Hi Siva,
try using the Solr Stats Component
http://wiki.apache.org/solr/StatsComponent
similar to
select/?&q=*:*&stats=true&stats.field={your-weight-field}&stats.facet={your-facet-field}
and get the sum field from the response. You may need to resort the weighted
facet counts to get a descending list of facet counts.
Note, there is a bug for using the Stats Component with multi-valued facet
fields.
For details see
https://issues.apache.org/jira/browse/SOLR-1782
Johannes
2011/1/24 Chris Hostetter <hossman_luc...@fucit.org>
>
> : prod1 has tag called “Light Weight” with weightage 20,
> : prod2 has tag called “Light Weight” with weightage 100,
> :
> : If i get facet for “Light Weight” , i will get Light Weight (2) ,
> : here i need to consider the weightage in to account, and the result will
> be
> : Light Weight (120)
> :
> : How can we achieve this?Any ideas are really helpful.
>
>
> It's not really possible with Solr out of the box. Faceting is fast and
> efficient in Solr because it's all done using set intersections (and most
> of the sets can be kept in ram very compactly and reused). For what you
> are describing you'd need to no only assocaited a weighted payload with
> every TermPosition, but also factor that weight in when doing the
> faceting, which means efficient set operations are now out the window.
>
> If you know java it would be probably be possible to write a custom
> SolrPlugin (a SearchComponent) to do this type of faceting in special
> cases (assuming you indexed in a particular way) but i'm not sure off hte
> top of my head how well it would scale -- the basic algo i'm thinking of
> is (after indexing each facet term wit ha weight payload) to iterate over
> the DocSet of all matching documents in parallel with an iteration over
> a TermPositions, skipping ahead to only the docs that match the query, and
> recording the sum of the payloads for each term.
>
> Hmmm...
>
> except TermPositions iterates over <term, <doc, freq, <position>>> tuples,
> so you would have to iterate over every term, and for every term then loop
> over all matching docs ... like i said, not sure how efficient it would
> wind up being.
>
> You might be happier all arround if you just do some sampling -- store the
> tag+weight pairs so thta htey cna be retireved with each doc, and then
> when you get your top facet constraints back, look at the first page of
> results, and figure out what the sun "weight" is for each of those
> constraints based solely on the page#1 results.
>
> i've had happy users using a similar appraoch in the past.
>
> -Hoss
--
Johannes Goll
211 Curry Ford Lane
Gaithersburg, Maryland 20878
