On Jan 21, 2011, at 10:57 PM, doug foxvog wrote: (01)
> On Fri, January 21, 2011 13:39, Ed Barkmeyer wrote:
>> Christopher Menzel wrote:
>>> On Jan 21, 2011, at 9:46 AM, doug foxvog wrote:
>>>> ...
>>>> A standard distinction between a set and a class, is that membership
>>>> in a [set] cannot change, while membership in a class can.
>>>>
>>>
>>> I think it's useful to distinguish two claims when it comes to the
>>> identity conditions of classes:
>>>
>>> (1) Classes are not extensional (i.e., distinct classes can have the
>>> same members/instances)
>>>
>>> (2) Classes can change their membership.
>>>
>>> In the formal semantics of a number of KR languages, (1) is true but,
>>> strictly speaking at least, (2) is not. Notably, classes in OWL are
>>> explicitly non-extensional: since a class is stipulated only to *have*
>>> an extension in OWL's formal semantics, nothing prevents distinct
>>> classes from having the same extension. The same is true of RDF.
>>> However, simply because there is no formal notion of change built into
>>> OWL's semantics, there is no possibility, within a given interpretation,
>>> that a class change its membership. As noted in an earlier message in
>>> this thread, without augmenting the notion of an OWL interpretation
>>> somehow, change can only be represented formally in terms of something
>>> like a series of interpretations that are thought of as temporally
>>> ordered. That said, (2) does seem to be a strong *intuitive* idea in
>>> the KR, AI, and database communities.
>
>> The particular problem I have recently got involved in is the intrusion
>> of temporal concepts into would-be ontologies in business applications.
>> In the supply-chain area, for example, it is important to be able to
>> talk about schedules and shipments being "late". Getting past the
>> indexical issues, which are fixed by translating the intuitive "now"
>> into specific time relationships, the particular problem is that
>> shipments and orders do change state, and actions are taken on the basis
>> of reclassification.
>
> Fine.
>
>> A major problem for us is that the industry folk throw these concepts
>> into what was an ontology for the "snapshot" model of decision-making --
>> the state of the world at the time the decision is to be made. This
>> gives rise to formalizing ideas like "proposition X is false at time A
>> and true at time B." And that problem arises from the idea that states
>> of things are characterized by propositions, which seems to be
>> fundamental to applications of ontologies.
>
> This is fine for representing states.
>
>> The 4D idea that a thing in
>> a different state is a different thing,
>
> What is the need to consider a thing in a different state a different
> thing? (02)
It has many advantages over time-state-based ontological frameworks, though it
also (of course) has its problems. The chief advantage can be summed up as its
ability to talk about things at different times in the same sentence. for
example, it is much easier to define notions such as rate of change if one is
able to talk about temporal 'slices' explicitly. (03)
The issue is not what counts as a 'thing', but whether one is allowed to talk
of temporal slices - things-in-a-state - as entities in the ontology at all.
(Could there be a class of them, for example?) Opponents of the 4D approach
typically reject such entities as incoherent, philosophically confused, etc.. (04)
>
> This is not what the ontological community considers to be a 4D idea. (05)
It is exactly that. (06)
> A 4D object, in this context, is an object that can have different
> state at different times. (07)
No, that is a continuant. In a very strict 4D model, there is no single 'thing'
that has the various states. (Although I hasten to add, it is very convenient
to be somewhat less strict than this. I mean only to make the conceptual
distinction.) (08)
>
>> and 'objects' are actually
>> sequences (or more generally, lattices) of things in states,
>
> This is a different model, that has a far different definition of "thing"
> than is generally used in computer ontologies.
>
>> is a means of producing a formal semantics,
>
> One could certainly produce a formal semantics using such definitions.
>
>> but it is totally out of line with the
>> intuition of the domain experts.
>
> Then it is probably not worth while to present such a model to them.
> If you want to use it "under the covers"/"inside the black box",
> because it makes calculations easier, fine. But don't inflict such
> a model on domain experts. I would suggest it also be hidden from
> the ontology builders and merely be maintained as part of the inference
> engine. (09)
But the engine will be using the concepts provided by the ontology itself. (010)
>
>> They cannot then "validate" the
>> ontology -- they don't understand it.
>
>> I have said in that forum that solving the problem is beyond my
>> expertise. It is my conviction that the problem is not really "time",
>> but rather "change of state" or "alternative states", and in that sense,
>> "time" is a means of labeling "alternative possible worlds".
>
> Time is certainly one way of marking alternative states.
>
>> All we are saying is that the intuitive notion of change is endemic to a
>> lot of ontology applications.
>
> Certainly.
>
>> We can usually constrain the immediate application
>> to avoid the problem or create a convenient work-around,
>
> If you are only using the data with a single reference time period,
> the of work-around need not consider viewing the data in another
> temporal context. But if the data is to be dealt with for another
> time period, that should affect the work-around method chosen. (011)
The very use of terms like 'temporal context' suggests that you have already
chosen one way to handle the ontological description of time and change. Which
of course is fine, but you should acknowledge that others prefer a different
ontology, and that yours is by no means the only one possible. (012)
Pat Hayes (013)
> _________________________________________________________________
> 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
>
> (014)
------------------------------------------------------------
IHMC (850)434 8903 or (650)494 3973
40 South Alcaniz St. (850)202 4416 office
Pensacola (850)202 4440 fax
FL 32502 (850)291 0667 mobile
phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes (015)
_________________________________________________________________
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 (016)
|