[
Lists Home |
Date Index |
Thread Index
]
On Sun, 27 Apr 2003 13:41:49 -0400, Thomas B. Passin <tpassin@comcast.net>
wrote:
> There may not be "the" semantic definition, but rather many that are not
> fully congruent. No matter what semantic classification you set up, some
> application might know better for its purposes. Therefore, make sure to
> leave enough info in the document that such an application would be able
> to do its job.
That's what I take from Walter as well. Or perhaps "to thine own semantics
be true" ... figure out what information YOU can extract from a certain
class of documents/messages to meet YOUR purposes. If you can re-use
someone else's semantics, or if your industry might standardize on your
semantics, so much the better, but don't get your hopes up.
It's hard for me to strongly disagree with Roger's points about the
downside of application-specific semantics, but the alternative: 1)
Extricate semantic definitions from applications
2) Standardize semantic vocabulary
seems a tiny bit over-optimistic :-) As Sean's post earlier noted, a lot
of "applications" simply don't have extricateable semantics (e.g. Word,
unless you think of "Heading 1 ... paragraph ... bulleted list" as
'semantics'). Likewise, a real-world application might have its semantics
so tightly coupled to a specific business' processes that it's not clear
what benefit one could get from extricating them. I'm sure that there are
a lot of examples in the middle between those extremes of over-generality
and over-specificicity, but as anyone who has participated in a
standardization effort can attest, the different politics, personalities,
perceptions, priorities, etc. of the standardization participants add an
AWFUL lot of devils to the the details that must be rigorously specified.
|
