[topicmapmail] Identities and names (WAS - A somewhat new topic maps format)
Daniel Rivers-Moore
Daniel.Rivers-Moore@rivcom.com
Fri, 8 Aug 2003 17:01:35 +0100
Bernard
Thanks for this. A few responses:
> [1] For the (hopefully improbable) reader still unaware of Lewis
Carroll's
> pionneering work in those difficult matters back in 1871, when the
oldest
> and wisest of us all were barely born, this is a *must* background
reading:
> "Alice Through the Looking Glass, Chapter VIII : It's My Own
Invention" .
Chapter VI (Humpty Dumpty) is also well worth reading, for what it has
to say about words and their meanings.
> Actually, Murray's page is not a proper topic map IMO, but a PSI set
...
OK
> > "X is called 'A'" is synonymous with "'A' is a name of X"
>
> I guess Lewis Carroll would have bought that one :)
Actually, I'm not sure he would. To quote a few snippets from the
chapter you cite:
"The name of the song is called 'Haddocks' Eyes'"
"That's what the name is _called_. The name really _is_ 'The Aged Aged
Man'"
"The _song_ is called 'Ways and Means'"
So it seems that for Lewis Carroll, the song's name and what it is
called are not the same thing, whereas I am asserting that if the song
is called X, then X is the (or a) name of the song.
Reading between Carroll's lines (and I haven't read Martin Gardner's
commentary), I imagine the distinction he had in mind between
something's name and what it is called is analogous to Murray's
distinction between a formal name and an informal name. Murray's rabbit
is _called_ 'Bunzilla' (that's what Murray actually calls it; it is the
rabbit's informal name), but its _name_ (formally) is 'Burroughs Deeply
Underhay' (... and what that formal name _means_ would seem to depend on
whether you see it written or hear it spoken! ;-) ).
> Right. But as Murray states in his answer, this name has some declared
> identification context and mechanism (or protocol), so it's not
exactly
> *any* kind of name ... See below.
Agreed. But then I also agree profoundly with both you and Murray that
naming always require specific context (domain of validity), and
protocols/processes (for dereferencing), so _no_ name is *any* kind of
name. Or, to put it differently, every name is a particular kind of
name. (Is that a tautology? ;-) )
> > Strictly, we know that in Topic Map terminology,
> > "http://www.altheim.com/bunny/#bid0537" is the subject indicator for
the
> > topic whose subject is the name "Bunzilla".
>
> Well, you should update your terminology here, to agree with most
recent
> both SAM and PubSubj TC specifications:
> "http://www.altheim.com/bunny/#bid0537" (the URI) is the subject
> identifier. The subject indicator is whatever human-readable resource
you
> get when pulling this handle through http protocol ... can of worms
...=20
Thanks for the clarification. I was using (I believe) the XTM 1.0
terminology. But I agree that the update is clearer, more
rigorous/correct, and extremely useful.
...
> Of course, the subject identifier barely indicates the subject to the
> (human) reader. Without leading to a subject indicator, the URI
> "http://www.altheim.com/bunny/#bid0537" does not say much. But note -
this
> is an overlooked thing - that getting to whatever subject indicator
you
> get through this URI is just another use of the same name for
> identification purpose, but in another context (human use) and another
> protocol (getting a human-readable resource through http).
Yes. Very good.
> And indeed, figuring out how subject identifiers can be used across
> different contexts and through different protocols is the crucial
> interoperability challenge we face now.
I suspect that there are other interoperability issues, but certainly
this would appear to be a fundamental one for building anything
approaching a semantically capable cross-application architecture, such
as the dreamed-of Semantic Web.
> It goes beyond the "Web identity crisis" about URIs, or this crisis is
> just one aspect of it, and in fact it goes beyond PSIs, because PSIs
are
> linked to this very specific protocol of dereferencing URIs through
http
> for human consumption.
Agreed. A few thoughts on this:
I suspect that one reason for the "Web identity crisis" is the popular
(mis)conception that the world consists of 'things' (or 'individuals')
some of which are spatiotemporal and some of which may be abstract
(though there are a number of attempts to map even the abstract
individuals to the spatiotemporal ones in various ways).
