Hyphenation decompounder token filter

Uses XML-based hyphenation patterns to find potential subwords in compound words. These subwords are then checked against the specified word list. Subwords not in the list are excluded from the token output.

This filter uses Lucene’s HyphenationCompoundWordTokenFilter, which was built for Germanic languages.

Example

The following analyze API request uses the hyphenation_decompounder filter to find subwords in Kaffeetasse based on German hyphenation patterns in the analysis/hyphenation_patterns.xml file. The filter then checks these subwords against a list of specified words: kaffee, zucker, and tasse.

GET _analyze
{
  "tokenizer": "standard",
  "filter": [
    {
      "type": "hyphenation_decompounder",
      "hyphenation_patterns_path": "analysis/hyphenation_patterns.xml",
      "word_list": ["Kaffee", "zucker", "tasse"]
    }
  ],
  "text": "Kaffeetasse"
}

The filter produces the following tokens:

[ Kaffeetasse, Kaffee, tasse ]

Configurable parameters

hyphenation_patterns_path

(Required, string) Path to an Apache FOP (Formatting Objects Processor) XML hyphenation pattern file.

This path must be absolute or relative to the config location. Only FOP v1.2 compatible files are supported.

For example FOP XML hyphenation pattern files, refer to:

word_list

(Required*, array of strings) A list of subwords. Subwords found using the hyphenation pattern but not in this list are excluded from the token output.

You can use the dictionary_decompounder filter to test the quality of word lists before implementing them.

Either this parameter or word_list_path must be specified.

word_list_path

(Required*, string) Path to a file containing a list of subwords. Subwords found using the hyphenation pattern but not in this list are excluded from the token output.

This path must be absolute or relative to the config location, and the file must be UTF-8 encoded. Each token in the file must be separated by a line break.

You can use the dictionary_decompounder filter to test the quality of word lists before implementing them.

Either this parameter or word_list must be specified.

max_subword_size
(Optional, integer) Maximum subword character length. Longer subword tokens are excluded from the output. Defaults to 15.
min_subword_size
(Optional, integer) Minimum subword character length. Shorter subword tokens are excluded from the output. Defaults to 2.
min_word_size
(Optional, integer) Minimum word character length. Shorter word tokens are excluded from the output. Defaults to 5.
only_longest_match
(Optional, boolean) If true, only include the longest matching subword. Defaults to false.

Customize and add to an analyzer

To customize the hyphenation_decompounder filter, duplicate it to create the basis for a new custom token filter. You can modify the filter using its configurable parameters.

For example, the following create index API request uses a custom hyphenation_decompounder filter to configure a new custom analyzer.

The custom hyphenation_decompounder filter find subwords based on hyphenation patterns in the analysis/hyphenation_patterns.xml file. The filter then checks these subwords against the list of words specified in the analysis/example_word_list.txt file. Subwords longer than 22 characters are excluded from the token output.

PUT hyphenation_decompound_example
{
  "settings": {
    "analysis": {
      "analyzer": {
        "standard_hyphenation_decompound": {
          "tokenizer": "standard",
          "filter": [ "22_char_hyphenation_decompound" ]
        }
      },
      "filter": {
        "22_char_hyphenation_decompound": {
          "type": "hyphenation_decompounder",
          "word_list_path": "analysis/example_word_list.txt",
          "hyphenation_patterns_path": "analysis/hyphenation_patterns.xml",
          "max_subword_size": 22
        }
      }
    }
  }
}