lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ype Kingma <ykin...@xs4all.nl>
Subject Re: scoring algorithm
Date Mon, 22 Sep 2003 04:29:40 GMT
On Tuesday 23 September 2003 00:12, Chris Hennen wrote:
> Hi,
>
> what is the purpose of "tf_q * idf_t / norm_q" in Lucene's scoring
> algorithm:
> score_d = sum_t( tf_q * idf_t / norm_q * tf_d * idf_t / norm_d_t)
>
> I dont understand, why the score has to be higher, when the frequency of a
> term in the query is higher. What is normalized by "norm_q"?

To give the user the possibility to assign a higher weight to a term in a
query, (by using a term weight or by repeating the term).
The norm_q compensates the total score for the query weights,
leaving the scores of two different queries somewhat comparable.

Kind regards,
Ype


Mime
View raw message