W3C home > Mailing lists > Public > semantic-web@w3.org > February 2005

Re: Madonna

From: Henry Story <henry.story@bblfish.net>
Date: Wed, 9 Feb 2005 17:46:30 +0100
Message-Id: <47af85968d4a8a4667b647d18617f79f@bblfish.net>
Cc: semantic-web@w3.org
To: Jeremy Wong <50263336@student.cityu.edu.hk>

Ok. Back on subject here. Thanks Jeremy.

I developed the Madonna example to illustrate the parallel between
personal identity and Atom Entries. Here is a section of a
conversation I had recently.

---------------------------------------------------------------
The RDF Semantics document says:

[[
  It assumes, implicitly, that URI references have the  same meaning
whenever they occur. To provide an adequate  semantics which would be
sensitive to temporal changes is a  research problem which is beyond
the scope of this document.
]]

That is really no problem. It is exactly what first order logic does.
And people like David Lewis, and many other great logicians have shown
that this is entirely sufficient not only for basing temporal reasoning
on, but the often even less well understood counterfactual reasoning. 
[1].

My guess is that the temporal reasoning problems people discover
are not so much related to RDF, as to the novelty of these tools
and the lack of experience people have with them. Over time I think
it is quite likely that the basics will be dealt with. I would like to
note also that the semantics does not says temporal reasoning
is impossible, only that it is beyond the research scope of the 
document.

The problem as I understand it lies just in the simplicity of that
claim, that a URI denotes some thing that does not change. People then
look at URIs such as <http://www.worldtimeserver.com/> and notice that
it changes, and so the conclusion is easily made that rdf does not model
the web correctly. But what we know when we fetch the above web
page is that we are getting a description of that resource. Since we
can get the resource over time, the problem is exactly the same as the
one I described with my Madonna post. Each of these returned
descriptions are descriptions of that resource returned at different
times, but the whole resource itself is described only by the whole set
of those description.

The parallel with Madonna, the material girl, is instructive.
When the government assigns a social security number to her, they
are naming one resource that does not change. But let us look how
modern skeptics following the ancients could use this to their
advantage to argue that the social security numbering system is broken.
At age 18 Madonna goes to get her drivers license. The clerk could
write down the following:

<urn:ss:123456789>
   |---name------> "Madonna"
   |---size------> 170cm
   |---sex-------> <female>

  Then at age 25 she may get stopped by the police who write
down the following

<urn:ss:123456789>
   |---name------> "Madonna"
   |---size------> 175cm
   |---sex-------> <female>

An ancient skeptic would have used this to show the inherent
contradiction in the whole social security system.

Of course the answer is that the social security number
only identifies the mereological fusion of all temporal
Madonna parts. If you wish to speak of one temporal instance
of Madonna, you need to introduce a blank node.

:M1
  |---ssn---> <urn:ss:123456789>
  |---name------> "Madonna"
  |---size------> 170cm
  |---sex-------> <female>
  |---age-------> "16 years"

ssn is a functional property of the temporal madonna part M1. It
is not an identity property, and neither is it inverse functional.
It is therefore possible to have numerous Madonna parts each with
different descriptions.

No problem really.

Notice the incredible similarity with my arguments concerning
the Atom feeds. An Atom feed describes various resources, Entries,
that can change over time.

If you insist on describing the resource directly with a URI
you will either force the URI to be the URIs of a time slice, going
against the syntax of http urls especially (since they can be used
over time), or you will say something false.

<http://www.worldtimeserver.com/>
    |----content----> "the time is 9:24am in Cuba"
    |----date-------> "14:24 French time"

<http://www.worldtimeserver.com/>
    |----content----> "the time is 10:00am in Cuba"
    |----date-------> "15:00 French time"


That of course is again a contradiction, and your ancient
skeptic would have used that to prove that time does not
exist.

