Ferret talk - Jan 2009 - Wildcard trouble

Hi-- I just ran into an odd situation.  If I do a search including the term:
         c*  -  I get 4 hits
         ca* - I get the same 4 documents
         co* - I get one new document, not found by c*

Does anyone know what might be going on, or have suggestions for debugging?

Thanks,
     --Paul

-- 
Paul Lynch
Aquilent, Inc.
National Library of Medicine (Contractor)

Hi Paul,

On 07.01.2009, at 01:17, Paul Lynch wrote:
> Hi-- I just ran into an odd situation.  If I do a search including  
> the term:
>         c*  -  I get 4 hits
>         ca* - I get the same 4 documents
>         co* - I get one new document, not found by c*
>
> Does anyone know what might be going on, or have suggestions for  
> debugging?
How does your full query look like? Ferret has a built in default  
limit of 512 for the number of terms wildcard queries (and other  
MultiTermQueries) can get expanded to. Any more terms matching your  
criteria will be dropped then, keeping the most relevant 512 terms.
You can override this value by specifying a max_terms value when  
constructing the query via the API:

query = WildcardQuery.new(:field, "c*",
                           :max_terms => 1024)

you might also try monkey patching the  
Ferret::Search::MultiTermQuery::default_max_terms method to return  
your custom limit so you dont need to use the query API to construct  
your queries (i.e. with aaf which doesn''t reliably work with query  
objects due to the DRb stuff involved).

It *might* also be a bug in Ferret - if the above doesn''t help, can  
you reproduce this with a simple test case?

cheers,
Jens

--
Jens Kr?mer
Finkenlust 14, 06449 Aschersleben, Germany
VAT Id DE251962952
http://www.jkraemer.net/ - Blog
http://www.omdb.org/     - The new free film database

-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 194 bytes
Desc: This is a digitally signed message part
URL:
<http://rubyforge.org/pipermail/ferret-talk/attachments/20090210/19ca694c/attachment.bin>

Jens,
  I think you are right.  There are 932 terms matching c* in my data table.
(The rest of my query is simple-- just onr or two other terms without
wildcards).  I tried setting the value of default_max_terms, but it did not
seem to have any effect.  (I think I was setting it correctly, because I
tried assigning a negative number and it immediately complained.)  However,
now that I know what that is doing, I''m not sure I want to increase the
value.
  Anyway, thank you very much for your help in sorting this out.
         --Paul

On Tue, Feb 10, 2009 at 4:15 AM, Jens Kraemer <jk at jkraemer.net> wrote:
> Hi Paul,
>
> On 07.01.2009, at 01:17, Paul Lynch wrote:
>
>  Hi-- I just ran into an odd situation.  If I do a search including the
>> term:
>>        c*  -  I get 4 hits
>>        ca* - I get the same 4 documents
>>        co* - I get one new document, not found by c*
>>
>> Does anyone know what might be going on, or have suggestions for
>> debugging?
>>
>
> How does your full query look like? Ferret has a built in default limit of
> 512 for the number of terms wildcard queries (and other MultiTermQueries)
> can get expanded to. Any more terms matching your criteria will be dropped
> then, keeping the most relevant 512 terms.
> You can override this value by specifying a max_terms value when
> constructing the query via the API:
>
> query = WildcardQuery.new(:field, "c*",
>                          :max_terms => 1024)
>
> you might also try monkey patching the
> Ferret::Search::MultiTermQuery::default_max_terms method to return your
> custom limit so you dont need to use the query API to construct your
queries
> (i.e. with aaf which doesn''t reliably work with query objects due
to the DRb
> stuff involved).
>
> It *might* also be a bug in Ferret - if the above doesn''t help,
can you
> reproduce this with a simple test case?
>
> cheers,
> Jens
>
> --
> Jens Kr?mer
> Finkenlust 14, 06449 Aschersleben, Germany
> VAT Id DE251962952
> http://www.jkraemer.net/ - Blog
> http://www.omdb.org/     - The new free film database
>
>
> _______________________________________________
> Ferret-talk mailing list
> Ferret-talk at rubyforge.org
> http://rubyforge.org/mailman/listinfo/ferret-talk
>


-- 
Paul Lynch
Aquilent, Inc.
National Library of Medicine (Contractor)
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://rubyforge.org/pipermail/ferret-talk/attachments/20090302/d7d4be02/attachment.html>