Dictionary decompounder token filter

The dictionary_decompounder token filter is used to split compound words into their constituent parts based on a predefined dictionary. This filter is particularly useful for languages like German, Dutch, or Finnish, in which compound words are common, so breaking them down can improve search relevance. The dictionary_decompounder token filter determines whether each token (word) can be split into smaller tokens based on a list of known words. If the token can be split into known words, the filter generates the subtokens for the token.

Parameters

The dictionary_decompounder token filter has the following parameters.

ParameterRequired/OptionalData typeDescription
word_listRequired unless word_list_path is configuredArray of stringsThe dictionary of words that the filter uses to split compound words.
word_list_pathRequired unless word_list is configuredStringA file path to a text file containing the dictionary words. Accepts either an absolute path or a path relative to the config directory. The dictionary file must be UTF-8 encoded, and each word must be listed on a separate line.
min_word_sizeOptionalIntegerThe minimum length of the entire compound word that will be considered for splitting. If a compound word is shorter than this value, it is not split. Default is 5.
min_subword_sizeOptionalIntegerThe minimum length for any subword. If a subword is shorter than this value, it is not included in the output. Default is 2.
max_subword_sizeOptionalIntegerThe maximum length for any subword. If a subword is longer than this value, it is not included in the output. Default is 15.
only_longest_matchOptionalBooleanIf set to true, only the longest matching subword will be returned. Default is false.

Example

The following example request creates a new index named decompound_example and configures an analyzer with the dictionary_decompounder filter:

  1. PUT /decompound_example
  2. {
  3.   "settings": {
  4.     "analysis": {
  5.       "filter": {
  6.         "my_dictionary_decompounder": {
  7.           "type": "dictionary_decompounder",
  8.           "word_list": ["slow", "green", "turtle"]
  9.         }
  10.       },
  11.       "analyzer": {
  12.         "my_analyzer": {
  13.           "type": "custom",
  14.           "tokenizer": "standard",
  15.           "filter": ["lowercase", "my_dictionary_decompounder"]
  16.         }
  17.       }
  18.     }
  19.   }
  20. }

copy

Generated tokens

Use the following request to examine the tokens generated using the analyzer:

  1. POST /decompound_example/_analyze
  2. {
  3.   "analyzer": "my_analyzer",
  4.   "text": "slowgreenturtleswim"
  5. }

copy

The response contains the generated tokens:

  1. {
  2.   "tokens": [
  3.     {
  4.       "token": "slowgreenturtleswim",
  5.       "start_offset": 0,
  6.       "end_offset": 19,
  7.       "type": "<ALPHANUM>",
  8.       "position": 0
  9.     },
  10.     {
  11.       "token": "slow",
  12.       "start_offset": 0,
  13.       "end_offset": 19,
  14.       "type": "<ALPHANUM>",
  15.       "position": 0
  16.     },
  17.     {
  18.       "token": "green",
  19.       "start_offset": 0,
  20.       "end_offset": 19,
  21.       "type": "<ALPHANUM>",
  22.       "position": 0
  23.     },
  24.     {
  25.       "token": "turtle",
  26.       "start_offset": 0,
  27.       "end_offset": 19,
  28.       "type": "<ALPHANUM>",
  29.       "position": 0
  30.     }
  31.   ]
  32. }