Question

i want to compute a final score for each document by summing the weighted scores of multiple fields scores,instead of getting the MAX as the edisMAx query parser does.

So if i got three fields;title and description, i want my final score to be like this:

FINAL SCORE = w1 * title_field_score + (1-w1) * description_field_score + (1-w1 -w2) * description_field_score

Where w1,w2 can be values between 0 to 1.

The default similarity with no any simialrity gives this ranking. This uses the following search handler:

  <requestHandler name="/select" class="solr.SearchHandler">
 <lst name="defaults">
   <str name="echoParams">explicit</str>
   <int name="rows">10</int>
   <str name="df">detail_page_text</str>

   <!-- Query settings -->
   <str name="defType">edismax</str>
   <str name="qf">
      job_title description detail_page_text
   </str>

now by using the tie as below, seems that we getting a ideal ranking.

<str name="tie">1.0</str>

So my question is if the use of the tie parameter is the right way to achieve a weigthed ranking score based on all fields equaly(or with additional weights).

Any tip would be much appreciated!
Cheers

Answer 1

Dismax tie is the parameter you should use for weighted scoring.

tie = 1.0 -> Sum of All the Individual Scores.

19.07% 0.69268036 (MATCH) **max plus 1.0 times** others of:
    1.96% 0.07105611 detail_page_text:senior
    4.16% 0.15101658 description:senior
    12.96% 0.4706077 job_title:senior

tie = 0.0 -> Max of the Individual Scores.

17.83% 0.7043001 (MATCH) **max** of:  
    - 0.028197879 detail_page_text:senior  
    - 0.04566047 description:senior  
    - 17.83% 0.7043001 job_title:senior