Japanese (kuromoji) Analysis for Elasticsearch

The Japanese (kuromoji) Analysis plugin integrates Lucene kuromoji analysis module into elasticsearch.

In order to install the plugin, run:

bin/plugin install elasticsearch/elasticsearch-analysis-kuromoji/2.5.0

You need to install a version matching your Elasticsearch version:

elasticsearch	Kuromoji Analysis Plugin	Docs
master	Build from source	See below
es-1.x	Build from source	2.6.0-SNAPSHOT
es-1.5	2.5.0	2.5.0
es-1.4	2.4.3	2.4.3
< 1.4.5	2.4.2	2.4.2
< 1.4.3	2.4.1	2.4.1
es-1.3	2.3.0	2.3.0
es-1.2	2.2.0	2.2.0
es-1.1	2.1.0	2.1.0
es-1.0	2.0.0	2.0.0
es-0.90	1.8.0	1.8.0

To build a SNAPSHOT version, you need to build it with Maven:

mvn clean install
plugin --install analysis-kuromoji \
       --url file:target/releases/elasticsearch-analysis-kuromoji-X.X.X-SNAPSHOT.zip

Includes Analyzer, Tokenizer, TokenFilter, CharFilter

The plugin includes these analyzer and tokenizer, tokenfilter.

name	type
kuromoji_iteration_mark	charfilter
kuromoji	analyzer
kuromoji_tokenizer	tokenizer
kuromoji_baseform	tokenfilter
kuromoji_part_of_speech	tokenfilter
kuromoji_readingform	tokenfilter
kuromoji_stemmer	tokenfilter
ja_stop	tokenfilter

Usage

Analyzer : kuromoji

An analyzer of type kuromoji. This analyzer is the following tokenizer and tokenfilter combination.

• kuromoji_tokenizer : Kuromoji Tokenizer
• kuromoji_baseform : Kuromoji BasicFormFilter (TokenFilter)
• kuromoji_part_of_speech : Kuromoji Part of Speech Stop Filter (TokenFilter)
• cjk_width : CJK Width Filter (TokenFilter)
• stop : Stop Filter (TokenFilter)
• kuromoji_stemmer : Kuromoji Katakana Stemmer Filter(TokenFilter)
• lowercase : LowerCase Filter (TokenFilter)

CharFilter : kuromoji_iteration_mark

A charfilter of type kuromoji_iteration_mark. This charfilter is Normalizes Japanese horizontal iteration marks (odoriji) to their expanded form.

The following ar setting that can be set for a kuromoji_iteration_mark charfilter type:

Setting	Description	Default value
normalize_kanji	indicates whether kanji iteration marks should be normalized	true
normalize_kana	indicates whether kanji iteration marks should be normalized	true

Tokenizer : kuromoji_tokenizer

A tokenizer of type kuromoji_tokenizer.

The following are settings that can be set for a kuromoji_tokenizer tokenizer type:

Setting	Description	Default value
mode	Tokenization mode: this determines how the tokenizer handles compound and unknown words. normal and search, extended	search
discard_punctuation	true if punctuation tokens should be dropped from the output.	true
user_dictionary	set User Dictionary file

Tokenization mode

The mode is three types.

• normal : Ordinary segmentation: no decomposition for compounds

• search : Segmentation geared towards search: this includes a decompounding process for long nouns, also including the full compound token as a synonym.

• extended : Extended mode outputs unigrams for unknown words.

Difference tokenization mode outputs

Input text is 関西国際空港 and アブラカダブラ.

mode	関西国際空港	アブラカダブラ
normal	関西国際空港	アブラカダブラ
search	関西 関西国際空港 国際 空港	アブラカダブラ
extended	関西 国際 空港	ア ブ ラ カ ダ ブ ラ

User Dictionary

Kuromoji tokenizer use MeCab-IPADIC dictionary by default. And Kuromoji is added an entry of dictionary to define by user; this is User Dictionary. User Dictionary entries are defined using the following CSV format:

<text>,<token 1> ... <token n>,<reading 1> ... <reading n>,<part-of-speech tag>

Dictionary Example

東京スカイツリー,東京 スカイツリー,トウキョウ スカイツリー,カスタム名詞

To use User Dictionary set file path to user_dict attribute. User Dictionary file is placed ES_HOME/config directory.

