link here
> (http://en.wikipedia.org/wiki/Non-well-founded_set_theory) which probably
> explains it in terms of logic. Personally, I can't understand a bloody word
> of it.
>
> I'm not sure the AI level-nine wizards entirely co-opted the term either.
> All the ISO15926 and BORO work was done without requirement for machine
> reasoning. Philosophers also talk about ontology quite a bit too.
>
> For me, it's much more important for an ontology to closely mirror the real
> world, because what I care about is building better information systems.
> Sure, there's got to be some formality in their development, but that
> doesn't mean it has to support reasoning. I get particularly worried when AI
> folks tell me I can't do things like have types whose instances are types,
> and relationships to relationships. They're bending their "ontology" to suit
> the tool at hand. If all you have is a war-hammer, everything starts to look
> like an orc.
>
> Cheers
> --
> Ian
>
> -----Original Message-----
> From: Christopher Menzel [mailto:cmenzel@xxxxxxxx]
> Sent: 26 October 2009 17:30
> To: ian@xxxxxxxxxxxxxxxx; [ontolog-forum]
> Subject: Re: [ontolog-forum] Just What Is an Ontology, Anyway?
>
> On Oct 26, 2009, at 11:43 AM, Ian Bailey wrote:
>
>> Er...what does ontology have to do with automated reasoning ?
>>
>
> Ever since the term was co-opted (not entirely without warrant) by the
> CS/AI community, a (perhaps the) central motivation has been to
> facilitate automated reasoning on large knowledge bases.
>
>
>> The scope of ontology is far wider than that, and there are lots of
>> ontologies out there that are really useful for real world
>> applications, but don't meet the narrow requirements for finite-time
>> reasoning.
>>
>
> Example being...? Do you really mean it's in a logic without a proof
> theory? Or do you simply mean that the ontology is not formally
> specified? I don't doubt that a semi-formal ontology couldn't be
> useful for, e.g., facilitating a common understanding of a domain
> among human agents. But, ultimately, complete clarity (and
> computational support) comes only when an informal ontology has been
> rendered in a logical language. And if you've got a genuine logical
> language, you'll have some sort of proof theory and hence something
> amenable to automated reasoning.
>
>
>> On the other hand, there are ontologies out there that have been
>> built only for reasoning, and are no use whatsoever in real world
>> applications...in fact there are rather a lot of these, mostly
>> funded by our taxes, unfortunately.
>>
>
> So there are bad, well-funded ontologies; nothing new there.
>
>
>> I'm not sure a complete proof theory is required either.
>>
>
> You are right; partial proof theories for well-specified fragments of
> a given logic could also be useful. The point was that one needs a
> rigorous proof theory for a logic to support any kind of automated
> reasoning.
>
>
>> The none-well-founded stuff seems to work quite well (assuming
>> that's what Chris meant by "proof").
>>
>
> I don't have any clear idea what you have in mind by "none-well-
> founded" stuff. I'm guessing you mean "non-well-founded" but I'm
> still not sure what you mean. Perhaps you are alluding to well-
> founded semantics (WFS)? That is indeed a framework that in general
> does not have a complete proof theory but there are a number of
> interesting completeness results for WFS-based systems when certain
> conditions are imposed on models.
>
> -chris
>
>
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>
>
>
> (08)
--
Mike Bennett
Director
Hypercube Ltd.
89 Worship Street
London EC2A 2BF
Tel: +44 (0) 20 7917 9522
Mob: +44 (0) 7721 420 730
www.hypercube.co.uk
Registered in England and Wales No. 2461068 (09)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (010)
|