Distance feature relevance score behaviour with boost > 1

Elastic Stack Elasticsearch
1

Hello,

I am confused about the calculation of relevance scores with distance_feature. In the documentation it says that:

The distance_feature query calculates a document’s relevance score as follows:
relevance score = boost * pivot / (pivot + distance)

I have the following query

{
  "_source": false,
  "query": {
    "distance_feature": {
      "field": "address.location",
      "origin": {
        "lat": 52.5065133,
        "lon": 13.1445545
      },
      "pivot": "100000m",
      "boost": __BOOST__
    }
  },
  "script_fields": {
    "distance": {
      "script": {
        "source": "doc['address.location'].arcDistance(params.lat,params.lon)",
        "params": {
          "lat": 52.5065133,
          "lon": 13.1445545
        }
      }
    }
  }
}

When running the query with boost=1.0, the documents are scored as expected.
The document below has score 1 * (100000/(100000+2398.9804327731676)) = 0.9765722.
All scores are in [0, 1].

{
  "_index" : "some_index",
  "_type" : "_doc",
  "_id" : "57114",
  "_score" : 0.9765722,
  "fields" : {
    "distance" : [
      2398.9804327731676
    ]
  }
}

Yet when running the same query with boost=2.0, some score values are bigger than 2.0.
For the same document the calculation does not work out anymore: 2 * (100000/(100000+2398.9804327731676)) = 1.9531444 yet the returned score is 3.9062889

{
  "_index" : "some_index",
  "_type" : "_doc",
  "_id" : "57114",
  "_score" : 3.9062889,
  "fields" : {
    "distance" : [
      2398.9804327731676
    ]
  }
}

Am I misunderstanding the docs or have a stupid mistake in my thought process :thinking:? I would be glad if someone could help me out with this :slightly_smiling_face:

EDIT Just to give some context: This is relevant for me because i sometimes use constant_score queries to make sure that certain hits are sorted to the top. In such a case it's useful to know that the distance_feature only adds to the score within a certain range. This is how I interpreted the formula, the additional score is in [0, boost].

Best regards

2

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.