Phrase suggester not capable of removing whitespace ("marktp latz" => "marktplatz")

Hi,

the phrase suggester does not seem to be capable of removing erroneously introduced whitespace.
E.g. marktp latz => gives no suggestions, while for marktp_latz => marktplatz actually is suggested.
The same for all cases where additional whitespace needs to be removed.

The term suggester however, is capable of suggesting marktp latz => marktplatz.

I played with all params of the phrase suggester without luck (confidence, max_errors, etc). Is it a limitation or am i doing something wrong?

index setup:

  {   
    settings: {
      analysis: {
        analyzer:   {   
          suggest: {
            type:      'custom',
            tokenizer: 'standard',
            filter: [   
              'lowercase',
              'shingle_filter'
            ]
          }
        },
        filter:     {   
          shingle_filter: {
          type:             'shingle',
          min_shingle_size: 2,
          max_shingle_size: 3
        }
      }
   }

mapping:

    ...
    suggestions:                {
      properties: {
        de: { type: 'text', analyzer: 'suggest' },
        en: { type: 'text', analyzer: 'suggest' },
        fr: { type: 'text', analyzer: 'suggest' }
      }
    }
   ...

query:

  "suggest": {
    "my_suggestions": {
      "text": "markt platz",
      "phrase": {
        "field": "suggestions.de"
      }
    }
  }
}

Thx in advance

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.