CODEWITHSUNDEEP
elasticsearch
Vignesh T.V.
Vignesh T.V.

Reputation: 1860

Get one document per bucket in elasticsearch aggregation

I have an aggregation in elasticsearch which gives a response like this:

{
  "took": 5,
  "timed_out": false,
  "_shards": {
    "total": 1,
    "successful": 1,
    "skipped": 0,
    "failed": 0
  },
  "hits": {
    "total": 1261,
    "max_score": 0,
    "hits": []
  },
  "aggregations": {
    "clusters": {
      "doc_count_error_upper_bound": 0,
      "sum_other_doc_count": 1073,
      "buckets": [
        {
          "key": 813058,
          "doc_count": 46
        },
        {
          "key": 220217,
          "doc_count": 29
        },
        {
          "key": 287763,
          "doc_count": 23
        },
        {
          "key": 527217,
          "doc_count": 20
        },
        {
          "key": 881778,
          "doc_count": 15
        },
        {
          "key": 700725,
          "doc_count": 14
        },
        {
          "key": 757602,
          "doc_count": 13
        },
        {
          "key": 467496,
          "doc_count": 10
        },
        {
          "key": 128318,
          "doc_count": 9
        },
        {
          "key": 317261,
          "doc_count": 9
        }
      ]
    }
  }
}

I want to get one document (either by top score or at random - anything works) for every bucket in the aggregation. How do I do that?

The query I am using to get the aggregation is this:

GET myindex/_search
{
    "size": 0,
    "aggs": {
        "clusters": {
            "terms": {
                "field": "myfield",
                "size": 100000
            }
        }
    },
    "query": {
                "bool": {
                    "must": [
                        {
                            "query_string": { "default_field": "field1", "query": "val1" }
                        },
                        {
                            "query_string": { "default_field": "field2", "query": "val2" }
                        }
                    ]
                }
            }
}

I am trying to implement a cluster based sentence similarity system and hence I need this. I pick one sentence from every cluster and check for similarity with a given sentence.

Upvotes: 2

Views: 1868

Answers (1)

Vignesh T.V.
Vignesh T.V.

Reputation: 1860

I was able to solve it by using the top hits aggregation given here: https://www.elastic.co/guide/en/elasticsearch/reference/current/search-aggregations-metrics-top-hits-aggregation.html

The sample query below:

GET myindex/_search
{
    "size": 0,
    "aggs": {
        "clusters": {
            "terms": {
                "field": "myfield",
                "size": 100000
            },
        "aggs": {
            "mydoc": {
                "top_hits": {
                    "size" : 1
                }
            }
        }
        }
    },
    "query": {
                "bool": {
                    "must": [
                        {
                            "query_string": { "default_field": "field1", "query": "val1" }
                        },
                        {
                            "query_string": { "default_field": "field2", "query": "val2" }
                        }
                    ]
                }
            }
}

Upvotes: 7

Related Questions