[WEB4LIB] metasearch engines

Nick Arnett listbot at mccmedia.com
Sat Oct 24 10:00:44 EDT 1998


At 07:26 AM 10/23/98 -0700, Nicholas Tomaiuolo wrote:

>Could I elicit views and opinions from you addressing the
>appropriateness of metasearch engine versus single engine use?

It is well known that people generally interpret search results with the
fewest irrelevant documents as the most accurate, even though the results
may omit many relevant documents.  Combine this with the fact that
searching more indexes nearly always increases the number of irrelevant
documents more than it increases relevant ones, and the outcome is that
people will usually perceive a meta-search engine as less accurate.  And
they generally are, *if* you spend equal time reviewing results on a single
engine and a meta-search engine.  This is compounded by the fact that the
engines score relevancy differently, so that a combined results list can
only be as precise as the least accurate engine. Naturally, there are
classes of users for whom these generalizations are not true.  Anyone who
wishes to be exhaustive is an exception, such as patent attorneys.

I like InfoSeek Express, because it seems to have good relevancy ranking
and is designed to help reduce the time that it takes to search and review
results -- it'll pre-fetch results. It does the meta-search from the local
machine, instead of on a server, so it presumably is more scalable.  Of
interest is the fact that it preserves the advertising from the engines it
searches, which should keep the various competitors happy, or at least not
unhappy.

Nick


More information about the Web4lib mailing list