Altavista: Required Terms

Gwen Galton gwen.galton at sheridanc.on.ca
Mon Mar 24 20:49:44 EST 1997


> Example:  The search "+nesbeitt" retrieves 468 documents.  (I realize these
> numbers aren't exact.)
> 
> The search "+nesbeitt +library" retrieves approximately "3000 documents."
> 
> The search "+nesbeitt +library +bridgewater" retrieves approximately "30000
> documents."   Etc.  You'd think that forcing more required terms would narrow 
> the search, not widen it.
> 
It could be that the relevancy criteria of Alta Vista which eventually 
defaults to OR over-rides the purity of Boolean Logic as we are used to 
seeing it.  I believe Alta Vista's relevancy criteria 
(in addition to words appearing in the head, URL, heading and first paragraph) 
call for:

First: words occur beside one another in the document (PHRASE OR 
^^^^^                                                  PROXIMITY)
Next: both words appear in document  (AND)
^^^^
Then: either word appears in document.  (OR)
^^^^

It seems that as you add terms, you get results based on the above 
criteria that eventually default to OR.

> The Advanced Query page, in which one includes AND in between the search
> terms, performs the search correctly.

The advanced query algorithm may not work in the same way. I must look 
into this.  This is a topic quite close to my heart.  Would be interested 
to hear other comments.  
Gwen...
------------------------------------------------------------------------
Gwen Galton		  		       | gwen.galton at sheridanc.on.ca
School of Computing and Information Management | Tel: 905-845-9430 x2485 
Sheridan College, 1430 Trafalgar Road,         | Fax: 905-815-4035	
Oakville, Ontario, Canada L6H 2L1	       | 
		Visit the School's Home Page at 
	http://www.sheridanc.on.ca/academic/computing/scim.htm
------------------------------------------------------------------------------
Gwen Galton Consultants: galtong at ican.net/ (905)823-3614/ Fax: (905)823-6202


More information about the Web4lib mailing list