How to search for a part of a word with ElasticSearch

How to Search for Parts of Words in ElasticSearch

In Elasticsearch, if you want to search for parts of words within documents, you can typically use several different methods. These techniques primarily leverage Elasticsearch's robust full-text search capabilities and its support for various analyzers. Here are some common methods:

1. Using wildcard Query

The wildcard query allows you to match parts of words using wildcards. For example, if you want to search for words containing the substring 'log' (e.g., 'biology', 'catalog', 'logistic', etc.), you can construct the following query:

json
{
  "query": {
    "wildcard": {
      "content": "*log*"
    }
  }
}

Here, content is the field name in the document, and *log* matches any word containing 'log'. The asterisk * is a wildcard representing any character sequence.

2. Using ngram Analyzer

To enable more flexible matching of parts of words during search, you can use the ngram analyzer when creating the index. The ngram analyzer splits words into multiple n-grams of specified lengths. For example, the word 'example' is split into ['ex', 'xa', 'am', 'mp', 'pl', 'le'].

Here's an example of creating an index with the ngram analyzer:

json
PUT /my_index
{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_ngram_analyzer": {
          "type": "custom",
          "tokenizer": "standard",
          "filter": ["lowercase", "my_ngram"]
        }
      },
      "filter": {
        "my_ngram": {
          "type": "nGram",
          "min_gram": 2,
          "max_gram": 3
        }
      }
    }
  },
  "mappings": {
    "properties": {
      "content": {
        "type": "text",
        "analyzer": "my_ngram_analyzer"
      }
    }
  }
}

With this analyzer, matching parts of words during search becomes more straightforward.

3. Using match_phrase Query

Although the match_phrase query is typically used for exact phrase matching, it can be adapted to search for parts of words within text by appropriately adjusting its parameters. This often involves combining it with the ngram analyzer or other tokenization approaches.

json
{
  "query": {
    "match_phrase": {
      "content": {
        "query": "part_of_word",
        "slop": 2
      }
    }
  }
}

These are just a few common methods; in practice, you can choose the appropriate method based on specific requirements and data characteristics. When using these query techniques, consider performance and index maintenance, and proper configuration and optimization are crucial in production environments.