Superset Similarity?

[Otis Gospodnetic <ot...@gmail.com>]

Hi,

Somebody asked about BM25 on solr-user and comparison to other relevancy
scoring methods.  Similarity has to be set at index time and should not be
changed... but is there any one Similarity that happens to be a superset of
them all, so that one can use that Similarity during indexing, and then
switch to other Similarities to query time?  I doubt it, but....maybe, just
maybe... :)

Thanks,
Otis
--
Search Analytics - http://sematext.com/search-analytics/index.html
Performance Monitoring - http://sematext.com/spm/index.html

Re: Superset Similarity?

[Robert Muir <rc...@gmail.com>]

On Fri, Nov 16, 2012 at 5:18 PM, Tom Burton-West <tb...@umich.edu> wrote:

> Hi Otis,
>
> I hope this is not off-topic,
>
> Apparently in Lucene similarity does not have to be set at index time:
>

Actually in the general case it does. IndexWriter calls the Similarity's
computeNorm method at index-time.
Its just that the models we added for that GSOC project all have the same
computeNorm implementation at the moment, so they are "compatible".

On the other hand if you want to use something like SweetSpotSimilarity
which has a different computeNorm, then you need to reindex with it.


>
> See http://lucene.apache.org/core/4_0_0/changes/Changes.html under Lucene
> 2959
> "All models default to the same index-time norm encoding as
> DefaultSimilarity, so you can easily try these out/switch back and
> forth/run experiments and comparisons without re-indexing."
>
> Does Solr expose this ability to change similarities without re-indexing?
>  i.e could you just change your schema?
>

yes thats right.

Re: Superset Similarity?

[Tom Burton-West <tb...@umich.edu>]

Hi Otis,

I hope this is not off-topic,

Apparently in Lucene similarity does not have to be set at index time:

See http://lucene.apache.org/core/4_0_0/changes/Changes.html under Lucene
2959
"All models default to the same index-time norm encoding as
DefaultSimilarity, so you can easily try these out/switch back and
forth/run experiments and comparisons without re-indexing."

Does Solr expose this ability to change similarities without re-indexing?
 i.e could you just change your schema?

Tom

http://www.hathitrust.org/blogs/large-scale-search

On Thu, Nov 15, 2012 at 11:31 PM, Otis Gospodnetic <
otis.gospodnetic@gmail.com> wrote:

> Hi,
>
> Somebody asked about BM25 on solr-user and comparison to other relevancy
> scoring methods.  Similarity has to be set at index time and should not be
> changed... but is there any one Similarity that happens to be a superset of
> them all, so that one can use that Similarity during indexing, and then
> switch to other Similarities to query time?  I doubt it, but....maybe, just
> maybe... :)
>
> Thanks,
> Otis
> --
> Search Analytics - http://sematext.com/search-analytics/index.html
> Performance Monitoring - http://sematext.com/spm/index.html
>