Custom analyzers in Elasticsearch

Introduction

Custom analyzers help you control how text is broken down and searched in Elasticsearch. They let you make search smarter and more accurate for your needs.

You want to ignore certain words like 'the' or 'and' in searches.

You need to make searches case-insensitive so 'Apple' and 'apple' match.

You want to break text into words differently, like splitting on hyphens.

You want to add special rules for stemming words to match similar forms.

You want to handle languages with special characters or accents properly.

Syntax

Elasticsearch

{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_custom_analyzer": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase", "stop"]
        }
      }
    }
  }
}

The type must be custom to create your own analyzer.

You choose a tokenizer to split text into words.

Examples

This analyzer splits text normally and makes all words lowercase.

Elasticsearch

{
  "settings": {
    "analysis": {
      "analyzer": {
        "simple_lowercase": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase"]
        }
      }
    }
  }
}

This analyzer also removes common stop words like 'and', 'the'.

Elasticsearch

{
  "settings": {
    "analysis": {
      "analyzer": {
        "stopword_remover": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase", "stop"]
        }
      }
    }
  }
}

This analyzer creates partial word matches for autocomplete using edge n-grams.

Elasticsearch

{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_edge_ngram": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase", "edge_ngram"]
        }
      },
      "filter": {
        "edge_ngram": {
          "type": "edge_ngram",
          "min_gram": 2,
          "max_gram": 5
        }
      }
    }
  }
}

Sample Program

This creates an index with a custom analyzer that lowercases text and removes stop words. Then it shows how the text is broken down by the analyzer.

Elasticsearch

PUT /my_index
{
  "settings": {
    "analysis": {
      "analyzer": {
        "custom_lowercase_stop": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase", "stop"]
        }
      }
    }
  }
}

GET /my_index/_analyze
{
  "analyzer": "custom_lowercase_stop",
  "text": "The Quick Brown Fox jumps over the lazy Dog"
}

Important Notes

Custom analyzers let you mix and match tokenizers and filters to fit your search needs.

Stop words are common words that usually don't add meaning and can be removed to improve search speed.

Always test your analyzer with the _analyze API to see how text is processed.

Summary

Custom analyzers control how Elasticsearch breaks down and processes text.

You can add filters like lowercase and stop words to improve search results.

Testing analyzers helps you understand how your text is indexed and searched.

Practice

1. What is the main purpose of a custom analyzer in Elasticsearch?

easy

A. To control how text is broken down and processed before indexing

B. To store documents in the database

C. To create visualizations of search results

D. To manage user access permissions

Solution

1. Step 1: Understand the role of analyzers

   Analyzers in Elasticsearch process text by breaking it into tokens and applying filters.

2. Step 2: Identify the purpose of custom analyzers

   Custom analyzers let you define specific tokenizers and filters to control text processing.

3. Final Answer:

   To control how text is broken down and processed before indexing -> Option A

4. Quick Check:

   Custom analyzer = text processing control [OK]

Hint: Custom analyzers shape text before search indexing [OK]

Common Mistakes:

• Confusing analyzers with data storage
• Thinking analyzers create visualizations
• Mixing analyzers with security settings

2. Which of the following is the correct way to define a custom analyzer with a lowercase filter in Elasticsearch settings?

easy

A. "analyzer": {"my_analyzer": {"type": "custom", "tokenizer": "standard", "filter": ["lowercase"]}}

B. "analyzer": {"my_analyzer": {"type": "standard", "tokenizer": "lowercase", "filter": ["custom"]}}

C. "analyzer": {"my_analyzer": {"type": "custom", "tokenizer": "lowercase", "filter": ["standard"]}}

D. "analyzer": {"my_analyzer": {"type": "filter", "tokenizer": "standard", "filter": ["lowercase"]}}

Solution

1. Step 1: Review custom analyzer syntax

   A custom analyzer requires "type": "custom", a tokenizer, and filters as an array.

2. Step 2: Match correct tokenizer and filter

   "standard" tokenizer and "lowercase" filter are valid and correctly placed in "analyzer": {"my_analyzer": {"type": "custom", "tokenizer": "standard", "filter": ["lowercase"]}}.

3. Final Answer:

   "analyzer": {"my_analyzer": {"type": "custom", "tokenizer": "standard", "filter": ["lowercase"]}} -> Option A

4. Quick Check:

   Custom analyzer syntax = "analyzer": {"my_analyzer": {"type": "custom", "tokenizer": "standard", "filter": ["lowercase"]}} [OK]

Hint: Custom analyzer needs type, tokenizer, and filter array [OK]

Common Mistakes:

• Using wrong type like 'filter' instead of 'custom'
• Swapping tokenizer and filter names
• Missing array brackets for filters

3. Given this custom analyzer configuration:

{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_analyzer": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase", "stop"]
        }
      },
      "filter": {
        "stop": {
          "type": "stop",
          "stopwords": ["the", "is"]
        }
      }
    }
  }
}

What is the output tokens when analyzing the text "The cat is on the mat" with my_analyzer?

medium

A. ["The", "cat", "is", "on", "the", "mat"]

B. ["cat", "is", "on", "mat"]

C. ["the", "cat", "is", "on", "the", "mat"]

D. ["cat", "on", "mat"]

Solution

1. Step 1: Apply lowercase filter

   All words become lowercase: "the", "cat", "is", "on", "the", "mat".

