Re: Problem with extended dismax, minus prefix (to mean NOT) and interaction with mm?

2010-05-25 Thread Erik Hatcher
This looks like a case where the extended dismax parser is creating a Lucene QueryParser parsed query rather than a disjunction maximum query. A case of too much magic maybe? Looks like this one should be parsed quite differently. Try dismax and see what you get, it'll be quite

Re: Problem with extended dismax, minus prefix (to mean NOT) and interaction with mm?

2010-05-25 Thread Chris Hostetter
: I'm running edismax (on both a 1.4 with patch and a branch_3x version) and : I'm seeing something I don't expect. ... : str name=rawquerystringdog cat -trilogy/str : str name=querystringdog cat -trilogy/str : str name=parsedqueryallfields:dog allfields:cat : -allfields:trilogi/str :

Problem with extended dismax, minus prefix (to mean NOT) and interaction with mm?

2010-05-24 Thread Bill Dueber
I'm running edismax (on both a 1.4 with patch and a branch_3x version) and I'm seeing something I don't expect. We have our mm set such that 2/2 must match and 2/3 must match (mm=2-1 567% A query of dog cat ...gets interpreted as dog AND cat But a query of dog cat -mouse ...gets