[sc34wg3] occurrence - basename fuzzy boarder

[Bernard Vatant]

Some thoughts following up Baltimore brainstorming:

Now that TNC is relaxed and typing of names is allowed, the (already fuzzy) boarder
between names and occurrences seems to fade away a little more. And well, that's good news
for those who have always considered this boarder to be a syntactic answer to a
non-question :)
A quick exchange on topicmaps IRC channel yesterday about "occurrence variants"
http://www.ontopia.net/topicmaps/irc/irclog/2002-12-18.html triggered further reflection
on that.

Let's say occurrences and basenames together form a class of "information items". We can
define at least two clear orthogonal subclasses - orthogonal to each other, more clearly
defined than, and orthogonal to, the fuzzy name-occurrence classification.

1. Identifier vs Non-identifier
Some types of names (e.g. codes) will be used as unique identifiers, and some will not.
Some types of occurrences could also be used as unique identifiers (think about biometric
identifiers, e.g. fingerprints), and some should not.
And it figures: where is the boarder between this name-occurrence superclass and subject
indicators for that matter?

2. Internal vs External
There is more difference between an external occurrence of type e.g. "Home Page" defined
by a <resourceRef> link and an internal occurrence of type "Short Description" defined by
a <resourceData>, than between the same "Short Description" and some basename of type
"Long Name".

Of course following this track would lead to something hardly compatible with XTM 1.0
syntax, but maybe compatible with the Reference Model. For SAM I don't figure clearly.

Bernard

-------------------------------------------------------------------
Bernard Vatant

Senior Consultant - Knowledge Engineering
Mondeca - www.mondeca.com

Chair - OASIS TM PubSubj Technical Committee
www.oasis-open.org/committees/tm-pubsubj/
-------------------------------------------------------------------