OOV queries

[Dmitry Kan <so...@gmail.com>]

Hi,

Solr is good at caching: even if first "cold" query takes longer time, the
subsequent one is much quicker, given that it shares the fq's of the first
query.

This is more like an idea question:

what about the out-of-vocabulary (OOV) queries? According to my quick
measurements, they take same amount of time when repeated and return
nothing.

Most of such queries are usually user typos, so I could imagine
implementing a round-trip to the server with an auto-corrected query, if
the first query returned 0 hits. This does not solve the underlining slow 0
hits query though.

Would it make sense to implement an OOV cache to speed things up? Does one
exist already in Lucene / Solr?

-- 
Dmitry Kan
Blog: http://dmitrykan.blogspot.com
Twitter: http://twitter.com/dmitrykan

Re: OOV queries

[Dmitry Kan <so...@gmail.com>]

Mike,

I did only measurements by hand, i.e. no scientific figures yet. Basically
the same oov query takes about same amount of time when repeated. Which is
surprising. Ok, to add to the mix: from a list of fqs that are on that
query, one is marked with cache=false and a cost > 100.

Dmitry




On Thu, Jun 5, 2014 at 4:17 PM, Michael Sokolov <
msokolov@safaribooksonline.com> wrote:

> It seems as if 0-hit queries should be pretty fast since they can
> terminate very early?  Are you seeing a big difference between first-time
> and subsequent (cached) no-match queries?
>
> -Mike
>
>
>
> On 6/5/2014 8:47 AM, Dmitry Kan wrote:
>
>> Hi,
>>
>> Solr is good at caching: even if first "cold" query takes longer time, the
>> subsequent one is much quicker, given that it shares the fq's of the first
>> query.
>>
>> This is more like an idea question:
>>
>> what about the out-of-vocabulary (OOV) queries? According to my quick
>> measurements, they take same amount of time when repeated and return
>> nothing.
>>
>> Most of such queries are usually user typos, so I could imagine
>> implementing a round-trip to the server with an auto-corrected query, if
>> the first query returned 0 hits. This does not solve the underlining slow
>> 0
>> hits query though.
>>
>> Would it make sense to implement an OOV cache to speed things up? Does one
>> exist already in Lucene / Solr?
>>
>>
>


-- 
Dmitry Kan
Blog: http://dmitrykan.blogspot.com
Twitter: http://twitter.com/dmitrykan

Re: OOV queries

[Michael Sokolov <ms...@safaribooksonline.com>]

It seems as if 0-hit queries should be pretty fast since they can 
terminate very early?  Are you seeing a big difference between 
first-time and subsequent (cached) no-match queries?

-Mike


On 6/5/2014 8:47 AM, Dmitry Kan wrote:
> Hi,
>
> Solr is good at caching: even if first "cold" query takes longer time, the
> subsequent one is much quicker, given that it shares the fq's of the first
> query.
>
> This is more like an idea question:
>
> what about the out-of-vocabulary (OOV) queries? According to my quick
> measurements, they take same amount of time when repeated and return
> nothing.
>
> Most of such queries are usually user typos, so I could imagine
> implementing a round-trip to the server with an auto-corrected query, if
> the first query returned 0 hits. This does not solve the underlining slow 0
> hits query though.
>
> Would it make sense to implement an OOV cache to speed things up? Does one
> exist already in Lucene / Solr?
>