Internal Revenus Service/AltaVista

Nick Arnett narnett at verity.com
Tue Apr 22 16:40:05 EDT 1997


At 12:54 PM 4/22/97 -0700, Laura B. Cohen wrote:

>I have to disagree here. Sometimes the *better* and more intelligent
>software cannot recognize phrases, for example. Just look at the
>case of DIALOG, which is an extremely powerful and accurate search
>engine (which of course does not relevancy rate its results).

What engine is Dialog using?  As far as I know, it is not very accurate in
comparison with state-of-the-art search.  There aren't any high-accuracy
engines that don't do some sort of relevancy ranking.  I believe Dialog is
still using something written years ago.  (If they're using Verity, just
give me a moment to fetch my hara-kiri sword.)

When I said "recognize phrases," I should have been more clear.  I meant
that linguistically advanced engines will recognize noun phrases and proper
names -- "Internal Revenue Service" is both.  This is quite different from
the typical default behavior of searching on the "bag of words" (the query
words, minus stop words, stemmed, in any order or proximity) v. searching
for the phrase (which has the same mthods but also ranks documents higher if
they contain the words in a similar phrase and order).  I suspect that
Silver Platter's default is to treat words as a phrase.  This is fine if it
is a phrase, but worse than useless if it is not -- if it were "taxes
government IRS," which are just evidence words whose order and proximity
aren't especially relevant.

>I don't think there's anything wrong with requiring people to take 
>a moment to learn how to use something. especially given today's 
>search technologies.

I agree that people should and expect that probably will learn a bit more.
With so many engines on the market and no agreement on query operators,it's
a difficult case to make right now.  Which one(s) should people learn?

Nick

---------------------------------------
Verity Inc. -- Connecting People with Information

Product Manager, Categorization and Visualization
408-542-2164; fax 408-541-1600; home office 408-733-7613
http://www.verity.com/

Verity Inc.
894 Ross Drive
Sunnyvale, CA 94089



More information about the Web4lib mailing list