Looking through the code and some example Suggesters, it seems that
theoretically, one can write a GeoSuggester and provide that as the Lookup
implementation (lookupimpl) that would factor in the geo score or extend
the SolrSuggestor to support spatial extensions in the same spirit as
"Filters" are supported today.

Sameer.

On Mon, Nov 9, 2015 at 11:47 AM, William Bell <billnb...@gmail.com> wrote:

> Yeah we have that working today. But the issue is we want to use
> http://lucidworks.com/blog/solr-suggester/
>
> And you cannot do a boost with that right?
>
>
>
> On Mon, Nov 9, 2015 at 12:41 PM, Sameer Maggon <sam...@measuredsearch.com>
> wrote:
>
> > Have you looked at the Spatial extensions for Solr? If you are indexing
> > Lat/Lon along with your documents, you can compute the distance from the
> > origin & use that distance as one of the boost factors to affect the
> score.
> > Typically, use cases around that combine the geo score with other factors
> > as a pure sort by geo score might not give you the relevant results.
> >
> > e.g. typing to search for "sushi restaurants" near Santa Monica, CA - you
> > might not want "thai restaurants" that are closest to you. (Local Search
> > use case)
> >
> > https://cwiki.apache.org/confluence/display/solr/Spatial+Search
> >
> > Thanks,
> > --
> > *Sameer Maggon*
> > www.measuredsearch.com <http://measuredsearch.com/>
> > Fully Managed Solr-as-a-Service | Solr Consulting | Solr Support
> >
> >
> >
> > On Mon, Nov 9, 2015 at 11:18 AM, William Bell <billnb...@gmail.com>
> wrote:
> >
> > > http://lucidworks.com/blog/solr-suggester/
> > >
> > >
> > > Wondering if anyone has uses these new techniques with a boost on
> > > geodist() inverted? So the rows that get returned that are closest
> > > need to come back first.
> > >
> > >
> > > We are still using Edge Grams since we have not figured out how to
> > > boost the results on geo spatial.
> > >
> > >
> > > Anyone have thoughts?
> > >
> > >
> > >
> > >
> > > --
> > > Bill Bell
> > > billnb...@gmail.com
> > > cell 720-256-8076
> > >
> >
>

Reply via email to