Search code examples
elasticsearchelasticsearch-dsl

Fuzzy Multi Match not working as expected Elastic Search


I am trying to write query for search as you type with fuzziness added.(elastic search 7.12)

{
"query": {
    "multi_match": {
        "query": "airl recl",
        "fields": [
            "tags",
            "display_text",
            "display_subtext"
        ],
        "type" : "most_fields",
        "operator": "and",
        "fuzziness": "AUTO:4,6",
        "prefix_length" :2 
    }
}

}

I have inserted docs with "airtel recharge" values. I am also using edge n gram(1:50) for above given 3 fields along with space analyzer.

  1. If i search with airl -> it works fine, getting result with airtel keyword.
  2. If i search with recl -> it works fine, getting result with recharge keyword.
  3. But when i search with "airl recl" in query, not getting any result.

space analyzer :

"words_with_spaces_analyzer" : {
          "filter" : [
            "lowercase",
            "asciifolding"
          ],
          "type" : "custom",
          "tokenizer" : "words_with_space"
        }
      },
      "tokenizer" : {
        "words_with_space" : {
          "pattern" : "([a-zA-Z0-9.-]+[\\s]*)",
          "type" : "pattern",
          "group" : "0"
        }
      }
    },

Mapping

   "display_text": {
                "type": "text",
                "fields": {
                    "raw": {
                        "type": "keyword"
                    }
                },
                "analyzer": "edge_nGram_analyzer",
                "search_analyzer": "words_with_spaces_analyzer"
            }

Can someone please help me in understanding why above given query behaves this way for multi token input whereas both the token giving output if run them separately ?


Solution

  • You need to use a whitespace analyzer, as a search_analyzer. This will break the search term "airl recl" into airl and recl. And, then will perform the search on these individual tokens.

    Adding a working example with index mapping, search query, and search result

    Index Mapping:

     {
      "settings": {
        "analysis": {
          "analyzer": {
            "my_analyzer": {
              "tokenizer": "my_tokenizer"
            }
          },
          "tokenizer": {
            "my_tokenizer": {
              "type": "edge_ngram",
              "min_gram": 2,
              "max_gram": 50,
              "token_chars": [
                "letter",
                "digit"
              ]
            }
          }
        },
        "max_ngram_diff": 50
      },
      "mappings": {
        "properties": {
          "name": {
            "type": "text",
            "analyzer": "my_analyzer",
            "search_analyzer": "whitespace"
          }
        }
      }
    }
    

    Index Data:

    {
      "name": "airtel recharge"
    }
    

    Search Query:

    {
      "query": {
        "multi_match": {
          "query": "airl recl",
          "fields": [
            "name"
          ],
          "type": "most_fields",
          "operator": "and",
          "fuzziness": "AUTO:4,6",
          "prefix_length": 2
        }
      }
    }
    

    Search Result:

    "hits": [
          {
            "_index": "67702617",
            "_type": "_doc",
            "_id": "1",
            "_score": 0.22729424,
            "_source": {
              "name": "airtel recharge"
            }
          }
        ]