You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by Qingdi <qi...@nextbio.com> on 2009/01/08 20:29:12 UTC

2 questions about solr spellcheck

Hi,

I use solr 1.3 and I have two questions about spellcheck.

1) if my index docs are like:
<doc>
<str name="NAME">university1</str>
<str name="TYPE">UNIVERSITY</str>
</doc>
<doc>
<str name="NAME">street1, city1</str>
<str name="TYPE">LOCATION</str>
</doc>
is it possible to build the spell check dictionary using field "NAME" but
with filter "TYPE"="UNIVERSITY"?
That is, I only want to include the university name in the dictionary. What
is the proper way to implement this?

2) my current data index size is about 11G, and the spelling dictionary
index size is about 6 G. After adding the spell check component, will the
spell checking have any impact on the runtime query performance and memory
usage? Should I increase the memory allocation for the solr server?

Thanks for your help.

Qingdi
-- 
View this message in context: http://www.nabble.com/2-questions-about-solr-spellcheck-tp21359183p21359183.html
Sent from the Solr - User mailing list archive at Nabble.com.


Re: 2 questions about solr spellcheck

Posted by Shalin Shekhar Mangar <sh...@gmail.com>.
On Fri, Jan 9, 2009 at 12:59 AM, Qingdi <qi...@nextbio.com> wrote:

>
> Hi,
>
> I use solr 1.3 and I have two questions about spellcheck.
>
> 1) if my index docs are like:
> <doc>
> <str name="NAME">university1</str>
> <str name="TYPE">UNIVERSITY</str>
> </doc>
> <doc>
> <str name="NAME">street1, city1</str>
> <str name="TYPE">LOCATION</str>
> </doc>
> is it possible to build the spell check dictionary using field "NAME" but
> with filter "TYPE"="UNIVERSITY"?
> That is, I only want to include the university name in the dictionary. What
> is the proper way to implement this?
>

It is not possible out of the box. However, there are a couple of ways to do
this.

1. You can create a copy field for 'NAME' (say 'NAME_SPELL') which has a
value only if "TYPE"="UNIVERSITY" for the document.
2. You can create your own implementation of the IndexBasedSpellChecker and
HighFrequencyDictionary which applies a filter query on "TYPE" and then uses
the terms to create the dictionary.

Option #1 would be probably be the easiest if you care only about
"TYPE"="UNIVERSITY".


> 2) my current data index size is about 11G, and the spelling dictionary
> index size is about 6 G. After adding the spell check component, will the
> spell checking have any impact on the runtime query performance and memory
> usage? Should I increase the memory allocation for the solr server?
>

I think the spelling index will have some impact. But the magnitude of the
impact and the memory needed depends on a number of factors such as type of
queries, query rate etc.


>
> Thanks for your help.
>
> Qingdi
> --
> View this message in context:
> http://www.nabble.com/2-questions-about-solr-spellcheck-tp21359183p21359183.html
> Sent from the Solr - User mailing list archive at Nabble.com.
>
>


-- 
Regards,
Shalin Shekhar Mangar.