so here are my results with my special stripped title without any special chars:
query: ( stitle:WOBENZYM AND stitle:100 ) #01 Wobenzym P 100 Stück [0.866468607375] #02 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #03 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #04 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #05 WOBENZYM N Drag magensaftres 200 St [0.649851455532] #06 WOBENZYM N Drag magensaftres 800 St [0.649851455532] #07 WOBENZYM N Drag magensaftres 100 St [0.649851455532] query: ( stitle:WOBENZYM OR stitle:100 ) #01 Wobenzym P 100 Stück [0.866468607375] #02 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #03 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #04 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #05 WOBENZYM N Drag magensaftres 200 St [0.649851455532] #06 WOBENZYM N Drag magensaftres 800 St [0.649851455532] #07 WOBENZYM N Drag magensaftres 100 St [0.649851455532] query: ( stitle:WOBENZYM~0.6 OR stitle:100~0.6 ) #01 Wobenzym P 100 Stück [0.866468607375] #02 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #03 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #04 WOBENZYM mono Tabletten magensaftresistent [0.758160031454] #05 WOBENZYM N Drag magensaftres 200 St [0.649851455532] #06 WOBENZYM N Drag magensaftres 800 St [0.649851455532] #07 WOBENZYM N Drag magensaftres 100 St [0.649851455532] so, unfortunately, different scores, but same order. i am starting to think it could be a problem searching for numbers, because when i use this query: query: ( stitle:WOBENZYM~0.6 OR stitle:ST~0.6 ) #01 WOBENZYM N Drag magensaftres 200 St [1] #02 WOBENZYM N Drag magensaftres 800 St [1] #03 WOBENZYM N Drag magensaftres 100 St [1] #04 Wobenzym P 100 Stück [0.157236973127] #05 WOBENZYM mono Tabletten magensaftresistent [0.137582351486] #06 WOBENZYM mono Tabletten magensaftresistent [0.137582351486] #07 WOBENZYM mono Tabletten magensaftresistent [0.137582351486] all seems fine here, but i have to be able to search for numbers, since its a search for drugs, and the user should be able to enter numbers for packaging sizes etc. any help greatly appreciated. regards benjamin kampfhering wrote: > > hallo everyone, > > i am having some trouble getting my results in the right order. > > my query is: > > ( title:WOBENZYM~0.6 OR title:100~0.6 ) > > and these are my results (score in brackets): > > #01 Wobenzym P 100 Stück [1] > #02 WOBENZYM mono Tabletten magensaftresistent [0.875] > #03 WOBENZYM mono Tabletten magensaftresistent [0.875] > #04 WOBENZYM mono Tabletten magensaftresistent [0.875] > #05 WOBENZYM N Drag.magensaftres. / 100 St [0.75] > #06 WOBENZYM N Drag.magensaftres. / 200 St [0.75] > #07 WOBENZYM N Drag.magensaftres. / 800 St [0.75] > > what i don't understand is, why does the #05 has a lower score than #02 - > #04, because #05 has the number 100 in it and #02 - #04 doesn't. i've read > some messages on the nabble java forum, and there seems to be an explain > class which can help determine how the score got calculated. is there an > equivalent for zend_search_lucene? i also tried to change my query a bit, > here are some of my results: > > query: ( title:WOBENZYM~0.6 AND title:100~0.6 ) > > #01 Wobenzym P 100 Stück [1] > #02 WOBENZYM mono Tabletten magensaftresistent [0.875] > #03 WOBENZYM mono Tabletten magensaftresistent [0.875] > #04 WOBENZYM mono Tabletten magensaftresistent [0.875] > #05 WOBENZYM N Drag.magensaftres. / 100 St [0.75] > #06 WOBENZYM N Drag.magensaftres. / 200 St [0.75] > #07 WOBENZYM N Drag.magensaftres. / 800 St [0.75] > > so no matter if its AND or OR, no difference in the score here. > > query: ( title:WOBENZYM~0.9 OR title:100~0.9 ) > > #01 Wobenzym P 100 Stück [1] > #02 WOBENZYM mono Tabletten magensaftresistent [0.875] > #03 WOBENZYM mono Tabletten magensaftresistent [0.875] > #04 WOBENZYM mono Tabletten magensaftresistent [0.875] > #05 WOBENZYM N Drag.magensaftres. / 100 St [0.75] > #06 WOBENZYM N Drag.magensaftres. / 200 St [0.75] > #07 WOBENZYM N Drag.magensaftres. / 800 St [0.75] > > and fuzzy to 0.9 doesn't change anything either. > > query: ( title:WOBENZYM OR title:100 ) > > #01 Wobenzym P 100 Stück [1] > #02 WOBENZYM mono Tabletten magensaftresistent [0.875] > #03 WOBENZYM mono Tabletten magensaftresistent [0.875] > #04 WOBENZYM mono Tabletten magensaftresistent [0.875] > #05 WOBENZYM N Drag.magensaftres. / 100 St [0.75] > #06 WOBENZYM N Drag.magensaftres. / 200 St [0.75] > #07 WOBENZYM N Drag.magensaftres. / 800 St [0.75] > > even completely without fuzzy, i get exactly the same scoring. > > i have the feeling the problem could be the slash in the title. maybe it > has a problem with that. i will build a new lucene index with some > stripped titles without special chars, and see what that does get me. i > will post the results later on. if someone has another idea, please let me > know. > > regards benjamin > -- View this message in context: http://www.nabble.com/understanding-lucene-score%2C-kind-of-weird-results%2C-explain-class-available--tp26092008p26092209.html Sent from the Zend Framework mailing list archive at Nabble.com.
