Facet counts for first N hits

[Noah Torp-Smith <no...@dbc.dk.INVALID>]

If I have a solr with millions of documents, and a query results in maybe 100000 hits, is there any way to get facet information for only the first 1000 hits (for example) for that query?

I googled and found a question from 11 years ago (https://solr-user.lucene.apache.narkive.com/AtnCbPV0/getting-facet-counts-for-10-000-most-relevant-hits#selection:3.655.119), but the links there are not working any longer. I guess I could fetch the 1000 documents and extract the documents on the client side, but I'd avoid that if solr could do it for me.

Thanks,

/Noah


--

Noah Torp-Smith (nots@dbc.dk)

Re: Facet counts for first N hits

[Joel Bernstein <jo...@gmail.com>]

Streaming Expressions can do this pretty easily. Checkout the rollup
expression:

https://solr.apache.org/guide/8_6/stream-decorator-reference.html#rollup

The example shows a rollup over the export handler results. to rollup over
the top N just remove the qt=/export and it will use the /select handler.


Joel Bernstein
http://joelsolr.blogspot.com/


On Fri, Jun 3, 2022 at 8:24 AM Dave <ha...@gmail.com> wrote:

> Yeah, my first thought would be to have the first query with no facets and
> a fl of just the id, limited to 1000, it’s a lot faster than you think if
> you only return the id and no facets. Then do a secondary search for just
> those ids and the facets added to that query using terms component as you
> may run out of OR clauses.
>
> https://solr.apache.org/guide/6_6/the-terms-component.html
>
>
>
> > On Jun 3, 2022, at 7:52 AM, Noah Torp-Smith <no...@dbc.dk.invalid> wrote:
> >
> > If I have a solr with millions of documents, and a query results in
> maybe 100000 hits, is there any way to get facet information for only the
> first 1000 hits (for example) for that query?
> >
> > I googled and found a question from 11 years ago (
> https://solr-user.lucene.apache.narkive.com/AtnCbPV0/getting-facet-counts-for-10-000-most-relevant-hits#selection:3.655.119),
> but the links there are not working any longer. I guess I could fetch the
> 1000 documents and extract the documents on the client side, but I'd avoid
> that if solr could do it for me.
> >
> > Thanks,
> >
> > /Noah
> >
> >
> > --
> >
> > Noah Torp-Smith (nots@dbc.dk)
>

Re: Facet counts for first N hits

[Dave <ha...@gmail.com>]

Yeah, my first thought would be to have the first query with no facets and a fl of just the id, limited to 1000, it’s a lot faster than you think if you only return the id and no facets. Then do a secondary search for just those ids and the facets added to that query using terms component as you may run out of OR clauses. 

https://solr.apache.org/guide/6_6/the-terms-component.html



> On Jun 3, 2022, at 7:52 AM, Noah Torp-Smith <no...@dbc.dk.invalid> wrote:
> 
> If I have a solr with millions of documents, and a query results in maybe 100000 hits, is there any way to get facet information for only the first 1000 hits (for example) for that query?
> 
> I googled and found a question from 11 years ago (https://solr-user.lucene.apache.narkive.com/AtnCbPV0/getting-facet-counts-for-10-000-most-relevant-hits#selection:3.655.119), but the links there are not working any longer. I guess I could fetch the 1000 documents and extract the documents on the client side, but I'd avoid that if solr could do it for me.
> 
> Thanks,
> 
> /Noah
> 
> 
> --
> 
> Noah Torp-Smith (nots@dbc.dk)