: I don't think there is a way of seeing the "boosts" from the index, as
: those are encoded as "norms" (together with length normalization). You can
: see the norms with Luke if you want to and in the debugQuery output the
...
:
http://lucene.apache.org/java/3_1_0/api/all/org/apache/lucene/search/Similarity.html
: for
: Lucene/Solr 3.1) you can see how the norm is calculated, In your debugQuery
: I can see that all the "fieldNorm" are 1.5 and I'm not so sure of why that
: can happen to you.
I haven't done the math to check, but this is most likely because of the
precision loss involved when encoding the norm values as a single "byte"
-- boost values that are small and very close together aren't going to
provide much differentiation.
Bottom Line: if you want fine control over boosting with some per doc
numeric value, add that value as a field, and then either: factor it
directly into the score with a boosting query, or add it as a sort option
(ie: if you just want to break ties when scores are identical do
"sort=my_boolean+desc,score+desc,my_boost_field+desc")
-Hoss
