0

I am working on indexing a large data set which has multiple name fields for a particular entity. I have defined the name field of type array and I am adding around 4 names in that. Some of the names have spaces in between and they are getting tokenized. Can I avoid that?

I know for String we have text as well as keyword type in Elastic but how do I define the type as keyword when I am having array as my data type? By default all the array fields are taken as text type. I want them to be treated as keyword type so they don't get tokenized while indexing.

Expected : If I store "Hello World" in an array, I should be able to search "Hello World".

Current behavior : It stores hello differently and world differently as it tokenizes that.

Divyesh Kalbhor
  • 385
  • 3
  • 19

1 Answers1

1

There is no data type for array in elastic search. Whenever you send an array as value of a property of type x then that property becomes an array accepting only the values of type x.

So for example you created a property as below:

{
   "tagIds": {
      "type": "integer"
   }
}

And you index a document with values as below:

{
   "tagIds": [124, 452, 234]
}

Then tagIds automatically become an array of integers.

For your case all you need to do is create a field say name with type as keyword. And make sure you always pass an array to this field even if it has to hold a single value to make sure it is always an array. Below is what you need: Mapping:

PUT test
{
  "mappings": {
    "_doc": {
      "properties": {
        "name": {
          "type": "keyword"
        }
      }
    }
  }
}

Indexing document:

PUT test/_doc/1
{    
    "name" : ["name one"]
}
Nishant
  • 7,504
  • 1
  • 21
  • 34