[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: [topicmaps-comment] on occurrence
this message is
also posted at:
***
Ray (Bradley) - in
eventChemistry
Occurrence is an
important concept to many - including the topic maps community. It is a
way of asking where does this abstract notion become reified? In topic
maps the "occurrence" can be something in the computer, like a file, or it can
be outside of the computer, like me.
I must say here
that the topic maps paradigm has not yet quite developed the agile-enough
standard so that general knowledge is being "encoded", not if we mean by
"knowledge" as most cognitive scientists mean. If fact, it
does seem improper to say that knowledge can exist "in" a
computer.
Information may be the organization of data, and data may be
occurrences of a specific type.
... And there is an
epistemic gap between the perception of information (by a living system) and the
experience of knowledge.
So the question
of even the possibility of a standard for "knowledge representation" can be
easily (and justifiably) re-visited from time to time... (Topic maps is
clearly a front runner, in my opinion; as is Ballard's system).
But this is an issue
of how we wish to use our language, is it not?
Dick Ballard has a
notion of encoding knowledge, and often I am at a loss to express my reservation
about the notion of "encoding". But Dick, I think, knows that he is
talking about encoding information structures, not knowledge encoding.
On the other hand,
if an encoded something-or-other is the primary trigger for an experience of
knowledge, then perhaps topic maps and Ballard's Mark 3 system, and the Cyc
system are "knowledge encoders". But what about Stanford's Protégée, used
primarily as a mdeical ontology. Who knows? What are the limitations
of these knowledge encodings? How might they be of value to society?
I am interested in
the topic maps' community viewpoints on this issues... if they are
interested in coming at the issue again.
What we seem to be
able to do is to encode abstractions that are understood by humans in some
predicable fashion, and that have some correspondence to some set of
"occurrences". The occurances may seem to have (or to actually have)
associations and properties.
In your recent post
to the eventChemistry forum
your
said:
"To
the degree that this holographic-like organization of
information/communication processing is a general feature of interaction
ALL in social settings, then the principles of communication that Karl
(Pribram) and I formulate in our "Communication and Stability in Social
Collectives" article should provide the theoretical basis for understanding, and
therefore modeling, interaction among terrorist groups and networks and
interaction among hackers."
The argument that
hypothesis are verified is not always straight forward... as this gets us
to the notion of what is science. Hypothesis about holographic-like
organizations of information/communication processes is more difficult to make
science out of that is one of the equations from Newton's work. The
reason may be due to the stratified nature of the experience of
knowledge by an individual, and of the sharing of knowledge in communities of
practice.
In a stratified
science, the notion of evidence has to be de-constructed and
re-constructed. I am not at all sure how this will work out. But I
do suspect that the problem of reification, as discussed within the topic maps
community, must address the concept of truthfulness or fidelity of an
abstraction (such as a topic) to something. A pattern of events occurring
in the computer, for example, might be a topic; the occurrence of which is
not be in the computer because the pattern is only recognized as a
pattern through the perception of an individual. The pattern might
even be a statistical artifact, or an artifact produced via a stochastic process
(such as scatter-gather methods in text understanding). So (oddly enough)
this "topic" has an occurrence that is
1) not in the
computer
and
2) not in the real
world either
In the incident
detection problem, the patterns of greatest importance are those that trigger in
the mind of an expert the sense that something might be going on. A
process of investigation must then follow this. The perceived pattern can
not be said to exist except as a perception. Pribram and I have have
recent discussion about this that involved the literature from
For example, given
that somehow the powers that be allow us to work on an event detection system
and we wish to "detect" the events being generated by terrorist in the Internet
(using common hacker tools). We need to be able to build patterns from the
abstraction produced by categorization of occurrences (the being together of
many instances of occurrence into an abstraction that in some way accounts for
any of the occurrences in the abstract.)
But the evidence
that you put forward in your book is that social organization can be mirrored in
some ways by taking a number of occurrences of a question about the relationship
between pairs of individuals and from the total information of this type to
construct a diagram. The diagrams so constructed can be be characterized
as having significance on issues such a the stability of the group.
A non-local
distribution of the structure of the community that you model is represented as
a graph. Yes?
There are
differences and similarities between the question gathering process and the
consequence representation of results in the graph.
|
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC