Re: A serious detail point
On Thu, 17 Apr 1997 11:53:17 -0400 Henry Thompson said:
>Here's the problem: I want to have tens of thousands of links in a
>single document, all pointing to (spans in) ONE target document.
>Without a way to specify a default containing resource OTHER THAN the
>document contain the linking element, I will have to reproduce the URL
>for my target document in the specification of EVERY ONE of those
>Or have I missed something?
Not that I know of.
But why is it plausible to suppose that it will be more common to
have tens of thousands of links all pointing at the same host
document, than scores of thousands of links all pointing at various
documents in a corpus or collection?
That is, why is the case where out-of-line pointers all point to
the same document worth optimizing for? I don't even think it's
going to be very common, let alone the dominant method of out-of-line
markup or annotation.
You may say, well it's because for each sample in the corpus I have
exactly one para-file with its grammatical analysis. And I say,
"you don't even bother to point to a lexicon or a set of
feature structures or anything at all where you document more fully
what your grammatical annotation means?"
I don't insist that you point both to the text being annotated and to
a lexicon or other standard object. But I do insist that when I do
wish to point both to a lexicon and to a text being annotated, that
I get the same special treatment you get. So now n=2. Who will
speak up next, for three- and four-sided annotation?
The update and maintenance problem is handled nicely by the general
entity mechanism you illustrate. The caching problem can be handled
with affinity groups / BOS / whatchamacallits that say "Cache this
one, I'm going to need it often".
I suffer from the fact that I didn't hear your talk and may not have
the right attitude. But on your showing so far, this looks like an
uncooked special case. I don't think we have any reason at all to
suppose that para-documents always pointing at a SINGLE other
document are likely to be a common enough case to optimize for. I'm
open to persuasion, but so far not persuaded.
-C. M. Sperberg-McQueen
ACH / ACL / ALLC Text Encoding Initiative
University of Illinois at Chicago