If you believe the primary constituents of the universe are 'things',
then it makes sense to look for a single 'thing' that is referenced by a
formal name in a particular context, independently of a process for
dereferencing it. The reference somehow 'exists', and the process is
only required to enable us to 'discover' it.
Wittgenstein opened his first and classic work, the Tractatus
Logico-Philosophicus[1] with the statement:
"1 The world is all that is the case."
and followed up with the clarifying statement:
"1.1 The world is the totality of facts, not of things."
This is a mind-bending change of perspective, and this change is part of
what makes the Tractatus such a difficult work to understand. However,
as you look further into it, he goes on to say things like:
"2 What is the case - a fact - is the existence of states of affairs."
"3 A logical picture of facts is a thought."
"4 A thought is a proposition with a sense."
"4.001 The totality of propositions is language."
"4.024 To understand a proposition means to know what is the case if
it is true. (One can understand it, therefore, without knowing whether
it is true.) It is understood by anyone who understands its
constituents."
"4.0311 One name stands for one thing, another for another thing, and
they are combined with one another. In this way the whole group - like a
_tableau vivant_ presents a state of affairs."
In other words, despite Wittgenstein's radical shift from things to
facts as the primary objects in the world, the understanding of
propositions (which are 'pictures' of these facts), depends on
understanding the words that make up the proposition, and these still do
reference unique 'things'.
In his later classic work, the Philosophical Investigations[2],
Wittgenstein adopted a position much closer to that of Humpty Dumpty in
Through the Looking Glass chapter 6. In this chapter, Humpty claimed
that he could make words mean whatever he wished. For the later
Wittgenstein, words get their meaning from the meanings of expressions
in which they occur, not the other way round as he had asserted earlier,
and it takes a social usage for a linguistic expression to acquire
meaning; there is no such thing as a 'private language' where a single
individual makes a word mean something outside a social context - so for
the later Wittgenstein, Humpty is wrong to make unilateral and whimsical
meaning attributions, but right to assert that words ultimately mean
just what they are used to mean, neither more nor less.
Why go into all this? - Because here we see the view that language
becomes essentially a _process_. The way a word or an expression in
natural language gains its meaning is through a socio-historical
process, centuries long, and ever-evolving. The way a formal expression
or term gains its meaning is also through a social process, but here a
formalised one and (hopefully!) of rather less duration. And the way a
term, word or expression is _dereferenced or interpreted_, and its truth
or validity decided upon, is also by a process, which often has to be
curtailed before it is complete (if indeed it can ever be complete).
In a formal context, this process is what you and Murray call a
'protocol'. In a natural language context, the process may be a
recursive one, where we verify things outwards from one context to
another broader, more authoritative, or prior one, and continue as long
as we choose, stopping when we are either sufficiently satisfied that we
choose not to make the effort of going further, or because we have
simply run out of time and have to act on the basis of the
interpretation we have now reached.
I like to see all this in terms of the following progression of
thinking:
from
'The world is the totality of things'
to
'The world is the totality of facts'
to
'The world is the totality of processes'.
This last statement certainly seems very much in accord with the views
of modern physics, biology, chaos theory, etc. and I believe is also
extremely helpful in conceptualising computer systems, and the semantics
of formal models and language. The recursive dereferencing of names
through different contexts, as far as is necessary to resolve ambiguity,
is a powerful example of just such a process.
(In an ideal world, I would like to see a procedural model of semantics,
but a declarative programming paradigm - but that's another story.)
Daniel
=20
[1] Logisch-Philosophische Abhandlung (1921), Translated by D.F. Pears
and B.F.McGuinnes. (c) Routledge & Kegan Paul, 1961.
