[sc34wg3] the scope of topics, subject indicators as subjects, and the meaning of facets

[Steven R. Newcomb]

Here are a few notes inspired by various recent
conversations.

Many people, when they are first exposed to the topic
map paradigm, somehow get the idea that topics have
scope.

Topics don't have scope.  Only their characteristics
have scope, and each such characteristic has its very
own scope.  The PMTM4 way of saying this is, "Only
assertions have scope (and every assertion has at least
one scope)."

The connection between a topic and each one of its
subject identity points (its subject indicators and/or
subject constituter) is always scopeless.  Every
subject identity point of a topic is absolutely a
subject identity point for that topic in all contexts
and for every purpose.

That doesn't mean that we don't need to distinguish
between subject identity points for application- and
context-specific purposes.  For example, for a single
topic, we may have a subject indicator in English and
another subject indicator in Swahili.  

For purposes of topic map processing, the notations or
languages in which subject indicators are notated is
irrelevant.  (Topic map engines don't read subject
indicators; for topic map processing purposes, a topic
map engine is only concerned with the question of
whether or not, for every pair of subject indicators,
they both exist at the same address.)

In the context of particular applications, however, we
may need to be able to distinguish between the English
and Swahili subject indicators, so that we can direct
Anglophone users to the English subject indicator, and
we can direct speakers of Swahili (Swahiliphones?) to
the Swahili subject indicator.  How can an application
do this, if there are no scopes to guide it?

The answer is that every subject indicator can itself
be a subject (of another topic).  In order to make
distinctions between subject indicators, all we have to
do is to create a topic for each one whose subject is
("constitutes") the piece of information that is that
subject indicator.  Once we have such a topic, we can
say any combination of things about it, by giving it
"characteristics" (in PMTM4 terms: by making it play
roles in assertions).  We can, for example, say that
the subject indicator that is expressed in English is
an occurrence of the topic whose subject is the English
language itself.

  Some historical notes: In retrospect, it is hard to
  understand how the blindingly obvious idea that a
  piece of information could itself constitute the
  subject of a topic could possibly have escaped our
  attention for so many years.  Nevertheless, that's
  what happened, even though I personally feel
  embarrassed to admit it.  Prior to this epiphany,
  while designing ISO 13250, we felt it necessary to
  provide special syntactic constructs, called
  "facets", in order to make assertions about pieces of
  information.

  People still love facets, and with good reasons.
  There is no reason to get rid of facets, and I don't
  think I've heard anyone calling for their demise, at
  least not recently.  Indeed, facets continue to be
  desirable, not only in order to maintain backward
  compatibility for existing topic maps, but also for
  other reasons.

  One of the beautiful (and widely underappreciated)
  outcomes of the TopicMaps.Org work is that now we can
  explain what facets really mean, and why they really
  make perfect sense in terms of a single consistent
  model (PMTM4) of what topic maps mean.

  Another beautiful outcome of the TopicMaps.Org work
  is the fact that it demonstrates the possibility and
  desirability of providing more than one syntax for
  the interchange of topic map information, and for
  providing for the possibility that different usage
  scenarios may demand the invention of even more
  syntaxes.  For worldwide information interchange, it
  is neither necessary nor possible for everyone to use
  exactly the same syntax.  The critically important
  thing for the standard to provide is a single common
  model of what all topic map syntaxes must ultimately
  be unambiguously interpretable as meaning.  The PMTM4
  model provides such a common substrate.  Like it or
  not, PMTM4 was just as much a product of the
  TopicMaps.Org work as was the XTM syntax.  To my way
  of thinking, PMTM4 is even more important than the
  XTM syntax, because it provides a platform on which
  to rationalize and explain the whole world of topic
  maps syntaxes, including not only the XTM syntax, but
  also the pre-existing ISO 13250 HyTime-based syntax,
  the NewsML syntax, and all future syntaxes for the
  interchange of topic map information.

-Steve

--
Steven R. Newcomb, Consultant
srn@coolheads.com

voice: +1 972 359 8160
fax:   +1 972 359 0270

1527 Northaven Drive
Allen, Texas 75002-1648 USA