Considering multiple fields when searching in Elasticsearch

Question

I want to implement a simple search query using Elasticsearch.

I have two fields, "title" and "description" that I would like to match the searched term with. Currently, I have the body shown below as the body for search body. How can I make it so that the search prioritizes the title match, but if there are matches in the description, they are still included in the search (with lower priority)? Thanks in advance.

body = {
   size: 200,
   from: 0,
   query: {
      prefix: {
        title: searchTerm
      }
   }
}

Tom Slabbaert · Accepted Answer · 2021-05-04 10:26:48Z

1

You have to use a constant score query with a score of 0 for the "other" field. Any other boost / function score usage will not reliably score a certain field over another field as the scoring is based on other parameters like text length for example, this means a constant boost (unless very very large) can not guarantee the behaviour you seek.

By using a constant score for each field you can control score manually, like so:

{
    size: 200,
    from: 0,
    query: {
        bool: {
            should: [
                {
                    prefix: {
                        title: searchTerm
                    }
                },
                {
                    constant_score: {
                        filter: {
                            prefix: {
                                description: searchTerm
                            }
                        },
                        boost: 0
                    }
                },
            ]
        }
    }
}

If you set description boost to be more than 0 then the score will be the combined score of both fields, by doing this you can prioritize documents that have that prefix in both fields over ones that have it in just the title field.

edited May 4, 2021 at 10:26

answered May 4, 2021 at 8:32

Tom Slabbaert

22.5k10 gold badges34 silver badges45 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Tom Slabbaert Over a year ago

regarding a simple boost usage for example a description=prefix could be prioritized over title= ... alot of text ... prefix ... alot more text. I recommend reading about how elastic scores documents, if you can assume the factors that affect scoring are somewhat negligible between the fields (i.e certain length restrictions, etc) then you could use a simple boost method with some sort of threshold as the boost reliably.

Bhavya · Accepted Answer · 2021-05-04 08:25:11Z

1

You can use a combination of bool/should clause along with the boost parameter

{
  "query": {
    "bool": {
      "should": [
        {
          "prefix": {
            "title": {
              "value": "searchterm"
            }
          }
        },
        {
          "prefix": {
            "description": {
              "value": "searchterm",
              "boost": 4
            }
          }
        }
      ]
    }
  }
}

answered May 4, 2021 at 8:25

Bhavya

16.2k3 gold badges23 silver badges52 bronze badges

Collectives™ on Stack Overflow

Considering multiple fields when searching in Elasticsearch

2 Answers 2

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related