Caching filters inside bq clauses when using DisMax

[Kaktu Chakarabati <ji...@gmail.com>]

Quick Question - are the filter queries used to create the document mask
inside bq clauses also cached using the filterCache?
this seems like an important optimization for sites that use bq heavily with
a small set of static boost selector filters..

thanks,
Chak

Re: Caching filters inside bq clauses when using DisMax

[Jayson Minard <ja...@gmail.com>]

This seems wrong still.  It *does* make sense to cache the bq as a filter
since it is re-used every time you issue a new query through dismax.  The
query changes with every q parameter change so that query re-executes but
why do force each bq to execute again?  Keep it, and apply its boost where
it intersects.

--
View this message in context: http://lucene.472066.n3.nabble.com/Caching-filters-inside-bq-clauses-when-using-DisMax-tp2844821p3652081.html
Sent from the Lucene - Java Developer mailing list archive at Nabble.com.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org

Re: Caching filters inside bq clauses when using DisMax

[Chris Hostetter <ho...@fucit.org>]

: got you. I thought perhaps the bq is first computed as a filter, e.g a
: bitmask over the entire corpus, and then intersected with results, in which

since the "bq" only affects teh scores of docs that alreayd match 9and 
doesn't alter the result set in any way) there wouldn't really be much 
point in caching a docset for it ... that doc set would never be 
intersected or unioned with any other doc set.


-Hoss

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org

Re: Caching filters inside bq clauses when using DisMax

[Kaktu Chakarabati <ji...@gmail.com>]

Hoss,
got you. I thought perhaps the bq is first computed as a filter, e.g a
bitmask over the entire corpus, and then intersected with results, in which
case it would've been beneficial to cache that bitmask,
I guess thats not how it works..


On Wed, Apr 20, 2011 at 3:25 PM, Chris Hostetter
<ho...@fucit.org>wrote:

>
> : Quick Question - are the filter queries used to create the document mask
> : inside bq clauses also cached using the filterCache?
> : this seems like an important optimization for sites that use bq heavily
> with
> : a small set of static boost selector filters..
>
> I really don't understand what you mean by "document mask"
>
> when bq is used, the query defined is added as an optional clause to the
> main query -- nothing can be cached independent of the main query, becaue
> the bq has no independent results.  it's just another clause of the mian
> query (factored into the final DocList)
>
> -Hoss
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
> For additional commands, e-mail: dev-help@lucene.apache.org
>
>

Re: Caching filters inside bq clauses when using DisMax

[Chris Hostetter <ho...@fucit.org>]

: Quick Question - are the filter queries used to create the document mask
: inside bq clauses also cached using the filterCache?
: this seems like an important optimization for sites that use bq heavily with
: a small set of static boost selector filters..

I really don't understand what you mean by "document mask"

when bq is used, the query defined is added as an optional clause to the 
main query -- nothing can be cached independent of the main query, becaue 
the bq has no independent results.  it's just another clause of the mian 
query (factored into the final DocList)

-Hoss

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org