Skip to main content

Mappings

Updated Dec 30, 2022 ·

Overview​

Mapping defines how Elasticsearch stores, indexes, and queries document fields.

  • Similar to SQL schemas, it sets field types and properties.
  • Ensures efficient storage and accurate searches.

A mapping is a schema definition. Elasticsearch has default settings, but you may customize it. For example, the following POST request:

curl -XPUT $ELASTIC_ENDPOINT:9200/movies -d '
{
  "mappings": {
    "properties": {
      "year": {
        "type": "date"
      }
    }
  }
}
'

Dynamic and Explicit Mapping​

Dynamic mapping in Elasticsearch automatically detects and maps fields in your data as they are indexed. This simplifies setup and allows Elasticsearch to adapt to changing data structures.

  • Automatically assigns data types based on content.
  • Maps new fields without manual intervention.

On the other hand, explicit mapping gives you full control over field definitions and data types.

  • Specifies field properties, such as data type and analyzers.
  • Ensures consistency in how data is indexed and queried.
  • Prevents unexpected mappings by defining them in advance.

Field Types​

Field types specify the format for each field in Elasticsearch.

curl -XPUT $ELASTIC_ENDPOINT:9200/movies -d '
{
  "mappings": {
    "properties": {
      "title": {
        "type": "text"
      },
      "release_date": {
        "type": "date"
      }
    }
  }
}
'

Field Index​

Field indexing controls how data is indexed and searched.

curl -XPUT $ELASTIC_ENDPOINT:9200/movies -d '
{
  "mappings": {
    "properties": {
      "title": {
        "type": "text",
        "index": true
      }
    }
  }
}
'

Field Analyzer​

Field analyzers define how text is processed for indexing and search, affecting how documents are stored and queried.

curl -XPUT $ELASTIC_ENDPOINT:9200/movies -d '
{
  "mappings": {
    "properties": {
      "description": {
        "type": "text",
        "analyzer": "standard"
      }
    }
  }
}
'

Sample Field Analyzers​

Field analyzers can include components like character filters, tokenizers, and token filters.

  • Character Filters

    • Modify text before tokenization.
    • Examples: removing punctuation or normalizing case.

  • Tokenizers

    • Break text into individual terms or tokens.
    • Split strings with whitespace, punctuations, or non-letters
    • Examples: include standard, whitespace, and keyword.

  • Token Filters

    • Process tokens after tokenization.
    • Lower-casing, stemming, synonyms, stopwords
    • Examples: remove stop words, modify token case, etc.

Common choices for analyzers:

  • Standard

    • Split on word boundaries, remove punctuation, lowercases
    • Good choice if language is unknown

  • Simple

    • Splits on anything that isn't a letter, and lowercase

  • Whitespace

    • Splits on whitespaces but don't lowercase

  • Language

    • Example: English
    • Accounts for language-specific stopwords and stemming.