lycos relevancy scoring - question

[vaughnk1 at westatpo.westat.com]

Barb

I've run into such inconsistancies before at Lycos.  I wrote to 
webmaster at lycos.com.  They answered pretty quickly.  Here is the reply I got 
which contains enough of my message to make sense of it.  It _may_ apply to your
term "music" but at least it will hint at some of the weaknesses in search 
engines yet to be ironed out.
Forgive the formatting of the message.  Too many mailers, too little time.


~~~~~~~~~~~FROM LYCOS WEBMASTER~~~~~~~~~~~~~~~~~~~~

At 02:17 PM 4/2/96 -0800, you wrote: >1.  SEARCH:  "westat research"
>retrieves only 1 irrelevant document.
>2.  SURF TO:  www.westat.com and you will find "westat" and "research" both 
used in 
>plain text near the top of the page (twice).
>3.  Now search Lycos:  "westat."  and www.westat.com comes up along with child 
pages.  
>URL #1 is www.westat.com mentioned above.  
>  Westat is fairly unconcerned about publicity but I am a librarian and Lycos 
has been 
>my #1 tool.  How is Lycos' engine doing this? 

I was pretty curious about this myself, so I had a long talk with our catalog 
developer about it.

His best guess is that, due to the size of the catalog, that "common" words can 
sometimes be underrepresented.  So even though the word "research" does appear 
in your file, it might not be searchable because so many files contain the word 
"research".  But few files contain the word "westat", so a search just on 
"westat" should always lead you
to your page.

This is a problem we're working on.  It might have something to do
with size limits buried deep in the search algorithm.  We apologize for any 
inconvenience.


**  Laurie D. T. Mann    ****    Lycos Webmaster Triage  ** Links to Lycos 
Services - just copy these links to your pages: <a 
href="http://www.lycos.com">Lycos catalog</a><br>
<a href="http://a2z.lycos.com">a2z directory</a><br>
<a href="http://www.pointcom.com">Point reviews</a><br>

~~~~~~~~~~~~~~END~~~~~~~~~~
______________________________ Reply Separator _________________________________
Subject: lycos relevancy scoring - question
Author:  BARB at DAYTON.LIB.OH.US at internet-e-mail
Date:    5/14/96 4:57 PM



 ... I did a search for old time music in Lycos in its default 
form - (match any term, display 10, loose match, standard 
output).  The first four have in their scoring 3 of 3 found.  
After that, all the hits have 2 of 3 found.  However, in hits 
21-30, there are some hits where all three words are noticeably 
present, even as far as being adjacent in the title.  

Can anyone help me with why this occurs?  I was wondering if a clue 
was in "Words Matched on Page:" that shows up in the detailed 
display.

I have looked in "Web Search Strategies" (book) and in the web4lib 
archive and didn't find an answer.  Thanks in advance for any 
information about these searhc results.

Barb

Barbara Kuhns
Dayton & Montgomery County Public Library 
Dayton, OH
barb at dayton.lib.oh.us

   * Kenneth Vaughn ~~~~~~~~~~  LIBRARIAN K  ~~~~~~~~~~ Tech Srvcs Libr *
   *       Westat, Inc. 1650 Research Blvd. Rockville MD 20850          *
   * vaughnk1 at westat.com  =  voice: 301-294-2881  =  fax: 301-294-2034  *