Elasticsearchelasticsearchsearchfull-textfuzzy-matchingproduction-ready

Fuzzy Multi-Field Search

Elasticsearch query DSL for fuzzy full-text search across multiple fields with boosting, highlighting, suggestions, and custom analyzers.

{
  "settings": {
    "number_of_shards": 2,
    "number_of_replicas": 1,
    "analysis": {
      "analyzer": {
        "custom_search_analyzer": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": [
            "lowercase",
            "asciifolding",
            "synonym_filter",
            "edge_ngram_filter"
          ]
        },
        "custom_index_analyzer": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": [
            "lowercase",
            "asciifolding",
            "word_delimiter_graph"
          ]
        }
      },
      "filter": {
        "synonym_filter": {
          "type": "synonym",
          "synonyms": [
            "js, javascript",
            "ts, typescript",
            "react, reactjs",
            "k8s, kubernetes"
          ]
        },
        "edge_ngram_filter": {
          "type": "edge_ngram",
          "min_gram": 2,
          "max_gram": 15
        }
      }
    }
  },
  "mappings": {
    "properties": {
      "title": {
        "type": "text",
        "analyzer": "custom_index_analyzer",
        "search_analyzer": "custom_search_analyzer",
        "fields": {
          "exact": { "type": "keyword" },
          "suggest": { "type": "completion" }
        }
      },
      "description": {
        "type": "text",
        "analyzer": "custom_index_analyzer"
      },
      "content": {
        "type": "text",
        "analyzer": "custom_index_analyzer"
      },
      "tags": {
        "type": "keyword",
        "fields": {
          "text": { "type": "text" }
        }
      },
      "category": { "type": "keyword" },
      "author": {
        "type": "text",
        "fields": { "keyword": { "type": "keyword" } }
      },
      "created_at": { "type": "date" },
      "popularity_score": { "type": "float" }
    }
  }
}

title: "Fuzzy Multi-Field Search" description: "Elasticsearch query DSL for fuzzy full-text search across multiple fields with boosting, highlighting, suggestions, and custom analyzers." category: "Elasticsearch" language: "json" tags: ["elasticsearch", "search", "full-text", "fuzzy-matching", "production-ready"] createdAt: "2024-12-12" updatedAt: "2025-01-09"

Overview

A comprehensive Elasticsearch configuration for implementing fuzzy full-text search with typo tolerance, field boosting, autocomplete suggestions, and faceted navigation. This setup handles real-world search challenges like misspellings, synonyms, and relevance tuning across multiple content fields.

Key design decisions:

  • Fuzziness: AUTO adjusts edit distance based on term length (0 for 1-2 chars, 1 for 3-5, 2 for 6+)
  • Field boosting prioritizes title matches over body content for better relevance
  • Edge n-grams enable prefix matching for type-ahead/autocomplete functionality
  • Synonyms expand queries to match related terms without requiring exact matches
  • Phrase matching with slop rewards results where search terms appear close together
  • Completion suggester provides fast autocomplete with fuzzy tolerance

Code

{
  "settings": {
    "number_of_shards": 2,
    "number_of_replicas": 1,
    "analysis": {
      "analyzer": {
        "custom_search_analyzer": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase", "asciifolding", "synonym_filter", "edge_ngram_filter"]
        },
        "custom_index_analyzer": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase", "asciifolding", "word_delimiter_graph"]
        }
      },
      "filter": {
        "synonym_filter": {
          "type": "synonym",
          "synonyms": ["js, javascript", "ts, typescript", "k8s, kubernetes"]
        },
        "edge_ngram_filter": {
          "type": "edge_ngram",
          "min_gram": 2,
          "max_gram": 15
        }
      }
    }
  },
  "mappings": {
    "properties": {
      "title": {
        "type": "text",
        "analyzer": "custom_index_analyzer",
        "search_analyzer": "custom_search_analyzer",
        "fields": {
          "exact": { "type": "keyword" },
          "suggest": { "type": "completion" }
        }
      },
      "description": { "type": "text", "analyzer": "custom_index_analyzer" },
      "content": { "type": "text", "analyzer": "custom_index_analyzer" },
      "tags": {
        "type": "keyword",
        "fields": { "text": { "type": "text" } }
      },
      "category": { "type": "keyword" },
      "created_at": { "type": "date" },
      "popularity_score": { "type": "float" }
    }
  }
}

Usage

{
  "query": {
    "bool": {
      "must": [{
        "multi_match": {
          "query": "kuberntes deploymnt",
          "type": "best_fields",
          "fields": ["title^3", "description^2", "content", "tags.text^2.5"],
          "fuzziness": "AUTO",
          "prefix_length": 2,
          "max_expansions": 50,
          "minimum_should_match": "75%"
        }
      }],
      "should": [
        { "match_phrase": { "title": { "query": "kuberntes deploymnt", "boost": 5, "slop": 2 } } },
        { "term": { "tags": { "value": "production-ready", "boost": 1.5 } } }
      ],
      "filter": [
        { "range": { "created_at": { "gte": "now-1y" } } }
      ]
    }
  },
  "highlight": {
    "pre_tags": ["<mark>"],
    "post_tags": ["</mark>"],
    "fields": {
      "title": { "number_of_fragments": 0 },
      "description": { "fragment_size": 150, "number_of_fragments": 3 }
    }
  },
  "suggest": {
    "did_you_mean": {
      "text": "kuberntes deploymnt",
      "phrase": {
        "field": "title",
        "size": 3,
        "direct_generator": [{ "field": "title", "suggest_mode": "popular" }]
      }
    }
  },
  "size": 20,
  "from": 0
}

Notes

  • Set prefix_length: 2 to avoid expensive fuzzy expansions on the first characters of every term
  • Use search_analyzer (with synonyms) separately from index_analyzer to keep the index lean
  • The completion field type is stored in an FST (finite state transducer) for sub-millisecond autocomplete
  • Consider function_score for more sophisticated relevance tuning (recency decay, popularity boosting)
  • Edge n-gram filters expand index size; monitor shard size and consider using search_as_you_type field type as an alternative
  • Always use filter context for non-scoring clauses (dates, categories) to leverage query caching
  • For multilingual content, use language-specific analyzers per field with multi-fields mapping