Future of meta-indices: site indexing proposal and Perl script

Robert S. Thau (rst@ai.mit.edu)
Tue, 22 Mar 1994 22:21:20 --100

Date: Tue, 22 Mar 1994 15:57-0500
From: John C. Mallery <jcma@reagan.ai.mit.edu>

In fact, you want to make assertions about the document using typed
links, eg. "abstract abstracts document". It would seem the mechanism
is available in the protocols, if not all the implmentations.

The power is the links not hacks to the markup language.

The trouble with this idea for my application is that, from a brutally
practical point of view, it's as much blood, sweat, and tears to maintain a
setup like this (links to abstracts, etc., in separate files) as it is to
maintain the IAFA template files manually. In fact, it's *more* trouble
--- you get two URIs per document to type in (and check for typos), one for
the keywords attribute and one for the description --- and on top of that,
you've suddenly got a slew of itsy-bitsy files to keep track of. The whole
thing turns into a configuration management nightmare.

The bottom line, as far as I'm concerned, is this: If people can get their
files into an index by adding two lines of <meta> tags, *and that's it*,
there's a chance that they will. If, instead, they have to create two
small files and a couple of cross-references, that chance is much reduced.

If I wanted a system about which it's easy to build neat theories, I might
do it your way regardless, but that's not my goal here. I'm looking to
save work (for myself, and eventually, the other people with files on the
main lab server), and I'm prepared for elegance to give way to that where
it has to. (If I wasn't, I suppose I'd be writing Lisp instead of Perl ;-).