Web Search Engines "Made Simple"

[Bob Duncan]

I have to wonder how good any search engine guide can be, print or
otherwise, considering many of the engines don't always work the way they
claim. I see many explanations of "how engine xyz works," but little about
whether the engine consistently fulfills its promise, i.e., is it *really*
working the way it should? (Perhaps this topic is covered in the new book?)
If a search mechanism consistently "misfires," then perhaps workarounds can
be applied, but when it behaves erratically, then what?

Two quick examples: 
A search on Infoseek using the query
	+date +rape [both terms required]
produced 134 hits, while the query
	"date rape" [date rape as a phrase]
produced 1375 hits. Unless I was asleep during IR101, there cannot be more
occurrences of a phrase than of the component parts of that phrase.
(Proximity is always more specific than a Boolean AND, which is what using
a + sign before both terms should create.) Also, using a + in front of the
double-quoted phrase produced fewer results than not using the +. (How can
a single phrase query not already be required?)  I'd be inclined to wonder
if different indexes are searched depending on the query formation, but
Infoseek doesn't always behave this way. (Only the one time I'm doing a
demo without performing my searches ahead of time...)

A search on HotBot using the query
	date rape
produced the same set of 142,000+ hits regardless of whether I told it to
look for "any of the words," "all the words," or "the exact phrase." When I
added a third term (drugs) the engine performed as *expected*, but one has
to wonder if it performed as it *should* have, and why it dropped the ball
on the original queries.

I've notified Infoseek and HotBot several times, but only get the automated
thank you.

I'm sure the new pub by Maze, Moxley, and Smith is useful, but let's hope
the changes to search engines will be more than just "cosmetic."

Getting a little weary of telling students, "always read the
tips/options/help page...and then assume it won't always work that way."

Bob Duncan


  ~'~'~'~'~'~'~'~'~'~'~'~'~'~'~'~'~
  Robert E. Duncan
  Reference/Instruction Librarian
  David Bishop Skillman Library
  Lafayette College
  Easton, PA  18042
  duncanr at lafayette.edu
  http://www.lafayette.edu/faculty/duncanr/