lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Erick Erickson <erickerick...@gmail.com>
Subject Re: Storing A Taxonomy In A Separate Core For Query Expansion
Date Thu, 08 Aug 2013 20:43:18 GMT
No reason why not. You don't even have to keep it in a separate core
if you have some field in your docs that had values
"items" or "taxonomy" you could always use the appropriate fq
when querying...

You could also to this from the app, query the taxonomy core and
do the substitutions in the app layer before sending to your items core.

Assuming the two cores are in the same JVM, it should be easy to query
the second core from a search component or query parser though.

Best
Erick


On Thu, Aug 8, 2013 at 7:52 AM, Michael Delaney <mickdelaney@gmail.com>wrote:

> Hi Guys,
>
> Our application contains two data sets:
>
> - Items
> - Item taxonomy
>
> What i'm trying to do is allow our admin team to maintain a taxonomy which
> allows for items to be found easier.
>
> For example:
>
> if someone searched for 'dog', and the taxonomy contained 'Spaniel' as
> a  narrower form of dog,
> then the search of the items core would return documents with Spaniel in
> the results, but lower in score.
>
> I've read this:
> http://nolanlawson.com/2012/10/31/better-synonym-handling-in-solr/
>
> - Here the author outlines how to handle this in a query parser, but he's
> focused on synonyms.
>
> - I'm trying to make it more general purpose, for example I mentioned
> narrower, there'll also be 'same as' type relations too.
>
> Anyways in the article above he uses the synonym file, what I'm trying to
> understand is:
>
> - Can I use another Core to store my taxonomy
>    - In a qparser or somewhere in the pipeline, examine my query
>    - Expand the terms based on the taxonomy (search it & find any related
> terms)
>    - Then use the expanded query to return the results
>
>
> I've also read this article here:
> http://java.dzone.com/articles/connecting-redis-solr-boosting
>
> - This explains how to use a custom function query connecting to redis to
> affect the scoring of a document.
>
> - It seems straight forward etc, but could the same result be achieved by
> keeping
> the referenced data not in redis but in another SOLR core???
>
>
> Any help would be greatly appreciated.
>
> Regards
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message