Boost score based on integer value - Elasticsearch

Question

I'm not very experienced with ElasticSearch and would like to know how to boost a search based on a certain integer value.

This is an example of a document:

{
    "_index": "links",
    "_type": "db1",
    "_id": "mV32vWcBZsblNn1WqTcN",
    "_score": 8.115617,
    "_source": {
        "url": "example.com",
        "title": "Example website",
        "description": "This is an example website, used for various of examples around the world",
        "likes": 9,
        "popularity": 543,
        "tags": [
            {
                "name": "example",
                "votes": 5
            },
            {
                "name": "test",
                "votes": 2
            },
            {
                "name": "testing",
                "votes": 1
            }
        ]
    }
}

Now in this particular search, the focus is on the tags and I would like to know how to boost the _score and multiply it by the integer in the votes under tags.

If this is not possible (or very hard to achieve), I would simply like to know how to boost the _score by the votes (not under tags)

Example, add 0.1 to the _score for each integer in votes

This is the current search query I'm using (for searching tags only):

{
    "query": {
        "nested": {
            "path": "tags",
            "query": {
                "bool":{
                    "should":{
                        "match":{
                            "tags.name":"example,testing,something else"
                        }
                    }
                }
            }
        }
    }
}

I couldn't find much online, and hope someone can help me out.

How do I boost the _score with an integer value?

Update

For more info, here is the mapping:

{
    "links": {
        "mappings": {
            "db1": {
                "properties": {
                    "url": {
                        "type": "text",
                        "fields": {
                            "keyword": {
                                "type": "keyword",
                                "ignore_above": 256
                            }
                        }
                    },
                    "title": {
                        "type": "text",
                        "fields": {
                            "keyword": {
                                "type": "keyword",
                                "ignore_above": 256
                            }
                        }
                    },
                    "description": {
                        "type": "text",
                        "fields": {
                            "keyword": {
                                "type": "keyword",
                                "ignore_above": 256
                            }
                        }
                    },
                    "likes": {
                        "type": "long"
                    },
                    "popularity": {
                        "type": "long"
                    },
                    "tags": {
                        "type": "nested",
                        "properties": {
                            "name": {
                                "type": "text",
                                "fields": {
                                    "keyword": {
                                        "type": "keyword",
                                        "ignore_above": 256
                                    }
                                }
                            },
                            "votes": {
                                "type": "long"
                            }
                        }
                    }
                }
            }
        }
    }
}

Update 2

Changed the tags.likes/tags.dislikes to tags.votes, and added a nested property to the tags

@Val I updated the question, and added the mapping. I don't know if this is nested or not. — Typewar
– Typewar, Commented Dec 23, 2018 at 16:35
Thanks, the issue is then that tags should be nested if you want to achieve what you expect, otherwise you cannot query specific nested tags in order to get their likes/dislikes values to boost the score. — Val
– Val, Commented Dec 23, 2018 at 16:37
@Val I would expect something like that. But by nesting the tags, does it mean that I can't set custom/new "names"? Would the mapping need to be HUGE to cover all the tags in the world? Or would I just need to change the mapping a bit? I'm considering making a db2, and slowly migrate all the documents to a new mapping structure — Typewar
– Typewar, Commented Dec 23, 2018 at 16:42
See this: elastic.co/guide/en/elasticsearch/reference/current/nested.html — Val
– Val, Commented Dec 23, 2018 at 16:43

Typewar · Accepted Answer · 2019-01-05 06:41:57Z

This took a long time to figure out. I have learnt so much on my way there.

Here is the final result:

{
    "query": {
        "nested": {
            "path": "tags",
            "query": {
                "function_score": {
                    "query": {
                        "bool": {
                            "should": [
                                {
                                    "match": {
                                        "tags.name": "example"
                                    }
                                },
                                {
                                    "match": {
                                        "tags.name": "testing"
                                    }
                                },
                                {
                                    "match": {
                                        "tags.name": "test"
                                    }
                                }
                            ]
                        }
                    },
                    "functions": [
                        {
                            "field_value_factor": {
                                "field": "tags.votes"
                            }
                        }
                    ],
                    "boost_mode": "multiply"
                }
            }
        }
    }
}

The array in should has helped a lot, and was glad I could combine it with function_score

Thomas Decaux · Accepted Answer · 2018-12-23 11:31:10Z

2

You are looking at function score query: https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html

And field value factor https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html#function-field-value-factor.

Snippet from documentation:

GET /_search
{
    "query": {
        "function_score": {
            "field_value_factor": {
                "field": "tags.dislikes",
                "factor": 1.2,
                "modifier": "sqrt",
                "missing": 1
            }
        }
    }
}

Or with script score because your nested tags field (not sure if field value score works fine with nested structure).

answered Dec 23, 2018 at 11:31

Thomas Decaux

22.8k3 gold badges129 silver badges139 bronze badges

Collectives™ on Stack Overflow

Boost score based on integer value - Elasticsearch

Update

Update 2

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

Update

Update 2

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related