[2] Philosophische Untersuchungen (1945), Translated and Edited by
G.E.M. Abscombe and R. Rhees. (c) Basil Blackwell and Mott, Ltd, 1958
-----Original Message-----
From: Bernard Vatant [mailto:bernard.vatant@mondeca.com]=20
Sent: 08 August 2003 11:13
To: Daniel Rivers-Moore; topicmapmail@infoloom.com
Subject: RE: [topicmapmail] Identities and names (WAS - A somewhat new
topic maps format)
Hello Daniel
> In response to your main message - yes, very important.
Seems that everyone agrees on that today. Good thing. More below.
> In response to your PS2 - you say "Or maybe the other way round ??"
Note that was written while re-reading Alice Through the Looking Glass,
in
the excellent version annotated by Martin Gardner. So ... It's *not* My
Own
Invention ... [1]
> I would say definitely the other way round.
Makes sense ...
> "Bunzilla" is the (or a) name of Murray's rabbit
> "Bunzilla" is also the (or a) name of Peter's rabbit
> "http://www.altheim.com/bunny/#bid0537" is what the name "Bunzilla" is
> called in the scope of Murray's topic map of bunny names and the
subject
> indicator formalism.
Actually, Murray's page is not a proper topic map IMO, but a PSI set ...
> "X is called 'A'" is synonymous with "'A' is a name of X"
I guess Lewis Carroll would have bought that one :)
> So you can also say
> "http://www.altheim.com/bunny/#bid0537" is a name of the name
> "Bunzilla".
Right. But as Murray states in his answer, this name has some declared
identification context and mechanism (or protocol), so it's not exactly
*any* kind of name ... See below.
> Strictly, we know that in Topic Map terminology,
> "http://www.altheim.com/bunny/#bid0537" is the subject indicator for
the
> topic whose subject is the name "Bunzilla".
Well, you should update your terminology here, to agree with most recent
both SAM and PubSubj TC specifications:
"http://www.altheim.com/bunny/#bid0537" (the URI) is the subject
identifier. The subject indicator is whatever human-readable resource
you
get when pulling this handle through http protocol ... can of worms ...
> But in ordinary language, a name serves precisely the function that a
> subject indicator does in the topic maps formalism - it indicates to
the
> hearer/reader what subject is being spoken of. A topic maps subject
> indicator is therefore a particular kind of name.
There again, it seems you confuse subject identifier and subject
indicator.
The subject identifier is indeed a name, and as such it can be used for
identification in a clearly defined context, and through a clearly
defined
protocol : for example the context of a Topic Map application, the
protocol
being interpretation of subjectIndicatorReference by a TM engine for
merging topics.
Of course, the subject identifier barely indicates the subject to the
(human) reader. Without leading to a subject indicator, the URI
"http://www.altheim.com/bunny/#bid0537" does not say much. But note -
this
is an overlooked thing - that getting to whatever subject indicator you
get
through this URI is just another use of the same name for identification
purpose, but in another context (human use) and another protocol
(getting a
human-readable resource through http).
So, in the notion of PSI, we have figured two uses of the
name-identifier
in two different contexts and through two different protocols (a
computer-only one, and a computer-human interaction one). Now, the same
subject identifier could be used also in other contexts and through
other
protocols as well - for example in ontologies, but in a way that remains
to
be defined. And indeed, figuring out how subject identifiers can be used
across different contexts and through different protocols is the crucial
interoperability challenge we face now. It goes beyond the "Web identity
crisis" about URIs, or this crisis is just one aspect of it, and in fact
it
goes beyond PSIs, because PSIs are linked to this very specific protocol
of
dereferencing URIs through http for human consumption.
Bernard
[1] For the (hopefully improbable) reader still unaware of Lewis
Carroll's
pionneering work in those difficult matters back in 1871, when the
oldest
and wisest of us all were barely born, this is a *must* background
reading:
"Alice Through the Looking Glass, Chapter VIII : It's My Own Invention"
.
The "Annotated Alice" by Martin Gardner should still be available
although
it was published in the 60's - Penguin Books - ISBN 0 14 00 1387 3.
Bernard Vatant
Senior Consultant
Knowledge Engineering
Mondeca - www.mondeca.com
bernard.vatant@mondeca.com