The edismax query handler will match the best field, not all fields. So for some documents it matches "t-shirt" and for some it matches "white".
Try creating a single field with all the description information. That field will have both "white" and "t-shirt". Then add pf2, pf3, and pf config lines. Those will improve phrase matches. It might look like this: <str name="qf">item_desc item_category item_color</str> <str name="pf2">item_desc^2 item_category^2 item_color^2</str> <str name="pf3">item_desc^2 item_category^2 item_color^2</str> <str name="pf">item_desc^4 item_category^4 item_color^4</str> wunder On Jun 23, 2014, at 9:08 PM, Aman Tandon <amantandon...@gmail.com> wrote: > I translated your query string(白色帽子) in which is in japanese i guess to > English, it shows me the meaning of string is white hat, may be that's why > you is getting the all records containing white only. Your query looks fine > to me in case if you pass the white T shirt as well. > > And i am not wrong here then if your are searching with double quotes > "white T shirt" then your df fields also not be tokenized. > > Anybody please correct me if i am wrong here. I am just a beginner in solr > :) > > With Regards > Aman Tandon > > > On Tue, Jun 24, 2014 at 9:27 AM, xiaoqi <belivexia...@gmail.com> wrote: > >> >> http://localhost:7080/solr/select?l=*,score&start=0&q=%E7%99%BD%E8%89%B2%E5%B8%BD%E5%AD%90&qf=item_category >> ^1.4+item_colour^0.5&bf=mul(ctr,2.5)&wt=xml&fq=item_id:[0+TO+*]&fq=&rows=60&defType=edismax&version=2&debugQuery=on >> >> >> >> -- >> View this message in context: >> http://lucene.472066.n3.nabble.com/Problem-limit-results-when-using-edismax-way-tp4143601p4143606.html >> Sent from the Solr - User mailing list archive at Nabble.com. >> -- Walter Underwood wun...@wunderwood.org
