-
Notifications
You must be signed in to change notification settings - Fork 473
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge branch 'main' into containerize-doc-building
- Loading branch information
Showing
109 changed files
with
3,642 additions
and
346 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,160 @@ | ||
| --- | ||
| layout: default | ||
| title: CJK bigram | ||
| parent: Token filters | ||
| nav_order: 30 | ||
| --- | ||
|
|
||
| # CJK bigram token filter | ||
|
|
||
| The `cjk_bigram` token filter is designed specifically for processing East Asian languages, such as Chinese, Japanese, and Korean (CJK), which typically don't use spaces to separate words. A bigram is a sequence of two adjacent elements in a string of tokens, which can be characters or words. For CJK languages, bigrams help approximate word boundaries and capture significant character pairs that can convey meaning. | ||
|
|
||
|
|
||
| ## Parameters | ||
|
|
||
| The `cjk_bigram` token filter can be configured with two parameters: `ignore_scripts`and `output_unigrams`. | ||
|
|
||
| ### `ignore_scripts` | ||
|
|
||
| The `cjk-bigram` token filter ignores all non-CJK scripts (writing systems like Latin or Cyrillic) and tokenizes only CJK text into bigrams. Use this option to specify CJK scripts to be ignored. This option takes the following valid values: | ||
|
|
||
| - `han`: The `han` script processes Han characters. [Han characters](https://simple.wikipedia.org/wiki/Chinese_characters) are logograms used in the written languages of China, Japan, and Korea. The filter can help with text processing tasks like tokenizing, normalizing, or stemming text written in Chinese, Japanese kanji, or Korean Hanja. | ||
|
|
||
| - `hangul`: The `hangul` script processes Hangul characters, which are unique to the Korean language and do not exist in other East Asian scripts. | ||
|
|
||
| - `hiragana`: The `hiragana` script processes hiragana, one of the two syllabaries used in the Japanese writing system. | ||
| Hiragana is typically used for native Japanese words, grammatical elements, and certain forms of punctuation. | ||
|
|
||
| - `katakana`: The `katakana` script processes katakana, the other Japanese syllabary. | ||
| Katakana is mainly used for foreign loanwords, onomatopoeia, scientific names, and certain Japanese words. | ||
|
|
||
|
|
||
| ### `output_unigrams` | ||
|
|
||
| This option, when set to `true`, outputs both unigrams (single characters) and bigrams. Default is `false`. | ||
|
|
||
| ## Example | ||
|
|
||
| The following example request creates a new index named `devanagari_example_index` and defines an analyzer with the `cjk_bigram_filter` filter and `ignored_scripts` parameter set to `katakana`: | ||
|
|
||
| ```json | ||
| PUT /cjk_bigram_example | ||
| { | ||
| "settings": { | ||
| "analysis": { | ||
| "analyzer": { | ||
| "cjk_bigrams_no_katakana": { | ||
| "tokenizer": "standard", | ||
| "filter": [ "cjk_bigrams_no_katakana_filter" ] | ||
| } | ||
| }, | ||
| "filter": { | ||
| "cjk_bigrams_no_katakana_filter": { | ||
| "type": "cjk_bigram", | ||
| "ignored_scripts": [ | ||
| "katakana" | ||
| ], | ||
| "output_unigrams": true | ||
| } | ||
| } | ||
| } | ||
| } | ||
| } | ||
| ``` | ||
| {% include copy-curl.html %} | ||
|
|
||
| ## Generated tokens | ||
|
|
||
| Use the following request to examine the tokens generated using the analyzer: | ||
|
|
||
| ```json | ||
| POST /cjk_bigram_example/_analyze | ||
| { | ||
| "analyzer": "cjk_bigrams_no_katakana", | ||
| "text": "東京タワーに行く" | ||
| } | ||
| ``` | ||
| {% include copy-curl.html %} | ||
|
|
||
| Sample text: "東京タワーに行く" | ||
|
|
||
| 東京 (Kanji for "Tokyo") | ||
| タワー (Katakana for "Tower") | ||
| に行く (Hiragana and Kanji for "go to") | ||
|
|
||
| The response contains the generated tokens: | ||
|
|
||
| ```json | ||
| { | ||
| "tokens": [ | ||
| { | ||
| "token": "東", | ||
| "start_offset": 0, | ||
| "end_offset": 1, | ||
| "type": "<SINGLE>", | ||
| "position": 0 | ||
| }, | ||
| { | ||
| "token": "東京", | ||
