You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-user@lucene.apache.org by Dino Korah <dc...@gmail.com> on 2007/10/19 18:39:45 UTC
Norm - please lit it up for me
Hi,
Could someone help me understand normalization factors for a field.
Also please tell me what are the situations where I should omit
normalization factors when adding a document.
Many thanks.
Dino Korah
Re: Norm - please lit it up for me
Posted by Daniel Naber <lu...@danielnaber.de>.
On Friday 19 October 2007 19:07, Karl Wettin wrote:
> doc[0] <text: hello hello hello>
> doc[1] <text: hello>
>
> With normalization doc[0] and doc[1] are equally important. Omitting
> normalization makes doc[0] (usually) three times as important as doc[1].
Not quite, as the normalization only refers to the length of the document.
But the fact that "hello hello hello" contains the searched term three
times makes it have a larger score no matter if normalization is set.
However, for this example, both docs have the same score with
f.setOmitNorms(true) when searching for "hello":
doc[0] <text: hello foo foo>
doc[1] <text: hello>
Regards
Daniel
--
http://www.danielnaber.de
---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org
Re: Norm - please lit it up for me
Posted by Karl Wettin <ka...@gmail.com>.
19 okt 2007 kl. 18.39 skrev Dino Korah:
> Could someone help me understand normalization factors for a field.
doc[0] <text: hello hello hello>
doc[1] <text: hello>
With normalization doc[0] and doc[1] are equally important. Omitting
normalization makes doc[0] (usually) three times as important as doc[1].
> Also please tell me what are the situations where I should omit
> normalization factors when adding a document.
Formula 1A is to omit normalization on fields that always contain a
single term such as primary key, timestamp, etc.
--
karl
---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org