On Tue, Aug 11, 2009 at 10:35 AM, leibniz<[email protected]> wrote: > > i saw someone said that this fieldNor factor is consisted out of :docboost > ,fieldboost and lengthNorm. > but i tried to computed like this,but the result is not the same as lucene's > output. The factors you use seem to be correct (see link below). I could imagine that you use different float values than the ones used in the scoring process. Lucene stores norm values as a 1 byte float value this will suffer from precision lost once the values are stored. Could that cause the differences?
> who can tell me in detail?thanks. You find the details for computing the norm value here: http://lucene.apache.org/java/2_4_1/api/core/org/apache/lucene/search/Similarity.html#formula_norm simon > -- > View this message in context: > http://www.nabble.com/how-is-the-fieldNorm-factor-computed-from--tp24913790p24913790.html > Sent from the Lucene - General mailing list archive at Nabble.com. > >
