lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Erick Erickson <erickerick...@gmail.com>
Subject Re: disable stemming on query parser.
Date Wed, 21 Dec 2011 13:11:48 GMT
Actually, 1M records isn't all that much for a Solr index, so I'd
simply test with the
copyfield alternative as it's much easier.

About compression: this simply compresses the *stored* data, which has
essentially no effect on index search speed, but will affect the size of
the file (*.fdt) that contains stored data. Here's a good reference:
http://lucene.apache.org/java/3_0_2/fileformats.html#file-names

The fields you copy *to* should probably not be stored (stored="false").

The idea (I thought there was a Solr patch for a new Filter, but I can't
find it) is to have something similar to the SynonymAnalyzer in Lucene
In Action that inserts the special token at, say, the end of the
term *as well as* sending the original term. Say you're indexing
"running". Your index process would put in "run" and "running#". Now
when searching, when you want exact match, you search for
the terms with the '#' at the end. Yes, it makes your index larger,
but whatever you do will make the index larger.

Best
Erick

On Wed, Dec 21, 2011 at 5:36 AM, meghana <meghana.ravani@amultek.com> wrote:
> Hi Dmitry ,
>
> If we add some unseen character sequence to array , doesn't it remove my
> stemming at all time? how we can manage stemmed and unstemmed words in the
> same field? i am a bit confused on this.
>
> also i tried with making compression on a field, which i use for copy field,
> what i read about compression on field , it should make your index size
> lower. and it lowers performance a bit while querying , but when i tried it
> on my local solr configuration (which have about 5000 records , and copy
> field size is more than 5000 char or may be much more).  it behave totally
> opposite of it. It increased my index file size and also performance does
> not decrease.  have any idea why it is behaved like this.
>
> like to make a note that this i tried with my local configuration of solr.
> in live solr , we have more than 10 lakh records , and copy field size is
> very big( about 5000 or much more char)
>
> Thanks in advance,
> Meghana
>
> --
> View this message in context: http://lucene.472066.n3.nabble.com/disable-stemming-on-query-parser-tp3591420p3603675.html
> Sent from the Solr - User mailing list archive at Nabble.com.

Mime
View raw message