example

Example Settings:

curl -XPUT 'http://localhost:9200/kuromoji_sample/' -d'
{
    "settings": {
        "index":{
            "analysis":{
                "tokenizer" : {
                    "kuromoji_user_dict" : {
                       "type" : "kuromoji_tokenizer",
                       "mode" : "extended",
                       "discard_punctuation" : "false",
                       "user_dictionary" : "userdict_ja.txt"
                    }
                },
                "analyzer" : {
                    "my_analyzer" : {
                        "type" : "custom",
                        "tokenizer" : "kuromoji_user_dict"
                    }
                }

            }
        }
    }
}
'

Example Request using _analyze API :

curl -XPOST 'http://localhost:9200/kuromoji_sample/_analyze?analyzer=my_analyzer&pretty' -d '東京スカイツリー'

Response :

{
  "tokens" : [ {
    "token" : "東京",
    "start_offset" : 0,
    "end_offset" : 2,
    "type" : "word",
    "position" : 1
  }, {
    "token" : "スカイツリー",
    "start_offset" : 2,
    "end_offset" : 8,
    "type" : "word",
    "position" : 2
  } ]
}

TokenFilter : kuromoji_baseform

A token filter of type kuromoji_baseform that replaces term text with BaseFormAttribute. This acts as a lemmatizer for verbs and adjectives.

example

Example Settings:

curl -XPUT 'http://localhost:9200/kuromoji_sample/' -d'
{
    "settings": {
        "index":{
            "analysis":{
                "analyzer" : {
                    "my_analyzer" : {
                        "tokenizer" : "kuromoji_tokenizer",
                        "filter" : ["kuromoji_baseform"]
                    }
                }
            }
        }
    }
}
'

Example Request using _analyze API :

curl -XPOST 'http://localhost:9200/kuromoji_sample/_analyze?analyzer=my_analyzer&pretty' -d '飲み'

Response :

{
  "tokens" : [ {
    "token" : "飲む",
    "start_offset" : 0,
    "end_offset" : 2,
    "type" : "word",
    "position" : 1
  } ]
}

TokenFilter : kuromoji_part_of_speech

A token filter of type kuromoji_part_of_speech that removes tokens that match a set of part-of-speech tags.

The following are settings that can be set for a stop token filter type:

Setting	Description
stoptags	A list of part-of-speech tags that should be removed

Note that default setting is stoptags.txt include lucene-analyzer-kuromoji.jar.

example

Example Settings:

curl -XPUT 'http://localhost:9200/kuromoji_sample/' -d'
{
    "settings": {
        "index":{
            "analysis":{
                "analyzer" : {
                    "my_analyzer" : {
                        "tokenizer" : "kuromoji_tokenizer",
                        "filter" : ["my_posfilter"]
                    }
                },
                "filter" : {
                    "my_posfilter" : {
                        "type" : "kuromoji_part_of_speech",
                        "stoptags" : [
                            "助詞-格助詞-一般",
                            "助詞-終助詞"
                        ]
                    }
                }
            }
        }
    }
}
'

Example Request using _analyze API :

curl -XPOST 'http://localhost:9200/kuromoji_sample/_analyze?analyzer=my_analyzer&pretty' -d '寿司がおいしいね'

Response :

{
  "tokens" : [ {
    "token" : "寿司",
    "start_offset" : 0,
    "end_offset" : 2,
    "type" : "word",
    "position" : 1
  }, {
    "token" : "おいしい",
    "start_offset" : 3,
    "end_offset" : 7,
    "type" : "word",
    "position" : 3
  } ]
}

TokenFilter : kuromoji_readingform

A token filter of type kuromoji_readingform that replaces the term attribute with the reading of a token in either katakana or romaji form. The default reading form is katakana.

