Zach Nelson
2006-08-31 16:34:02 UTC
I'm trying to tune and refine my query building routine and having
trouble with something that should be really simple. I have 2 word
docs, one named flavors.doc and one named colors.doc. Inside both are
some appropriate words (colors and flavors). They both have the word
'orange' in them because it applies to both concepts.
When I search for 'orange' I expect to see both docs in the results,
and they do show up. However, if I search for 'flavor orange' I would
expect to see flavors.doc ranked higher because of the match on both
the filename and the contents. However, both are ranked 0.
When I search for 'flavor', just flavors.doc is returned with a rank of
0 and a hitcount of 1. When I search for 'flavor orange' both are
returned with a rank of 0, and both hitcounts are 1 even though there
should be a hit for both 'flavor' (in the filename) and 'orange' in the
contents.
Here is my query:
( {vector RankMethod=Jaccard}{ve} {weight value=0.40}*flavor*{/vector}
or {prop name=filename}{weight value=0.90}*flavor*{/prop} )
As you can see, a filename hit is weighted higher, yet both ranks and
hitcounts come out 0 and flavors is listed as result #2 (even though
I'm sorting by doclastsaved[d] and flavors.doc is newer).
Also, side question: How can I filter results based on a specific value
of a property? For example, only show results where IsAdmin = 0. Right
now I've got that in an AND condition but not 100% comfortable that it
will always filter-out any docs where IsAdmin = 1.
Thanks in advance!
trouble with something that should be really simple. I have 2 word
docs, one named flavors.doc and one named colors.doc. Inside both are
some appropriate words (colors and flavors). They both have the word
'orange' in them because it applies to both concepts.
When I search for 'orange' I expect to see both docs in the results,
and they do show up. However, if I search for 'flavor orange' I would
expect to see flavors.doc ranked higher because of the match on both
the filename and the contents. However, both are ranked 0.
When I search for 'flavor', just flavors.doc is returned with a rank of
0 and a hitcount of 1. When I search for 'flavor orange' both are
returned with a rank of 0, and both hitcounts are 1 even though there
should be a hit for both 'flavor' (in the filename) and 'orange' in the
contents.
Here is my query:
( {vector RankMethod=Jaccard}{ve} {weight value=0.40}*flavor*{/vector}
or {prop name=filename}{weight value=0.90}*flavor*{/prop} )
As you can see, a filename hit is weighted higher, yet both ranks and
hitcounts come out 0 and flavors is listed as result #2 (even though
I'm sorting by doclastsaved[d] and flavors.doc is newer).
Also, side question: How can I filter results based on a specific value
of a property? For example, only show results where IsAdmin = 0. Right
now I've got that in an AND condition but not 100% comfortable that it
will always filter-out any docs where IsAdmin = 1.
Thanks in advance!