0

I have few fields indexed as title, bookname, description, author, topic and others and also there is one field "TiBoDe" which contains all values from title, bookname, description.

Now if I do like TiBoDe:game it will not give relevant results So I want to achieve that if someone does like TiBoDe:game I want to boost Ti more than BoDe So I can get more relevant docs returned

Note: TiBoDe is one single fieldname which contains the value of three fields

Abhishek Patil
  • 1,373
  • 3
  • 30
  • 62
  • looks like TiBoDe is copy field...in this case it is not possible...its difficult to identify which values belongs to which field – Abhijit Bashetti Oct 19 '20 at 10:49
  • Is it possible if boost those three fields and send them one by one for scoring like title:game boost=10 , bookname:game boost=5 something like that whenever someone do like TiBoDe:game else any idea how can I achieve better results with this copyfield t – akshay kale Oct 19 '20 at 11:05
  • boostin on individual fields is possible ...would suggest...then use edismax parser and then apply the field level bossting...no need of copy field – Abhijit Bashetti Oct 19 '20 at 11:07
  • Edismax not possible in my case will have to change a lot of things Still I'll try to work on boosting thing Thanks for the help – akshay kale Oct 19 '20 at 11:17
  • edismax will allow you to attach boosts to each field, meaning that hits in title will be scored higher than in bookname or description, etc. In your case you can add multiple query parts to simulate it with the regular query parser, for example `title:game^10 bookname:game^5` etc. – MatsLindh Oct 19 '20 at 21:58
  • @MatsLindh I'm not able to split the query like let's say we do like TiBoDe:game I want to send it one by one for scoring with boost assingned Ti:game, Bi:game like this Can I achieve the results this way? I have provided boost value hard coded for each field in BooleanWeight – akshay kale Oct 20 '20 at 04:40
  • 1
    As Abhijit said if you're indexing it into one common field, you can't score different tokens in that field differently - except if you use payloads and a payload scorer (but that will require you to index content in a very specific format, and in that case it usually becomes better to split the content appropriately across multiple fields instead and score hits in the different fields differently) – MatsLindh Oct 20 '20 at 08:10

0 Answers0