Lists Home |
Date Index |
email@example.com (Roger L. Costello) writes:
> Should the semantic definition (2.1) be:
> (a) hardcoded and buried within each application,
> or, should it be:
> (b) declaratively stated in a separate document,
> using a standard, well-defined vocabulary (i.e., OWL).
>I argued for the later, (b).
This seems to me a false choice. There's a range between (a) and (b),
and probably 95% of all XML documents and applications fall inside the
range, not at either of its endpoints.
I'd phrase the two perspectives somewhat more neutrally as "local" and
Local semantics tend to be far richer, and easier to use without deep
thought - think about the level of communication we have with family
members, or even co-workers. Global semantics are useful, but tend to
either be watered-down (think AP stories) or to require negotiation
(turning to dictionaries for words out of range, or just plain asking).
OWL seems to be an effort to reduce the cost of global semantics, but I
have serious doubts about the value of that project as expressed here
There's a second range I'll describe as "minimal" or "ambitious",
describing the scope of the project. Minimal projects include the small
and naturally limited vocabularies people write for a particular program
(config files are a classic), while ambitious vocabularies try to cover
as much as possible of a given field (maybe TEI is a good example).
It's possible for a project to have global reach and minimal scope, or
local reach and enormous scope. Some projects start out local or
minimal and creep their way up to something larger through broad
acceptance and new needs, but the track record for such projects is
pretty mixed - conflicts definitely become an important (and valuable)
part of the picture as range and scope expand.
IMHO, costs increase dramatically as a project moves from local to
global and minimal to ambitious. Abolishing negotiation in favor of
prior agreement reduces one kind of cost, but also limits flexibility
and generally has both initial agreement costs and ongoing costs as
users try to map their understandings to official semantics.
The classic 80/20 (80% of capability, 20% of cost) description seems to
fit this story pretty well, though maybe it becomes 64/4 if you plot it
on two axes. Maybe that's too much to hope for.
Ring around the content, a pocket full of brackets
Errors, errors, all fall down!
http://simonstl.com -- http://monasticxml.org