Re: [topicmaps-comment] RE: [sc34wg3] Re: PMTM4 and XTM Layer 1.0

[Bernard Vatant <bernard.vatant@mondeca.com>]

Let me be a little provocative : Any discourse about *what the 
subject of a topic is* is definitely pointless. 
What is needed are ways of *validating an agreement* about the 
fact that "we" have the same subject of conversation. And "we" 
here includes every actor exchanging information: this holds for 
human-human conversation, system-system conversation, and 
human-system conversation. 

What we can get from analysis of assertions expressed by 
characteristics, names, occurrences, roles or whatever, is a weak, 
local and transient form of agreement, valid in the limited context of 
the ongoing conversation. I agree with Tom: that is indeed how 
people do most of the time, because it's quite effective in 
everyday's conversation, where the context is generally implicit, 
and explicited by the speakers as often as necessary.
But the conflicts in communication (and conflicts in general) come 
out of irrelevant extension of this agreement outside of its implicit 
validation context. And this irrelevant extension is made because 
the context is not formally defined, and/or because the actors are 
not really aware that there is a context at all. That is what has 
happened so far in the three above quoted types of conversation, 
with all the known consequences ...

How can we transfer the flexibility of this local and transient 
agreement process into system-system conversation and human-
system conversation, and avoid at the same time the potential 
resulting conflicts and ambiguities? That must be grounded on 
formal reference to subject identifiers, the most possible 
independent from characteristics. Those identifiers won't ever be 
able to tell what the subject "is", but should tell who has set an 
agreement to use it, for what and in what context. For example, I 
know a standard is a standard not by looking at what it *is*, but at 
who has made it, who has recommended it, and who used it and 
for what. And if I use it myself and make that usage known, it's a 
little more of a standard.

In the choice of a tool for identification, our main concern should 
therefore be: will this tool provide a valid process of agreement in 
the context it will be used? rather than: will this tool really identify 
the subject?

Bernard


On 11 Oct 2001, at 1:08, Thomas B. Passin wrote:

[Steven R. Newcomb]

I have always strongly resisted
>           this dangerous and false doctrine, and I'm still resisting it,
even in this
>           case.  The doctrine is:
>
>              We should be able to tell what the subject of a topic is
>              by analyzing its characteristics (i.e., in PMTM4 terms,
>              by analyzing all the associations in which it plays
>              roles).
>

But Steve, this is exactly how people know about most things (including most
words).  Were it otherwise, you would not have needed this long posting but
only one or two sentences.  For people, everything has connotations and
associations, and all of those adjust one's understanding of the thing.
It's not exactly "analyzing its characteristics" but it's in the same
ballpark.

How is a computer or even a person supposed to know accurately what is
intended by a subjectIndicatorRef?  The only way is for a person to read or
look at it and try to extract some gist, essence, or other understanding,
then write that into a program (if we are talking code), or for a person or
computer to apply some heuristics to the addressable resource to at least
get a label or descriptive phrase.

Having an ontology does not really address it either.  The hierarchy
animal-mammal-dog does not tell you what a dog is - there is no way to
distinguish a cat from a dog this way.  It may tell you how to reason about
a dog, or how to use it correctly in an association or sentence, but that's
not the same as knowing what it is.

It's is really a matter of "intensive" vs "extensive" definitions, and there
is a need for both.  Extensive is harder for a computer but probably more
natural for people.

Cheers,

Tom P


----------------------------------------------------------------
To subscribe or unsubscribe from this elist use the subscription
manager: <http://lists.oasis-open.org/ob/adm.pl>

***********************************
Bernard Vatant - Consultant
bernard.vatant@mondeca.com
Mondeca - "Making Sense of Content"
www.mondeca.com
***********************************