TermsComponent results don't change after documents removed from index

View: New views
3 Messages — Rating Filter:   Alert me  

TermsComponent results don't change after documents removed from index

by Bill Au :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Should the results of the TermsComponent change after documents have been
removed from the index?  I am thinking about using the prefix of
TermsComponent to implement auto-suggest.  But I noticed that the prefix
counts in TermsComponent don't change after documents have been deleted.
The deletes are done with the standard update handler using a
delete-by-query.  Since the TermsComponent is showing the number of
documents matching the terms, the number should be decreasing when documents
are deleted.

I can reproduce this using the sample in the tutorial and the TermsComponent
prefix query in the Wiki:
http://wiki.apache.org/solr/TermsComponent

The output of the TermsComponent prefix doesn't change even after I removed
all the documents:

java -Ddata=args -jar post.jar "<delete><query>id:*</query></delete>"

What am I doing wrong?

Bill

Re: TermsComponent results don't change after documents removed from index

by Koji Sekiguchi-2 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Bill Au wrote:

> Should the results of the TermsComponent change after documents have been
> removed from the index?  I am thinking about using the prefix of
> TermsComponent to implement auto-suggest.  But I noticed that the prefix
> counts in TermsComponent don't change after documents have been deleted.
> The deletes are done with the standard update handler using a
> delete-by-query.  Since the TermsComponent is showing the number of
> documents matching the terms, the number should be decreasing when documents
> are deleted.
>
> I can reproduce this using the sample in the tutorial and the TermsComponent
> prefix query in the Wiki:
> http://wiki.apache.org/solr/TermsComponent
>
> The output of the TermsComponent prefix doesn't change even after I removed
> all the documents:
>
> java -Ddata=args -jar post.jar "<delete><query>id:*</query></delete>"
>
> What am I doing wrong?
>
> Bill
>
>  
This is a feature of Lucene... docFreq is not changed until segments
containing
deletions are merged. You can do optimize to correct docFreq.

Koji

--
http://www.rondhuit.com/en/


Re: TermsComponent results don't change after documents removed from index

by Bill Au :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Thanks for pointing that out.  The TermsComponent prefix query is running
much faster than the facet prefix query.  I guess there is yet another
reason to optimize the index.

Bill

On Tue, Nov 3, 2009 at 5:09 PM, Koji Sekiguchi <koji@...> wrote:

> Bill Au wrote:
>
>> Should the results of the TermsComponent change after documents have been
>> removed from the index?  I am thinking about using the prefix of
>> TermsComponent to implement auto-suggest.  But I noticed that the prefix
>> counts in TermsComponent don't change after documents have been deleted.
>> The deletes are done with the standard update handler using a
>> delete-by-query.  Since the TermsComponent is showing the number of
>> documents matching the terms, the number should be decreasing when
>> documents
>> are deleted.
>>
>> I can reproduce this using the sample in the tutorial and the
>> TermsComponent
>> prefix query in the Wiki:
>> http://wiki.apache.org/solr/TermsComponent
>>
>> The output of the TermsComponent prefix doesn't change even after I
>> removed
>> all the documents:
>>
>> java -Ddata=args -jar post.jar "<delete><query>id:*</query></delete>"
>>
>> What am I doing wrong?
>>
>> Bill
>>
>>
>>
> This is a feature of Lucene... docFreq is not changed until segments
> containing
> deletions are merged. You can do optimize to correct docFreq.
>
> Koji
>
> --
> http://www.rondhuit.com/en/
>
>