Web search tools - another study

[Ian Winship, Univ Northumbria Info Services]

This is relevant to recent postings on cataloguing the internet.


==========================================================================
Ian Winship                             
Information Services Dept.              |    e-mail: ian.winship at unn.ac.uk
University of Northumbria at Newcastle  |    phone:  0191 227 4132
City Campus Library                     |    fax:    0191 227 4563
Newcastle upon Tyne                     |
NE1 8ST                                 |
UK                                      |
===========================================================================



From: LEIGHTON at VAX2.Winona.MSUS.EDU
Subject: World Wide Web indexes: a study
To: Multiple recipients of list PACS-L <PACS-L at UHUPVM1.UH.EDU>

----------------------------Original message----------------------------
Pardon the cross posting on PACS-L and GOVDOC-L

I would like to announce the results of a study that I did in May and
June of 1995. I compared the precision of four of the larger free
World Wide Web indexing services: Infoseek (the service available
for free), Lycos, Webcrawler and WWWWorm. For precision among
the top ten hits (sites) returned, I used a two-way ANOVA to judge
significance.

I did this project for a graduate course in Computer Science.
Although I did not get an outstanding grade on the project, I
feel that there are real results and that they are important
enough to post on the Internet.

THE GIST OF THE RESULTS

Cutting to the chase, my basic finding was that Lycos and the free part
of Infoseek have about the same precision, with Lycos just a nose ahead.
I was surprised and pleased with the precision of both. Webcrawler gave
me surprisingly bad precision, surprising because it is currently supported
by America Online.

WWWWorm turned out to be a bit of a downer. My initial searching indicated
that it was big league, but more extensive efforts proved that my impression
was wrong. There are many smaller indexing services with it in this minor
league: Jumpstart, Nikos, etc. I suspect that they do not enjoy enough
institutional support for hardware to develop rich databases. These
services are fairly abysmal, often retrieving nothing for a subject
unrelated to computer science. WWWWorm was good enough that it
usually retrieved at least one or two hits, which for my queries
had high precision.

As far as response time between Infoseek and Lycos, Infoseek was
better, especially during the workday when traffic is up. However,
Infoseek would block the free users if the paid traffic was too high,
which it occasionally was. Also, at that time, the free Infoseek only
returned the top ten hits, whereas now, it will provide many more.

I have the complete report available on the Web at:

http://www.winona.msus.edu/services-f/library-f/webind.htm

Please comment if you have problems with the methods or conclusions. I
will try to make constructive criticism available on the Web too. You
may email me at:

Leighton at vax2.winona.msus.edu


H. Vernon Leighton
Government Documents Librarian
Winona State University
Winona MN 55987
(507) 457-5148
Leighton at vax2.winona.msus.edu