ICU tokenizer

Tokenizes text into words on word boundaries, as defined in UAX #29: Unicode Text Segmentation. It behaves much like the standard tokenizer, but adds better support for some Asian languages by using a dictionary-based approach to identify words in Thai, Lao, Chinese, Japanese, and Korean, and using custom rules to break Myanmar and Khmer text into syllables.

PUT icu_sample
{
  "settings": {
    "index": {
      "analysis": {
        "analyzer": {
          "my_icu_analyzer": {
            "tokenizer": "icu_tokenizer"
          }
        }
      }
    }
  }
}

Rules customization

edit

This functionality is marked as experimental in Lucene

You can customize the icu-tokenizer behavior by specifying per-script rule files, see the RBBI rules syntax reference for a more detailed explanation.

To add icu tokenizer rules, set the rule_files settings, which should contain a comma-separated list of code:rulefile pairs in the following format: four-letter ISO 15924 script code, followed by a colon, then a rule file name. Rule files are placed ES_HOME/config directory.

As a demonstration of how the rule files can be used, save the following user file to $ES_HOME/config/KeywordTokenizer.rbbi:

.+ {200};

Then create an analyzer to use this rule file as follows:

PUT icu_sample
{
  "settings": {
    "index": {
      "analysis": {
        "tokenizer": {
          "icu_user_file": {
            "type": "icu_tokenizer",
            "rule_files": "Latn:KeywordTokenizer.rbbi"
          }
        },
        "analyzer": {
          "my_analyzer": {
            "type": "custom",
            "tokenizer": "icu_user_file"
          }
        }
      }
    }
  }
}

GET icu_sample/_analyze
{
  "analyzer": "my_analyzer",
  "text": "Elasticsearch. Wow!"
}

The above analyze request returns the following:

{
   "tokens": [
      {
         "token": "Elasticsearch. Wow!",
         "start_offset": 0,
         "end_offset": 19,
         "type": "<ALPHANUM>",
         "position": 0
      }
   ]
}

« ICU normalization character filter ICU normalization token filter »