Of course we can identify via a blank node a time slice
of the resource

:T1
  |----content----> "the time is 9:24am in Cuba"
  |----date-------> "14:24 French time"
  |---funcId------> <http://www.worldtimeserver.com/>


And with another blank node identify another time slice
of the resource

:T2
  |----content----> "the time is 10:00am in Cuba"
  |----date-------> "15:00 French time"
  |---funcId------> <http://www.worldtimeserver.com/>

There you go. We have it down to a T. There is no
contradiction here. Not within a document and not across
documents.


[1] http://www.amazon.com/exec/obidos/tg/detail/-/0631224254/
--------------------------------------------------------

My formalization of the madonna example is close to yours
except that I did not intend M1 and M2 to be time instants
but rather blank nodes, with time properties.

Thanks again,

	Henry Story


On 9 Feb 2005, at 17:22, Jeremy Wong wrote:

> Let's see the following triples...
>
> madonna rdf:type owl:Class
> madonna1 rdf:type madonna
> madonna2 rdf:type madonna
>
> madonna socialSecurityNumber "123"
> madonna1 atMoment "M1"
> madonna2 atMoment "M2"
> madonna2 fingernail "clipped"
> madonna2 toenail "clipped"
>
>
> --
> Jeremy
>
> ----- Original Message ----- From: "Henry Story" 
> <henry.story@bblfish.net>
> To: <semantic-web@w3.org>
> Cc: "Roy T. Fielding" <fielding@gbiv.com>; "Atom Syntax" 
> <atom-syntax@imc.org>; "bloged" <users@bloged.dev.java.net>
> Sent: Wednesday, February 09, 2005 5:42 PM
> Subject: Madonna
>
>
>>
>> Just a point of clarification about identity. (I thought the example
>> is fun enough that it may be of interest)
>>
>> As we all know Madonna is a material girl, and she lives in a material
>> world [1].
>>
>> Madonna is a US citizen also. Let us imagine the day she received her
>> social security card through the post. That day Madonna, the material
>> girl, had a very specific material constitution. Let us call the set
>> of cells, organic and inorganic material that composed Madonna that
>> moment M1.
>>
>> Now let us stipulate that two material things are equal if they are
>> made of exactly the same matter.
>>
>> That afternoon Madonna goes to the beauty salon where she gets her 
>> hair
>> cut and fingernails and toe nails clipped. The resulting material girl
>> call her M2 is no longer the same as M1, as M2 is M1 less some hair
>> and nail material. M2 may also contain a little more water (she was
>> drinking some beauty drink during the process). So M1 is not the same
>> material thing as M2.
>>
>> Both M1 and M2 are also girls, and have a functional relation to a
>> social security number. Ie M1 and M2 each only have 1 social security
>> number. There are in fact a very large number of such material 
>> Madonnas
>> Mx that have a functional relation to that same social security 
>> number.
>> So clearly the inverse relation (from social security to temporal 
>> Material Madonnas) is not functional.
>>
>> Now let us think of the mereological fusion [2] of all of these
>> Material Madonnas, and let us call that MM. MM also has a functional
>> relation to the social security number. But it is also clear that
>> there is an inverse functional relation from social security number
>> to MM.
>>
>> So the social security number identifies madonna in two ways.
>>
>> 1. via a functional relation
>>     temporalId:  temporal material madonnas --> social security number
>> 2. via an inverse functional relation
>>     fusionId: MM --> social security number
>>
>> Given a social security number we can via the fusionId relation
>> get a hold of the mereological fusion of that person, and from
>> the mereological fusion we can find the individual temporal parts,
>> each of which of course have a temporalId relation to the same social
>> security number.
>>
>> The above can probably be also put in terms of set theory, and
>> so in terms of rdf graphs. In which case it clearly shows how
>> rdf can speak about temporal and non temporal identity.
>>
>> Henry Story
>>
>> [1] http://www.lyricsfreak.com/m/madonna/86925.html
>>     audio clip: http://tinyurl.com/5k9et
>> [2] http://plato.stanford.edu/entries/mereology/
>>
Received on Wednesday, 9 February 2005 16:46:33 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:44 UTC