This is the kind of use case the RankQuery API was created for. It allows you to write your own Lucene ranking collector and plug it in. It's an expert level java API so you'll need to program in Java and understand a lot about how Lucene collectors work, but it's cool stuff to learn.
Joel Bernstein Search Engineer at Heliosearch On Thu, Jul 24, 2014 at 2:59 PM, babenis <babe...@gmail.com> wrote: > Could you possibly elaborate on what that function could look like and how > to > use it? > > I have an ecommerce site with lots of products and some categories have 50 > times more products than others, and i would like to "shuffle" resultset in > a way that if the search is conducted by parent category id then all child > categories found in the resultset should be shuffled in a way that some > categories will have a little more results, while others have a little less > of the results and products from the same brand should be scattered across > the resultset, but to still be able to use pagination (ie 12 results per > page, page 1 2 and 3 should have different products, etc.)? > > ideally i'd like to use products tags as well to separate similar products > from each other even further > > thank you so much for any help > > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Shuffling-results-tp497372p4149092.html > Sent from the Solr - User mailing list archive at Nabble.com. >
