: I'm using the edismax handler, although my question is probably the same for
: dismax. When the user types a long query, I use the mm parameter so that
: only 75% of terms need to match. This works fine, however, sometimes documents
: that only match 75% of the terms show up higher in my results than documents
: that match 100%. I'd like to set a boost so that documents that match 100%
: will be much more likely to be put ahead of documents that only match 75%. Can
: anyone give me a pointer of how to do this? Thanks,
this is essentially the default behavior -- mm just sets a minimum
number of clauses to be considered a match, but the coord factor still
applies and penalizes docs based on how many clauses they don't match.
if you are seeing docs that match fewer terms score higher then docs
matching more terms it is likely because of the boosts you already have
specified (either in the qf, or maybe using the bf), but the discrepency
could be based on other standard scoring factors as well (lengthNorm,
index time doc boosts, the IDF of the terms, etc...
this is where it beocmes neccessary to start looking at score explanations
and really thinking through the data.
-Hoss
