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 David Ryan <he...@gmail.com> on 2011/10/05 22:19:07 UTC

Scoring in Lucene

Hi,

The defaulting scoring in Lucene  uses tf x idf^2 instead of tf x idf .
Does any have have insight that why not using tf x idf?


Here is the note on score calculation.
https://lucene.apache.org/java/3_4_0/api/core/org/apache/lucene/search/Similarity.html

Re: Scoring in Lucene

Posted by Doron Cohen <cd...@gmail.com>.

To my understanding this stems from V(q) · V(d) (see the "*Conceptual
Scoring Formula*") - the elements in those vectors are *Tf-idf* values, and
so, implementation wise (see the "*Practical Scoring Function*"),  idf(t) is
multiplied by itself: once for the query and once for the document.

HTH,
Doron

On Wed, Oct 5, 2011 at 10:19 PM, David Ryan <he...@gmail.com> wrote:

> Hi,
>
> The defaulting scoring in Lucene  uses tf x idf^2 instead of tf x idf .
> Does any have have insight that why not using tf x idf?
>
>
> Here is the note on score calculation.
>
> https://lucene.apache.org/java/3_4_0/api/core/org/apache/lucene/search/Similarity.html
>