Hi Viksit,
Why don't you try breaking the query and running a boolean boost query.
Building something like
("A B C"~1000)^100 OR ("A B"~1000 OR "B C"~1000 OR "A C"~1000)^10 OR (A OR B
OR C)Though this is not a fool proof way to do it and a manual merge is the right way. Also, I remember a similar question being posted in the past. You might want to have a look @ the archives. -- Anshum Gupta Naukri Labs! http://ai-cafe.blogspot.com The facts expressed here belong to everybody, the opinions to me. The distinction is yours to draw............ On Tue, Sep 29, 2009 at 10:56 AM, Viksit Gaur <vik.list.nu...@gmail.com>wrote: > Hi all, > > I'm trying to achieve the following, and wondered if I could get feedback > on how best to achieve it. > > Given an example phrase P - "Squeamish Ossifrage Monster", I'd like to > search a corpus such that in a list of results, > > - Docs with all 3 words in the phrase are ranked at the top > > - Docs with atleast 2 of the words in that order are ranked next > -- (Say, "Ossifrage Monster" and "Squeamish Ossifrage" but not "Squeamish > Monster") > > - Docs with only one of the words come next > -- Is there a way to put these into one result set, and the first 2 kinds > in another? > > The naive solution of course would be to take a phrase P, and then separate > out all its terms (Pt_1, Pt_2, Pt_3) and then do those 3 searches manually - > seems like a colossal waste. > > Is this possible with SpanQuery in some way? I also came across an article > somewhere which said that QueryParser doesn't support SpanQuery - in which > case, what would be the best way to actually implement this? > > Cheers > Viksit > > --------------------------------------------------------------------- > To unsubscribe, e-mail: java-user-unsubscr...@lucene.apache.org > For additional commands, e-mail: java-user-h...@lucene.apache.org > >
