Partial matching against dates

[snoopy20 (Andrew)]

I'm sure this is is easy but couldn't find a simple answer.

If I have 'XXXX2000' in an index and I search '01022000' how can I accept that as a partial match, whereby the more of the word that matches the higher the score.

For example there could also be 01XX2000 stored, so here the match would be higher.

[dadoonet (David Pilato)]

If you use ngrams to analyze your text, you will produce a lot of sub tokens. The more sub tokens match, the higher the score will be.

My 2 cents.

[snoopy20 (Andrew)]

ngrams ... googling!

[snoopy20 (Andrew)]

Hi David,

Quick eyes over, does this look correct (ngram from 1-8)?

      "settings": {
        "index": {
          "max_ngram_diff": 50
        },
        "analysis": {
          "analyzer": {
            "my_analyzer": {
              "filter": [
              ],
              "type": "custom",
              "tokenizer": "my_tokenizer"
            }
          },
          "tokenizer": {
            "my_tokenizer": {
              "token_chars": [
                "letter",
                "digit"
              ],
              "min_gram": 1,
              "type": "ngram",
              "max_gram": 8
            }
          }
        }
      },
      "mappings": {
        "properties": {
          "dob": {
            "type": "text",
            "analyzer": "my_analyzer",
            "fields": {
              "keyword": {
                "type": "keyword"
              }
            }
          }
        }
      }
    }

[snoopy20 (Andrew)]

Tested and is working. Thank you!

[snoopy20 (Andrew)]

Hello again,

Requirements change means I need to revisit this. The date in question I can split onto the three indexes or combine. With this in mind, what is the best way to do:

DOB_full = "10102000" or DOB_YearMonth = "102000" or DOB_Year = "2000"

and for a bonus point a means to weigh/boost them differently.

[snoopy20 (Andrew)]

One more, this is along the lines of what I'm after, however the syntax is incorrect with

reason":"[match] unknown token [START_OBJECT] after [match_all]

should.push({
  match: {
    query: {
      match_all: {}
    },
    functions: [
      {
        filter: {
          match: { fields:['dob1'], query }
        },
        weight: 60
      },
      {
        filter: {
          simple_query_string: { fields:['dob1'], query: '**'+query.slice(2) }
        },
        weight: 40
      },
      {
        filter: {
          simple_query_string: { fields:['dob1'], query: '****'+query.slice(4) }
        },
        weight: 20
      }
    ],
    score_mode: "max",
  }
});

[snoopy20 (Andrew)]

Sorted I think.

function_score: {
  query: { match_all: {} },
  functions: [
    {
      filter: {
        match: {
          dob1: query
        }
      },
      weight: 1.1
    },
    {
      filter: {
        simple_query_string: {
          fields: ['dob1'],
          query: '**'+query.slice(2)
        }
      },
      weight: 1
    },
    {
      filter: {
        simple_query_string: {
          fields: ['dob1'],
          query: '****'+query.slice(4)
        }
      },
      weight: 0.85
    }
  ],
  score_mode: "max",
}