CODEWITHSUNDEEP
elasticsearchelasticsearch-aggregation
user4906240
user4906240

Reputation: 625

Elastic Search Group by field

I have the following data in my index.

{
"id":1,
"car_name" : "ABC-101"
},
{
"id":2,
"car_name" : "DEF-102"
},
{
"id":3,
"car_name" : "ABC-103"
}

Mapping of my index is

{
  "car": {
    "mappings": {
      "_doc": {
        "properties": {
          "car_name": {
            "type": "text",
            "fielddata": true
          }
        }
      }
    }
  } 
}

Whey i run the following query

 localhost:9200/car/_doc/_search?pretty

With following request body

{
"size" : 0,
"aggs" : {
    "genres" : {
        "terms" : { 
            "field" : "car_name"
        }
    }
}

}

I get the following response

"buckets": [
    {
      "key": "ABC",
      "doc_count": 2
    },
    {
      "key": "DEF",
      "doc_count": 1
    },
    {
      "key": "101",
      "doc_count": 1
    },
    {
      "key": "102",
      "doc_count": 1
    },
    {
      "key": "103",
      "doc_count": 1
    }
]

Why it is not bring the actual keys which are ABC-101 and DEF-102 why ABC and 101 are treated as separated keys.

Upvotes: 2

Views: 17821

Answers (1)

A l w a y s S u n n y
A l w a y s S u n n y

Reputation: 38552

By default, string fields are analyzed in elasticsearch. It means that "ABC-101" is indexed as 2 terms "ABC" and "101". Your query is also analyzed, and it's also converted into 2 terms "ABC" and "101" no matter which special characters are in between them.

That's why they are matching all the strings separated by - like ABC, 101, DEF, 102 and so on.

e.g

  {
    "car": {
      "car_name": "string",
      "fields": {
        "raw": {
          "type": "string",
          "index": "not_analyzed"
        }
      }
    }  
  }

If you want to search this field exactly as it is, you should re-index it as "index":"not_analyzed"

You can make it work with keyword on car_name field to match exactly

{
  "size" : 0,
  "aggs" : {
      "genres" : {
          "terms" : { 
              "field" : "car_name.keyword"
         }
      }
   }
}

Upvotes: 6

Related Questions