[WEB4LIB] Very strange AltaVista result

[Greg Notess]

First of all, AltaVista is not very good at counting. See my AltaVista
Inconsistencies page
<http://www.notess.com/search/features/av/inconsistent.shtml> on Search
Engine Showdown for more details. However, in this case, part of the answer
is due to the site clustering on AltaVista simple searches. Try the search
with Boolean operators in the advanced search and the numbers make more sense:
	diabetes and domain:no is 1936
	Norwegian language limit reduces it to 1497
	English language limit to 374

So what happens on the simple search? Look at record number 1 of the 95
which for me is <http://www.uib.no/isf/sats/quality/quality6.htm>. Then
click on "More pages from this site" and you will get another 136 pages at
www.uib.no which are not listed in the 95 because results have been
clustered by site. So 95 is more the number of sites for which results have
been clustered. See my AltaVista review
<http://www.notess.com/search/features/av/review.html> under Sorting for
more details.

However, this does not explain the increase in size when adding the
Norwegian language limit and the decrease for +diabetes and Norwegian
language limit. On both of these, the Advanced Search seems to work better.
In the meanwhile, it makes another section to add to my AltaVista
Inconsistencies page. As you can see from others listed there, some of the
inconsistencies last only for a short while but others are ongoing.

-- 
 Greg R. Notess   greg at notess.com
 406-994-6563 (w) 406-585-2287 (h)
    Search Engine Showdown <http://notess.com/search/>
    Internet columnist for Online and EContent
    Author of "Government Information on the Internet"
    Reference Librarian - Montana State University

>I have just been giving a seminar on Internett searching
>to several Norwegian librarians, and during that, I got some of
>the strangest search results yet while trying out AltaVista.
>
>1) I search for +diabetes. Result: 1,044,090 pages found.
>2) I search for +diabetes +domain:no to get results
>from Norwegian servers. Result:95 pages found.  Not much, but some.
>3) Then I limit further by keeping the search string, but limiting the
>language to Norwegian. Result:  1498 pages found. 
>??????
>4) Finally, I drop the +domain:no and limit only to +diabetes in 
>Norwegian language. Result: 92 pages found. 
>???????
>
>Can anyone explain what is going on? I looked at the results under
>3) and they all looked OK - in Norwegian and in domain:no. It seems
>2) and 4) are badly reduced.