The following are settings that can be set for a kuromoji_readingform token filter type:

Setting	Description	Default value
use_romaji	true if romaji reading form output instead of katakana.	false

Note that elasticsearch-analysis-kuromoji built-in kuromoji_readingform set default true to use_romaji attribute.

example

Example Settings:

curl -XPUT 'http://localhost:9200/kuromoji_sample/' -d'
{
    "settings": {
        "index":{
            "analysis":{
                "analyzer" : {
                    "romaji_analyzer" : {
                        "tokenizer" : "kuromoji_tokenizer",
                        "filter" : ["romaji_readingform"]
                    },
                    "katakana_analyzer" : {
                        "tokenizer" : "kuromoji_tokenizer",
                        "filter" : ["katakana_readingform"]
                    }
                },
                "filter" : {
                    "romaji_readingform" : {
                        "type" : "kuromoji_readingform",
                        "use_romaji" : true
                    },
                    "katakana_readingform" : {
                        "type" : "kuromoji_readingform",
                        "use_romaji" : false
                    }
                }
            }
        }
    }
}
'

Example Request using _analyze API :

curl -XPOST 'http://localhost:9200/kuromoji_sample/_analyze?analyzer=katakana_analyzer&pretty' -d '寿司'

Response :

{
  "tokens" : [ {
    "token" : "スシ",
    "start_offset" : 0,
    "end_offset" : 2,
    "type" : "word",
    "position" : 1
  } ]
}

Example Request using _analyze API :

curl -XPOST 'http://localhost:9200/kuromoji_sample/_analyze?analyzer=romaji_analyzer&pretty' -d '寿司'

Response :

{
  "tokens" : [ {
    "token" : "sushi",
    "start_offset" : 0,
    "end_offset" : 2,
    "type" : "word",
    "position" : 1
  } ]
}

TokenFilter : kuromoji_stemmer

A token filter of type kuromoji_stemmer that normalizes common katakana spelling variations ending in a long sound character by removing this character (U+30FC). Only katakana words longer than a minimum length are stemmed (default is four).

Note that only full-width katakana characters are supported.

The following are settings that can be set for a kuromoji_stemmer token filter type:

Setting	Description	Default value
minimum_length	The minimum length to stem	4

example

Example Settings:

curl -XPUT 'http://localhost:9200/kuromoji_sample/' -d'
{
    "settings": {
        "index":{
            "analysis":{
                "analyzer" : {
                    "my_analyzer" : {
                        "tokenizer" : "kuromoji_tokenizer",
                        "filter" : ["my_katakana_stemmer"]
                    }
                },
                "filter" : {
                    "my_katakana_stemmer" : {
                        "type" : "kuromoji_stemmer",
                        "minimum_length" : 4
                    }
                }
            }
        }
    }
}
'

Example Request using _analyze API :

curl -XPOST 'http://localhost:9200/kuromoji_sample/_analyze?analyzer=my_analyzer&pretty' -d 'コピー'

Response :

{
  "tokens" : [ {
    "token" : "コピー",
    "start_offset" : 0,
    "end_offset" : 3,
    "type" : "word",
    "position" : 1
  } ]
}

Example Request using _analyze API :

curl -XPOST 'http://localhost:9200/kuromoji_sample/_analyze?analyzer=my_analyzer&pretty' -d 'サーバー'

Response :

{
  "tokens" : [ {
    "token" : "サーバ",
    "start_offset" : 0,
    "end_offset" : 4,
    "type" : "word",
    "position" : 1
  } ]
}

TokenFilter : ja_stop

A token filter of type ja_stop that provide a predefined "japanese" stop words. Note: It is only provide "japanese". If you want to use other predefined stop words, you can use stop token filter.

Example Settings:

example

curl -XPUT 'http://localhost:9200/kuromoji_sample/' -d'
{
    "settings": {
        "index":{
            "analysis":{
                "analyzer" : {
                    "analyzer_with_ja_stop" : {
                        "tokenizer" : "kuromoji_tokenizer",
                        "filter" : ["ja_stop"]
                    }
                },
                "filter" : {
                    "ja_stop" : {
                        "type" : "ja_stop",
                        "stopwords" : ["_japanese_", "ストップ"]
                    }
                }
            }
        }
    }
}'

Example Request using _analyze API :

curl -XPOST 'http://localhost:9200/kuromoji_sample/_analyze?analyzer=katakana_analyzer&pretty' -d 'ストップは消える'

Response :

{
  "tokens" : [ {
    "token" : "消える",
    "start_offset" : 5,
    "end_offset" : 8,
    "type" : "word",
    "position" : 3
  } ]
}

License

This software is licensed under the Apache 2 license, quoted below.

Copyright 2009-2014 Elasticsearch <http://www.elasticsearch.org>

Licensed under the Apache License, Version 2.0 (the "License"); you may not
use this file except in compliance with the License. You may obtain a copy of
the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
License for the specific language governing permissions and limitations under
the License.