lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Grant Ingersoll <gsing...@apache.org>
Subject Re: tf*idf scoring
Date Tue, 03 Nov 2009 13:24:08 GMT
Inline below

On Nov 3, 2009, at 2:30 AM, Markus Jelsma - Buyways B.V. wrote:

> Hello list,
>
>
> I have a question about Lucene's calculation of tf*idf value. I first
> noticed that Solr's tf does not compare to tf values based on
> calculation elsewhere such as
> http://odin.himinbi.org/idf_to_item:item/comparing_tf%3Aidf_to_item%
> 3Aitem_similarity.xhtml or http://en.wikipedia.org/wiki/Tf%E2%80%93idf
>
> The tf values returned by Solr are always integers and not normalized
> against the length of the corpus whilst the field in which it resides
> does not have omitNorms="true".
>
> Consider the following documents where the field subject is of the
> standard text_ws type:
>
> <result name="response" numFound="6" start="0">
>    <doc>
>        <str name="subject">a b c</str>
>    </doc>
>    <doc>
>        <str name="subject">d e f</str>
>    </doc>
>    <doc>
>        <str name="subject">x y z</str>
>    </doc>
>    <doc>
>        <str name="subject">a d x</str>
>    </doc>
>    <doc>
>        <str name="subject">a e z</str>
>    </doc>
>    <doc>
>        <str name="subject">c f z</str>
>    </doc>
> </result>
>
> Now, Solr's TermVector results for the first document:
>
> <lst name="doc-0">
>    <str name="uniqueKey">0</str>
>        <lst name="subject">
>        <lst name="a">
>            <int name="tf">1</int>
>            <lst name="positions">
>                <int name="position">0</int>
>             </lst>
>            <int name="df">3</int>
>            <double name="tf-idf">0.3333333333333333</double>
>        </lst>
>        <lst name="b">
>            <int name="tf">1</int>
>            <lst name="positions">
>                <int name="position">1</int>
>            </lst>
>            <int name="df">1</int>
>            <double name="tf-idf">1.0</double>
>        </lst>
>        <lst name="c">
>            <int name="tf">1</int>
>            <lst name="positions">
>                <int name="position">2</int>
>            </lst>
>            <int name="df">2</int>
>            <double name="tf-idf">0.5</double>
>        </lst>
>    </lst>
> </lst>
>
>
> According to different algorithms, the tf for term c would be 3 / 1 =
> 0.33 instead of 1 returned by Solr.

I don't follow.  The TF (term frequency) is the number of times the  
term c occurs in that particular document, i.e. 1 time.

> Also, the tf*idf value i get is 0.5
> for term c and i get 0.333 for term a. It looks like tf*idf is  
> quotient
> of document frequency and term frequency.

Yes, indeed.  IDF == Inverse Document Frequency, in other words, 1/DF.

>
> If i calculate tf*idf, for term c in the first document, according to
> other algorithms it would be:
>
> tf = 3 / 1 = 0.333

3/1 = 3, no?  I don't see where in your docs above you could even get  
a 3 for the letter c.

> idf = ln(6 / 3) = 1.0986
> tf*idf = 0.333 * 1.0986 = 0.3658
>

I think the formulas you are looking at are doing operations to  
normalize the values, whereas the Solr/Lucene stuff above is telling  
you their raw values.  Note, Lucene/Solr does length normalization,  
etc. too, it just isn't encoded into the TF or DF.  For more on  
Lucene's scoring, see http://lucene.apache.org/java/2_9_0/scoring.html

--------------------------
Grant Ingersoll
http://www.lucidimagination.com/

Search the Lucene ecosystem (Lucene/Solr/Nutch/Mahout/Tika/Droids)  
using Solr/Lucene:
http://www.lucidimagination.com/search


Mime
View raw message