[Web4lib] Creating a Google-based site search capability

[K.G. Schneider]

> I've used that method several times for sites I've administered. It's
> quick, easy and free as in beer. One possible drawback is that users
> are sent to Google for the search results instead of viewing results
> within the context of the rest of your site, which may confuse some
> users. Another option there is to use the Google API to display
> search results on your site directly; although, this method is a bit
> more complex to setup.

It's certainly better than no search, and also better than some search tools
you could use. Google doesn't seem to mind (nor could they do much). 

In addition to the problems above, you may have search needs on your site
that are not well met by Google, which is designed for full-text search and
follows the search-uber-alles paradigm. The Google mini-appliance could at
least allow you to load synonyms so those flat files on your website could
be found by the widest terms available--e.g. check-out versus "circulation."
It's supposedly quite easy to implement. See its features here:

http://www.google.com/enterprise/mini/administrator_features.html

But it's not the only fish in the sea. It sounds as if site search for your
website is something to budget and plan for. There are several vendor
products (Thunderstone, Verity, and Google) and several other non-commercial
products of varying complexity and/or limitations. The vendor products all
seem to offer easy administration and search analyses--things useful for
justifying the site itself. Just a thought. 

Karen G. Schneider
kgs at bluehighways.com