W3C home > Mailing lists > Public > public-prov-wg@w3.org > January 2012

Re: complementOf -> viewOf: proposed text

From: Graham Klyne <GK@ninebynine.org>
Date: Thu, 19 Jan 2012 15:12:04 +0000
Message-ID: <4F1832C4.6090005@ninebynine.org>
To: public-prov-wg@w3.org
Luc,

(This is a side conversation: I don't think it affects our point of agreement.)

I think our discussions have established, both intuitively and with some degree 
of formality, that alternativeOf is transitive if each value related by 
alternativeOf is ultimately a specialization of some unique "real-world" thing 
(that is not itself a specializations of anything).  If an entity can specialize 
multiple things which are not themselves specializations, then transitivity 
cannot be demonstrated (*).

So I think it's not so much about not understanding our data model, but having a 
data model that does not try to over-constrain its domain of discourse.  Note 
that, ultimately, no language grounded in first order logic that can describe an 
infinite domain can never uniquely constrain its domain of discourse (i.e. its 
models) - cf. http://en.wikipedia.org/wiki/Löwenheim–Skolem_theorem

(*) What I think we may not know is if there's any model in which alternatives 
don't all ultimately specialize the same real-world thing, but for which 
alternativeOf is transitive.

#g
--

On 19/01/2012 12:54, Luc Moreau wrote:
> I am personally not worried about not defining the property as transitive.
> For specific applications, one can always define it transitive.
>
> I am worried about us not understanding our model enough, which prevents us
> from deciding whether its transitive or not.
>
> Luc
>
> On 19/01/2012 10:43, Graham Klyne wrote:
>> Works for me :)
>>
>> I might even go further: drop the transitivity property until and unless a
>> specific requirement comes up.
>>
>> #g
>> --
>>
>> On 19/01/2012 09:52, Luc Moreau wrote:
>>> Hi Graham, Paolo, all
>>>
>>> Given this, and to allow us to progress on the document, can we, for now, remove
>>> the transitivity property, and add a note in the document, stating that the
>>> transitivity property is still
>>> under investigation?
>>>
>>> Cheers,
>>> Luc
>>>
>>> On 01/19/2012 09:27 AM, Graham Klyne wrote:
>>>> Hmmm... this is starting to feel to me like a philosophical rathole.
>>>>
>>>> I think we may be muddling things and roles, as maybe illustrated by your:
>>>>
>>>> So similarly I would not like to conclude alternateOf(Bush, Obama)
>>>>
>>>> This feels like a replay of the old Fregian "Hesperus and Phosporus" sense and
>>>> reference discussion.
>>>>
>>>> All this complexity is leading me to a view that while transitivity of
>>>> alternativeOf may be appealing at some levels of intuition, it may carry too
>>>> many traps and, absent a compelling requirement, we'd be better to leave it.
>>>>
>>>> Which I think is what Paolo is suggesting.
>>>>
>>>> #g
>>>> --
>>>>
>>>> On 18/01/2012 08:55, Stian Soiland-Reyes wrote:
>>>>> On Tue, Jan 17, 2012 at 18:01, Graham Klyne<GK@ninebynine.org> wrote:
>>>>>>> alternateOf(paoloInCafe, customerOnRedChair)
>>>>>>> alternateOf(stianInCafe, customerOnRedChair)
>>>>>> Hmmm... I'm not sure these actually match my intuition about alternateOf;
>>>>>> i.e. that they're both versions of some real-world thing. What real-worlkd
>>>>>> thing would that be?
>>>>>
>>>>> It would be something like the atoms of the living person who sits
>>>>> within the confines of the red chair. Perhaps it is more a case of
>>>>> specialization than alternateOf in this case. (and so a strong case
>>>>> for why specializationOf is not a subproperty of alternateOf)
>>>>>
>>>>> But this thing with the atoms is not true. A customer is not a set of
>>>>> atoms. A cafe *customer* is a concept which depends on the
>>>>> interactions with the cafe. While Paolo was in the cafe, he sat in the
>>>>> red chair and ordered coffee - and so for a period (the full lifetime
>>>>> of paoloInCafe) he also became customerOnRedChair.
>>>>>
>>>>>
>>>>> This would probably be fine then:
>>>>>
>>>>> specializationOf(paoloInCafe, customerOnRedChair)
>>>>> specializationOf(paoloInCafe, paolo)
>>>>>
>>>>> -->
>>>>> alternateOf(paolo, customerOnRedChair)
>>>>>
>>>>> which makes sense - they are both talking about the same thing.
>>>>>
>>>>>
>>>>>
>>>>> but if we also have the equivalent assertions about Stian - but the
>>>>> old characterisation interval of paoloInCafe never overlaps that of
>>>>> stianInCafe - then I feel they should *not* be alternateOf each other,
>>>>> because they did not exist at the same time.
>>>>>
>>>>> So similarly I would not like to conclude alternateOf(Bush, Obama)
>>>>>
>>>>> .. because if we do, then as far as I can tell there is not much value
>>>>> in alternateOf() any more.
>>>>>
>>>>>
>>>>>
>>>>> And that is perhaps my point. We can't have a single hierarchical
>>>>> structure organizing everything that exists (and talk about "the same
>>>>> real world thing"), because we include in "exists" various abstract
>>>>> concepts and simplifications that are not easily mappable to our
>>>>> understanding of the physical world.
>>>>>
>>>>> I am sure we can agree that this email message can be characterised by
>>>>> an entity. However you can't easily map that entity to electrons on
>>>>> the wire or photons coming out of the screen - although we are of
>>>>> course aware that the message would not exists without those.
>>>>>
>>>>>
>>>
>>
>
Received on Thursday, 19 January 2012 17:56:05 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:58:11 UTC