You are viewing a plain text version of this content. The canonical link for it is here.

Posted to solr-user@lucene.apache.org by 刘健 <li...@duokan.com> on 2013/08/20 03:50:35 UTC

Custom Sort(0.2relervanceScore + 0.8numberic_field_value)

Hello:

     I want to get final search result sorted by (0.2*relervance score + 0.8* specified_numberic_field) .

I have known that if I use “bf”in edismax (e.g. bf=field(value)), I can get a result sorted by(relervance sore + field(value)) ,but I don`t know how to Implement the result sorted by (0.2*relervance score + 0.8*specified_numberic_field) .

Thanks!

Leo

Re: Custom Sort(0.2relervanceScore + 0.8numberic_field_value)

Posted by 刘健 <li...@duokan.com>.

Thank you very much!

Then could you tell me how to implement relervance_score*numberic_field/(relervance_score + numberic_field)  ? I think it's better to sort by harmmean 

------------------ Original ------------------
From:  "Jack Krupansky"<ja...@basetechnology.com>;
Date:  Tue, Aug 20, 2013 10:47 AM
To:  "solr-user"<so...@lucene.apache.org>; 

Subject:  Re: Custom Sort(0.2*relervanceScore + 0.8*numberic_field_value)

Edismax applies the multiplicative boost ("boost") after applying the 
additive boost functions ("bf").

I think (0.2*relervance score + 0.8* specified_numberic_field) should be 
equivalent to:

0.2*(relevance score + (0.8/0.2)* specified_numeric_field)
or
0.2*(relevance score + 4.0* specified_numeric_field)

Unless the actual numeric range of the scores matters, just use:

relevance score + 4.0* specified_numeric_field

or

bf=mul(4.0, specified_numeric_field)

-- Jack Krupansky

-----Original Message----- 
From: 刘健
Sent: Monday, August 19, 2013 9:50 PM
To: solr-user
Subject: Custom Sort(0.2*relervanceScore + 0.8*numberic_field_value)

Hello:

     I want to get final search result sorted by (0.2*relervance score + 
0.8* specified_numberic_field) .

I have known that if I use “bf”in edismax (e.g. bf=field(value)), I can get 
a result sorted by(relervance sore + field(value)) ,but I don`t know how to 
Implement the result sorted by (0.2*relervance score + 
0.8*specified_numberic_field) .

Thanks!

Leo

Re: Custom Sort(0.2relervanceScore + 0.8numberic_field_value)

Posted by Jack Krupansky <ja...@basetechnology.com>.

Edismax applies the multiplicative boost ("boost") after applying the 
additive boost functions ("bf").

I think (0.2*relervance score + 0.8* specified_numberic_field) should be 
equivalent to:

0.2*(relevance score + (0.8/0.2)* specified_numeric_field)
or
0.2*(relevance score + 4.0* specified_numeric_field)

Unless the actual numeric range of the scores matters, just use:

relevance score + 4.0* specified_numeric_field

or

bf=mul(4.0, specified_numeric_field)

-- Jack Krupansky

-----Original Message----- 
From: 刘健
Sent: Monday, August 19, 2013 9:50 PM
To: solr-user
Subject: Custom Sort(0.2*relervanceScore + 0.8*numberic_field_value)

Hello:

     I want to get final search result sorted by (0.2*relervance score + 
0.8* specified_numberic_field) .

I have known that if I use “bf”in edismax (e.g. bf=field(value)), I can get 
a result sorted by(relervance sore + field(value)) ,but I don`t know how to 
Implement the result sorted by (0.2*relervance score + 
0.8*specified_numberic_field) .

Thanks!

Leo