On 07/08/07, Mitchell Katherine A. <kaamitch_at_jeffco.k12.co.us> wrote:
> > -----Original Message-----
> > Ted P Gemberling wrote:
> > > too. Actually, in some ways getting no hits is a useful thing. It's
> > > the "real world." It shows you've gone down one blind alley
> > and need
> > > to try another path.
> > It's a blind alley only if we are unable to provide exits
> > from it. You seem to be suggesting that we should
> > intentionally refrain from automatically providing
> > alternatives in a zero-hit search--that is, intentionally
> > frustrate the user--because it will be of educational value
> > to the user.
> >
> > That's not really what you're suggesting, is it?
> Maybe what Ted's saying is that it *can* be useful to know that the
> library you're searching actually doesn't have anything on, say, humane
> raccoon removal (another story). I just did a keyword search on this in
> my local PL catalog, and when the system found nothing with humane AND
> raccoon AND removal, it changed the search to humane OR raccoon OR
> removal, and gave me 328 hits. Some of which might be useful, but I'm
> not going to wade through them all. I'd rather be given the option to
> re-try my search in Google, or Amazon, another local library, etc. Back
> to moving beyond "books we own" catalogs.
You're searching for a book containing content on "humane raccoon
removal" with those exact terms ANDed in a keyword search? This
sounds, to me, like a very good argument for including tables of
contents, indexes, and other rich content in the catalog metadata so
that there would be a hope in a very hot place that your search would
actually retrieve any results. I just tried that same search in
worldcat.org and got zero hits, as expected. Google Books doesn't do
much better with that phrase.
Assuming that richer metadata is off the table (a very bad assumption
for new records in next generation catalogs, but let's consider the
constraints of those millions of existing records out there and
blue-sky some ideas on how improve search using those existing records
without a massive uplift effort to enrich them--a discussion that
probably deserves a thread of its own), I think that it's a small
salvation that the catalog provided feedback to you on the search
strategy it used to broaden the search.
"Sorry, your search resulted in 0 hits; I've ORed the terms instead of
ANDing them." is at least one step towards helping the patron find a
result that could meet their needs, and it offers concrete feedback
that the original search yielded 0 hits. Finding a reasonable way to
give you useful options (like offering to OR all synonyms for humane;
using an ontology to recognize that a raccoon is_a animal and is_a
wildlife and ferreting out relevant subject keywords on that basis;
determining that the phrase "raccoon removal" often appears in
discussions of animal control on teh interwebs or in a massive
database of full-text) that could lead you to "humane wildlife
removal" (which does give useful results in worldcat), giving you
explicit feedback and control along the way, is how I _think_ such a
system could work.
--
Dan Scott
Laurentian University
Received on Tue Aug 07 2007 - 15:03:16 EDT