[sc34wg3] CTM: Realistic use cases or toy examples?

[Steve Pepper]

* Lars Heuer
|
| I started a pre-Kyoto vs. post-Kyoto comparison for CTM based
| on our CTM use cases:
|
|     <http://www.semagia.com/tmp/ctm-comparison.html>
| 
| I hope this helps a bit to see the differences.

Thanks. It is somewhat helpful, but it is also rather
misleading. At least, *I* think it is misleading ;-)

Others may disagree. If so, the disagreement probably reflects
entrenched positions. On the other hand it might reflect
different priorities and expectations. If so, it might help us
move beyond the current impasse.

The reason why I find the table misleading is that almost all
the use cases are atypical and unrealistic (according, of
course, to my expectation of what a typical CTM document will
contain).

In a typical CTM document (as opposed to toy examples) most
topics will have multiple "characteristics" - typically two,
three, four or more.

In terms of identifiers, many topics, sometimes the majority,
will have exactly one subject identifier, no subject locators,
and maybe an item identifier; of the rest, all but a few will
have a single item identifier and nothing else; there might be
one or two subject locators and one or two topics with multiple
subject identifiers; there will be none with multiple item
identifiers (unless the topic map is the result of a merge, in
which case it is less likely to be in CTM).

This description of what I regard as a "typical" CTM document is
based on my experience using LTM to hand-author a very large
number of topic maps.

Turning to the use cases, only #14 is typical. All the others
contain too little information to be considered typical: they
are toy examples.

This explains why the curly braces and the semi-colons feel
unnecessary in examples ##1-13. If there's only a single
statement or identifier, it really seems like overkill to
enclose it in curlies and require that it be terminated with a
semi-colon.

However, the semi-colon is not intended as a terminator; its
purpose is to separate multiple statements clearly from each
other. And the curlies are there to demarcate a block that
consisting of several statements.

Neither of these may be *necessary* from the point of view of
parsers (and those who speak on their behalf). However, CTM is
for people, not parsers.

Steve
 
--
Conference Chair, Topic Maps 2008
Oslo, April 2-4 2008
www.topicmaps.com