Fields with same name but different boosts

[Andrzej Bialecki <ab...@getopt.org>]

Hello,

I have the following problem: in my application I'm trying to store set 
of keywords and their weights. Since the number of keywords is variable 
(and can be as high as 40) I decided to use a single field to store it. 
In other words, I want to use a single field with multiple values to 
store a keyword histogram of the document.

Now, I'm wondering how do I encode the weight of keywords... If I do the 
following:

Field f = Field.Keyword("kw", "value1");
f.setBoost(10.0);
doc.add(f);
f = Field.Keyword("kw", "value2");
f.setBoost(20.0);
doc.add(f);

Now the question is: what is the boost value for the fields when I 
search? Is it equivalent to "value1^10.0 value2^20.0" (which is my 
intention), or rather "value1^20.0 value2^20.0"?

If the latter, do you have any suggestions how to achieve the original 
effect?

Thanks in advance!

-- 
Best regards,
Andrzej Bialecki

-------------------------------------------------
Software Architect, System Integration Specialist
CEN/ISSS EC Workshop, ECIMF project chair
EU FP6 E-Commerce Expert/Evaluator
-------------------------------------------------
FreeBSD developer (http://www.freebsd.org)




---------------------------------------------------------------------
To unsubscribe, e-mail: lucene-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: lucene-user-help@jakarta.apache.org

Re: Fields with same name but different boosts

[Doug Cutting <cu...@lucene.com>]

Andrzej Bialecki wrote:
> Now, I'm wondering how do I encode the weight of keywords... If I do the 
> following:
> 
> Field f = Field.Keyword("kw", "value1");
> f.setBoost(10.0);
> doc.add(f);
> f = Field.Keyword("kw", "value2");
> f.setBoost(20.0);
> doc.add(f);
> 
> Now the question is: what is the boost value for the fields when I 
> search? Is it equivalent to "value1^10.0 value2^20.0" (which is my 
> intention), or rather "value1^20.0 value2^20.0"?

I think the boost will be 200.0.  Boosts are multiplicative.

> If the latter, do you have any suggestions how to achieve the original 
> effect?

The only way to do this is to repeat occurences of each word.

So you might:

   for (int i = 0; i < 10; i++) {
     doc.add(new Field("kw", "value1", false, true, false);
   }

   for (int i = 0; i < 20; i++) {
     doc.add(new Field("kw", "value1", false, true, false);
   }

You might also consider using a custom Similarity implementation so that 
you can control the interpretation of these frequencies.  For example, 
you might, instead of 10 and 20, be able to just use 1 and 2 and then, 
in your Similarity.tf() implementation, turn this into whatever value 
you want used in the scoring.

Doug


---------------------------------------------------------------------
To unsubscribe, e-mail: lucene-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: lucene-user-help@jakarta.apache.org