2. Step 2: Apply stop filter with stopwords ["the", "is"]

   Remove "the" and "is" from tokens, leaving "cat", "on", "mat".

3. Final Answer:

   ["cat", "on", "mat"] -> Option D

4. Quick Check:

   Lowercase + stop filter removes 'the' and 'is' [OK]

Hint: Lowercase then remove stopwords from tokens [OK]

Common Mistakes:

• Ignoring lowercase filter effect
• Not removing stopwords correctly
• Confusing original text with tokens

4. Identify the error in this custom analyzer definition:

{
  "analysis": {
    "analyzer": {
      "my_analyzer": {
        "type": "custom",
        "tokenizer": "standard",
        "filter": "lowercase"
      }
    }
  }
}

medium

A. Missing stopwords definition

B. The tokenizer name is invalid

C. The filter should be an array, not a string

D. The analyzer type should be "standard"

Solution

1. Step 1: Check filter syntax in custom analyzer

   Filters must be provided as an array, e.g., ["lowercase"], not a string.

2. Step 2: Verify other parts

   Tokenizer "standard" and analyzer type "custom" are correct; stopwords are optional.

3. Final Answer:

   The filter should be an array, not a string -> Option C

4. Quick Check:

   Filter must be array syntax [OK]

Hint: Filters always use array brackets [] in custom analyzers [OK]

Common Mistakes:

• Using string instead of array for filters
• Changing tokenizer to invalid types
• Assuming stopwords are mandatory

5. You want to create a custom analyzer that removes English stopwords, lowercases text, and also applies a stemmer filter named english_stemmer. Which of the following is the correct way to define this analyzer in Elasticsearch settings?

hard

A. {
"analysis": {
"analyzer": {
"custom_english": {
"type": "standard",
"tokenizer": "standard",
"filter": ["lowercase", "stop", "english_stemmer"]
}
},
"filter": {
"stop": {
"type": "stop",
"stopwords": "_english_"
},
"english_stemmer": {
"type": "stemmer",
"language": "english"
}
}
}
}

B. {
"analysis": {
"analyzer": {
"custom_english": {
"type": "custom",
"tokenizer": "standard",
"filter": ["lowercase", "stop", "english_stemmer"]
}
},
"filter": {
"stop": {
"type": "stop",
"stopwords": "_english_"
},
"english_stemmer": {
"type": "stemmer",
"language": "english"
}
}
}
}

C. {
"analysis": {
"analyzer": {
"custom_english": {
"type": "custom",
"tokenizer": "standard",
"filter": ["stop", "lowercase", "english_stemmer"]
}
},
"filter": {
"stop": {
"type": "stop",
"stopwords": "_english_"
},
"english_stemmer": {
"type": "stemmer",
"language": "english"
}
}
}
}

D. {
"analysis": {
"analyzer": {
"custom_english": {
"type": "custom",
"tokenizer": "standard",
"filter": ["lowercase", "english_stemmer", "stop"]
}
},
"filter": {
"stop": {
"type": "stop",
"stopwords": "_english_"
},
"english_stemmer": {
"type": "stemmer",
"language": "english"
}
}
}
}

Solution

1. Step 1: Confirm analyzer type and tokenizer

   The analyzer must be "custom" with "standard" tokenizer to apply filters correctly.

2. Step 2: Check filter order and definitions

   Filters should be ["lowercase", "stop", "english_stemmer"] in that order to lowercase first, remove stopwords next, then stem.

3. Step 3: Verify filter definitions

   Stop filter uses "_english_" stopwords, and stemmer filter is correctly defined with language "english".

4. Final Answer:

   { "analysis": { "analyzer": { "custom_english": { "type": "custom", "tokenizer": "standard", "filter": ["lowercase", "stop", "english_stemmer"] } }, "filter": { "stop": { "type": "stop", "stopwords": "_english_" }, "english_stemmer": { "type": "stemmer", "language": "english" } } } } -> Option B

5. Quick Check:

   Custom analyzer with correct filter order and definitions = { "analysis": { "analyzer": { "custom_english": { "type": "custom", "tokenizer": "standard", "filter": ["lowercase", "stop", "english_stemmer"] } }, "filter": { "stop": { "type": "stop", "stopwords": "_english_" }, "english_stemmer": { "type": "stemmer", "language": "english" } } } } [OK]

Hint: Filter order matters: lowercase, stop, then stemmer [OK]

Common Mistakes:

• Using wrong analyzer type like 'standard'
• Incorrect filter order causing wrong results
• Missing or wrong stopwords definition