Re: [topicmaps-comment] TMs & XTM [Was: skills to create topic maps]

["Steven R. Newcomb" <srn@coolheads.com>]

Tony.Coates@reuters.com writes:
> On 29/11/2001 19:34:16 "H. Holger Rath" wrote:

> >- And finally (I couldn't resist) about TMs and
> >XML. XTM is the interchange syntax and I would never
> >ever edit or tweak a TM in XML.  Same as you would
> >produce an ASCII dump out of Oracle, edit it in
> >emacs and load it again into Oracle. Sorry to say,
> >but XTM could also have been ASN.1-based, which is
> >somehow unreadable binary stuff - it would fulfil
> >the same purpose of a standardized interchange (!)
> >format. We selected XML because it is hype and it
> >makes some sense to have XTM (one of these senses
> >was the marketing effect!), not because it is a
> >readable format you can edit TMs in.

> What a great loss, though, if XTM is nothing more
> than an opaque transmission format between topic map
> engines.  If everyone took that view, I think it
> would kill a lot of the existing enthusiasm for topic
> maps, because it would suggest that there is no
> standard way of connecting your application to the
> information in a topic map.  The great value of XML
> is that there are so many tools available that can do
> useful things with it.  I think you should give more
> thought to the value of being able to intercept XTM
> travelling between topic map engines and apply
> standard XML tools to it.  Assuming that XTM is just
> a private format for the use of topic map engines
> greatly undervalues its possibilities, in my opinion.

Topic Maps are essentially graphs.  There can be any
number of syntaxes for interchanging these graphs.

The XTM 1.0 syntax is a really good syntax for
interchanging topic map graphs: redundancy is
minimized, and human readers of the XML instances can
understand the topic maps that are represented by them
pretty intuitively.  The XTM 1.0 syntax is also *very*
good for introductory learning about what topic maps
are and how they work.  For these reasons, and others,
I believe the XTM 1.0 syntax is the best choice for the
standardized interchange of topic maps.  It's a "market
opening" syntax, and (thanks be to God) it's powerful
enough to keep on working well for the foreseeable
future.

It is impossible to design a syntax that meets *all*
requirements; requirements conflict with one another.
The requirement that redundancy be minimized is a
perfectly valid requirement for interchange syntaxes,
and XTM 1.0 respects this requirement.  XTM 1.0
therefore does *not* meet the requirement that Tony
implies in his comment, above, namely that DOM
implementations be able to be used directly to gain
access to the topic map graph.  For Tony and others
like him, the good news is that it *is* possible to
interchange topic maps by means of an XML syntax that
accurately and directly reflects the topic map graph,
so that DOM implementations can directly use the topic
map without needing a Topic Map engine to translate the
XML.  In other words, topic maps can be published in
XML in a way that obviates the need for a topic map
engine.  There's a price, though: the XML instances are
necessarily chockfull of redundant information, and the
tasks of creating and maintaining such instances cannot
be accomplished by hand.  (The smallest tweak is very
likely to render the instance inconsistent with
itself.)

A syntax for Topic Maps that can meet Tony's
requirement can be found at
http://www.topicmaps.net/TMGraphAPI3.htm.  It's very
redundant, but something like it could be used in the
way that Tony wants.

-- 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