OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

geolang-comment message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]


Subject: Re: [geolang-comment] First proposals for ISO 639 and 3166 available


At 15:50 30/08/02 +0100, Murray Altheim wrote:
>[lots of stuff about the separation between the map and the territory]

Thanks for quoting me in your defence, but I agree with Lars Marius
on this issue.

Yes, there are two layers, and the map/territory analogy is useful,
but it only works up to a point.

The way I prefer to think about this is that the two layers represent
what (for want of better terms) I will call "knowledge"[1] and
"information" respectively. The knowledge layer captures "knowledge"
that is (or could be) present in the information layer.

One of the key decisions in designing any topic map application is to
decide how much of the knowledge to lift out of the information (the
contents of the external occurrences) and put into the knowledge layer.
Internal occurrences (along with names and associations) are part of
the knowledge that has been lifted into the layer above.

In the Italian Opera topic map I do this with, for example, the birth
dates of composers. They are represented as (internal) occurrences.
(They could have been associated topics as well, but that would have
been overkill for the purpose of my application.) The date of birth
of a composer is "information that is relevant to that topic". Because
it can be considered core knowledge about that topic and because it's
just a short string, it made more sense to create internal rather than
external occurrences.

Is that piece of knowledge/information part of the territory or the
map? Simply asking the question like that demonstrates, at least to
me, that the map/territory analogy breaks down at a certain point.

> > I don't see how you can dispute that. Name strings are information
> > resources, and they are clearly relevant to the things the name.  So
> > names are a specialization of occurrences.
>
>Your conclusion does not logically follow from your thesis for a
>number of reasons. Non-transitivity of instance-of relations and such.

I don't see how this invalidates the notion of names as specializations
of occurrences. Can you explain?

>In the topic map model names are not occurrences of the topic, they are
>the names (or labels) of the topic. Do you not make any distinction
>between a topic name and a topic occurrence?

Of course we do: Names are specialized information resources whose
pertinence to their topics is that they provide names or labels.

>If your argument were correct, everything in a topic map, being resources,
>could be some specialization of occurrence of a topic. We could have
>designed topic maps so that there were no topic names, that topic
>names were simply an occurrence within the scope of "topic name." That's
>not how it was done, nor does that make sense within the topic map model.

We could have designed it that way, yes. And we could have gone one step
further and said that occurrences were just specializations of associations,
so "Away with them, as well!" ... and we would have ended up with RDF. The
reason we didn't is that the distinction between names, occurrences and
associations provides an extra level of semantics that is extremely
useful. (We don't disagree about that, I assume. Just about the exact
nature of occurrences.)

If, as you believe, all occurrences are part of the information layer
(or "territory"), why did XTM make explicit the notion of "internal
occurrences" (in the form of <resourceData> elements)? Surely the mere
existence of resourceData (i.e. internal occurrences) rebuts all your
arguments?

> > | I think the XTM specification is starting to be read like the bible,
> > | wherein one can seemingly prove anything.
> >
> > True, and that's why we are replacing it.
>
>When the Protestant Reformation occurs and you publish your New
>Testament, I'll likely stay a Catholic if it causes all existing
>software to break. I hope the changes are by way of clarifications,
>not redefinitions.

I think you can safely assume that that is the case.

> > Sure, but who says topic maps are about that separation? As far as I
> > know, only you. Personally I don't agree that that is a consideration
> > in this case.
>
>Then perhaps you should talk to Steve Newcomb and Michel Biezunski.
>If you can't do that, read Jack and Sam's book, which has a chapter
>written by Steve on this subject. During one of the TopicMaps.Org
>AG meetings, Steve N. drew a diagram of a big gulf between that map
>and its territory, his whole point about global information interchange
>being reflected in how humans attempt to represent knowledge being
>the need to bridge that gulf, and that topic maps attempt to do just
>that. Ask Steve P. about it, I believe he was there.

That diagram was not at all about the two layers we are discussing
here, i.e. the "map" or "knowledge layer" (mostly topics and associations)
and the "territory" or "information layer" (external occurrences). It was
about another "gulf" altogether: that between topics and the subjects
that they represent.

>Or look at Steve
>Pepper's [TAO] or Steve N's [SN] Michel Biezunski's diagrams [MB]
>showing topic occurrences being down in "the territory" (eg., the
>Web), with names and the associations (interrelations) between
>topics being up in the map.

Those diagrams, as I said above, are a useful oversimplification that
is absolutely necessary in order to get the core concepts across to
people new to the topic map paradigm. They cannot pretend to capture
the full complexity of the topic map model.

>Is there some difficulty in understanding why topic maps attempt to
>create a bridge between the map and the territory? It's reflected in
>everything that topic maps have been about since the beginning. It's
>why they're called maps. I'm not sure how we're disconnecting here,
>as these definitions are hardly mine, nor do I think I'm alone in
>my understanding of them.

The term "map" is a metaphor which shouldn't be taken too literally.
It's as simple as that.

Steve

[1] I absolutely refuse to be drawn into yet another discussion about
the exact meaning of the term "knowledge" ;-) That shouldn't be
necessary in order to understand what I am trying to say here.

--
Steve Pepper, Chief Executive Officer <pepper@ontopia.net>
Convenor, ISO/IEC JTC1/SC34/WG3  Editor, XTM (XML Topic Maps)
Ontopia AS, Waldemar Thranes gt. 98, N-0175 Oslo, Norway.
http://www.ontopia.net/ phone: +47-23233080 GSM: +47-90827246
Goldfarb/Pepper/Ensign: "SGML Buyer's Guide" (Prentice-Hall, 1998)



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]


Powered by eList eXpress LLC