At 12:33 PM 12/4/1998 -0500, you wrote:
>The searching power that this addition to conventional search engines
>will afford us will be great. Some editors (FM+SGML) and browsers
>(Dyntext) currently offer this sort of functionality.
>The primary problem with offering XML-enhanced search engines is that
>users won't know which elements to search for. You may have defined your
>author element as 'author', but I may have defined it as 'auth' or
><PERSON role="author">. We could establish a
>tag library, with standardized element names for search engines, or rely
>on RDF. Comments?
If you use a search protocol such as Z39.50 which allows users to specify
standardized access points (author, title, corporate entity, date of
publication, etc.), and allows servers to inform client software of
available access points, you remove quite a bit of this problem for users
by placing the burden of mapping from the protocol's access points to local
naming conventions on the search engine. For it to be globally effective,
of course, you'd have to persuade people to abandon the HTTP protocol or
provide both HTTP and Z39.50 access to resources. I'm not sure which would
be more difficult: persuading people to globally implement Z39.50 access or
persuading people to globally standardize element names. But I suspect
that in the long run, abandoning
HTTP for something with more sophistication is the better route.
Jerome McDonough -- [log in to unmask] | (......)
Library Systems Office, 386 Doe, U.C. Berkeley | \ * * /
Berkeley, CA 94720-6000 (510) 642-5168 | \ <> /
"Well, it looks easy enough...." | \ -- / SGNORMPF!!!
-- From the Famous Last Words file | ||||