Elasticsearch is not able to search "5600" when product name has "TW5600"

Question

My Product name: Epson EH TW5600 Full HD Home Cinema Projector | Lens Shift | 2500 Lumens | Bluetooth Audio

Search Term: "5600" or "Epson 5600"

In the above condition, my elasticsearch is not able to search "5600". It returns empty products.

I was using solr in my Magento 1 and it was working fine.

Any suggestions on this?

Answer 1

You know your data better than Elasticsearch. In order to retrieve great output you will have to plan your indices ahead.

    DELETE /products
    # create index containing one document
    PUT /products/_doc/1
    {
        "product": "Epson EH TW5600 Full HD Home Cinema Projector | Lens Shift | 2500 Lumens | Bluetooth Audio"
    }
    # let's search for a portion of a string-token
    GET products/_search
    {
      "query": {
        "match": {
          "product": "5600" 
        }
      }
    }
    # noting found!

    # how is our field mapped?
    GET /products/_mapping
    # no analyzer specified => standard is utilized.
    # standard tokenizer is used by default
    GET /_analyze
    {
      "tokenizer" : "standard",
      "text" : "Epson EH TW5600 Full HD Home Cinema Projector | Lens Shift | 2500 Lumens | Bluetooth Audio"
    }
    # there is no token for '5600' and there will be no hits.

    # The standard tokenizer will not do. We need a different tokenizer!
    # it's up to you to decide the tokens.
    GET /_analyze
    {
      "tokenizer": {
        "type": "ngram",
        "min_gram": 3,
        "max_gram": 3,
        "token_chars": [
          "letter",
          "digit"
        ]
      },
      "text": "Epson EH TW5600 Full HD Home Cinema Projector | Lens Shift | 2500 Lumens | Bluetooth Audio"
    }

    # we cannot change properties of an index if there are documents. Index has to go.
    DELETE /products
    # create index with custom analyzer and field-mapping
    PUT products
    {
      "settings": {
        "analysis": {
          "analyzer": {
            "my_analyzer": {
              "tokenizer": "my_tokenizer"
            }
          },
          "tokenizer": {
            "my_tokenizer": {
              "type": "ngram",
              "min_gram": 3,
              "max_gram": 3,
              "token_chars": [
                "letter",
                "digit"
              ]
            }
          }
        }
      },
      "mappings": {
        "_doc": {
          "properties": {
            "product": {
              "type": "text",
              "analyzer": "my_analyzer"
            }
          }
        }
      }
    }
    # create document
    PUT /products/_doc/1
    {
        "product": "Epson EH TW5600 Full HD Home Cinema Projector | Lens Shift | 2500 Lumens | Bluetooth Audio"
    }
    # how is our field mapped?
    GET /products/_mapping
    # 'my_analyzer' is utilized

    # let's search for a portion of a string-token
    GET products/_search
    {
      "query": {
        "match": {
          "product": "560" 
        }
      }
    }

    # document found!

    # anything containing '560' (3 letters) will be a hit.
    # '56' will not hit.
    # 'anything-before-560-anything-after' will be a hit!

    # tokens produced.
    POST products/_analyze
    {
      "analyzer": "my_analyzer",
      "text" : "Epson EH TW5600 Full HD Home Cinema Projector | Lens Shift | 2500 Lumens | Bluetooth Audio"
    }