[WEB4LIB] Re: metasearch engines

[Cramer, Steve]

After several years of encouraging netizens not to spend much time
with metasearch engines, I have recently begun to appreciate them
(the meta-engines, that is!)

Several meta-engines (Dogpile, Inference Find, Metafind) now offer
interesting and advanced search options, including:

--specifying the number of seconds you want the meta-engine to
search (so meta-searching can be quite speedy);
--customizing the engines and directories searched, and the *order*
of engines and directories searched;
--sorting (alphabetically by title or by keyword, or by domain),
and clustering related hits (like Northern Light's folders);
--and removing duplicate hits.

Obviously such features don't overcome the basic limitations of
meta-searching described by Ward and John. But given how unique the
HotBot, AltaVista, and other engines' databases are (see Greg
Notess' "Measuring the Size of Internet Databases" from Online,
Oct/Nov 1997), metasearch engines can be valuable tools for
searching obscure topics or hard-to-find keywords.

Also, as libraries and vendors begin to offer integrated searching
of multiple databases via z39.50, we may be able to learn something
from the meta search engines -- another reason not to dismiss
Dogpile and friends.

Steve Cramer
Reference Librarian
Perkins Library, Duke University
Durham, NC USA
scramer at mail.lib.duke.edu