Something you could do via function queries. Performance (for 500+ words) is a doubtful.
1) With a separate float field (myfieldwordcount) that holds the count of words from your query field (myfield): http://localhost:8983/solr/collection1/select?wt=xml&indent=true&defType=func &fl=id,myfield &q={!frange l=0.75}div(sum(exists(query({!dismax qf=myfield v='word1'})),exists(query({!dismax qf=myfield v='word2'})),exists(query({!dismax qf=myfield v='word3'}))),myfieldwordcount) The value for myfieldwordcount will need to worked out externally and added to Solr for every document. 2) Using field norm as an approximation for word count:* *(Use default boost =1.0, see: http://wiki.apache.org/solr/SolrRelevancyFAQ#index-time_boosts) * * http://localhost:8983/solr/collection1/select?wt=xml&indent=true&defType=func &fl=id,myfield &q={!frange l=0.75}product(sum(exists(query({!dismax qf=myfield v='word1'})),exists(query({!dismax qf=myfield v='word2'})),exists(query({!dismax qf=myfield v='word3'}))),norm(myfield),norm(myfield)) Regards, Aloke On Wed, Oct 9, 2013 at 7:21 PM, shahzad73 <shahzad...@yahoo.com> wrote: > Please help me formulate the query that will be easy or do i have to build > a > custom filter for this ? > > Shahzad > > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Find-documents-that-are-composed-of-words-tp4094264p4094372.html > Sent from the Solr - User mailing list archive at Nabble.com. >
