Hmmmm. Could you expand a bit more on the problem you're trying to solve? The index organization you're hinting at seems close enough to a set of database tables to make me wonder if you're using an inappropriate index structure given the problem you want to solve.
Not that I know enough about your problem/solution to have a valid opinion, but there's at least a chance that this is an XY problem.... Best Erick On Wed, Jun 17, 2009 at 4:52 PM, MilkDud <jf...@limewire.com> wrote: > > Yea, not using stopwords at all. I do have tracks specified in the pf > param > along with a few other fields. That said, with a phrase query I lose the > ability to search for an artist and track combined. Two solutions i've > thought of include indexing at the track level only (right now i have > separate documents at the track, artist, and album level) or having a field > that contains both the artist and track name concatenated, allowing for > phrase queries containing bother artist and track names. > > > Michael Ludwig-4 wrote: > > > > MilkDud schrieb: > >> > >> That part I understand and is what I have now. It's the fact that > >> since tracks is multivalued, and i search for a track "love me", i > >> will also get back artists that have the words love and me in separate > >> tracks. > > > > Jason, > > > > are you sure "me" isn't in a stopword list used to analyze your query? > > Append debugQuery=true to find out whether by any chance it is removed > > from your query phrase. In that case, your phrase won't survive parsing, > > and all you'll be left with is "love" :-) > > > > But I guess there are quite a lot of "love" titles :-) > > > >> Now with a phrase query with a small ps and a large posIncGap that > >> could word. But then I lose the ability to search for artist and > >> track name together. > > > > Another thing, are you sure you have enabled "pf" for "track"? > > > > Michael Ludwig > > > > > > -- > View this message in context: > http://www.nabble.com/Searching-across-multivalued-fields-tp24056297p24076620.html > Sent from the Solr - User mailing list archive at Nabble.com. > >