| "start_offset": 0, | ||
| "end_offset": 2, | ||
| "type": "<DOUBLE>", | ||
| "position": 0, | ||
| "positionLength": 2 | ||
| }, | ||
| { | ||
| "token": "京", | ||
| "start_offset": 1, | ||
| "end_offset": 2, | ||
| "type": "<SINGLE>", | ||
| "position": 1 | ||
| }, | ||
| { | ||
| "token": "タワー", | ||
| "start_offset": 2, | ||
| "end_offset": 5, | ||
| "type": "<KATAKANA>", | ||
| "position": 2 | ||
| }, | ||
| { | ||
| "token": "に", | ||
| "start_offset": 5, | ||
| "end_offset": 6, | ||
| "type": "<SINGLE>", | ||
| "position": 3 | ||
| }, | ||
| { | ||
| "token": "に行", | ||
| "start_offset": 5, | ||
| "end_offset": 7, | ||
| "type": "<DOUBLE>", | ||
| "position": 3, | ||
| "positionLength": 2 | ||
| }, | ||
| { | ||
| "token": "行", | ||
| "start_offset": 6, | ||
| "end_offset": 7, | ||
| "type": "<SINGLE>", | ||
| "position": 4 | ||
| }, | ||
| { | ||
| "token": "行く", | ||
| "start_offset": 6, | ||
| "end_offset": 8, | ||
| "type": "<DOUBLE>", | ||
| "position": 4, | ||
| "positionLength": 2 | ||
| }, | ||
| { | ||
| "token": "く", | ||
| "start_offset": 7, | ||
| "end_offset": 8, | ||
| "type": "<SINGLE>", | ||
| "position": 5 | ||
| } | ||
| ] | ||
| } | ||
| ``` | ||
|
|
||
|
|
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,96 @@ | ||
| --- | ||
| layout: default | ||
| title: CJK width | ||
| parent: Token filters | ||
| nav_order: 40 | ||
| --- | ||
|
|
||
| # CJK width token filter | ||
|
|
||
| The `cjk_width` token filter normalizes Chinese, Japanese, and Korean (CJK) tokens by converting full-width ASCII characters to their standard (half-width) ASCII equivalents and half-width katakana characters to their full-width equivalents. | ||
|
|
||
| ### Converting full-width ASCII characters | ||
|
|
||
| In CJK texts, ASCII characters (such as letters and numbers) can appear in full-width form, occupying the space of two half-width characters. Full-width ASCII characters are typically used in East Asian typography for alignment with the width of CJK characters. However, for the purposes of indexing and searching, these full-width characters need to be normalized to their standard (half-width) ASCII equivalents. | ||
|
|
||
| The following example illustrates ASCII character normalization: | ||
|
|
||
| ``` | ||
| Full-Width: ＡＢＣＤＥ １２３４５ | ||
| Normalized (half-width): ABCDE 12345 | ||
| ``` | ||
|
|
||
| ### Converting half-width katakana characters | ||
|
|
||
| The `cjk_width` token filter converts half-width katakana characters to their full-width counterparts, which are the standard form used in Japanese text. This normalization, illustrated in the following example, is important for consistency in text processing and searching: | ||
|
|
||
|
|
||
| ``` | ||
| Half-Width katakana: ｶﾀｶﾅ | ||
| Normalized (full-width) katakana: カタカナ | ||
| ``` | ||
|
|
||
| ## Example | ||
|
|
||
| The following example request creates a new index named `cjk_width_example_index` and defines an analyzer with the `cjk_width` filter: | ||
|
|
||
| ```json | ||
| PUT /cjk_width_example_index | ||
| { | ||
| "settings": { | ||
| "analysis": { | ||
| "analyzer": { | ||
| "cjk_width_analyzer": { | ||
| "type": "custom", | ||
| "tokenizer": "standard", | ||
| "filter": ["cjk_width"] | ||
| } | ||
| } | ||
| } | ||
| } | ||
| } | ||
| ``` | ||
| {% include copy-curl.html %} | ||
|
|
||
| ## Generated tokens | ||
|
|
||
| Use the following request to examine the tokens generated using the analyzer: | ||
|
|
||
| ```json | ||
| POST /cjk_width_example_index/_analyze | ||
| { | ||
| "analyzer": "cjk_width_analyzer", | ||
| "text": "Ｔｏｋｙｏ 2024 ｶﾀｶﾅ" | ||
| } | ||
| ``` | ||
| {% include copy-curl.html %} | ||
|
|
||
| The response contains the generated tokens: | ||
|
|
||
| ```json | ||
| { | ||
| "tokens": [ | ||
| { | ||
| "token": "Tokyo", | ||
| "start_offset": 0, | ||
| "end_offset": 5, | ||
| "type": "<ALPHANUM>", | ||
| "position": 0 | ||
| }, | ||
| { | ||
| "token": "2024", | ||
| "start_offset": 6, | ||
| "end_offset": 10, | ||
| "type": "<NUM>", | ||
| "position": 1 | ||
| }, | ||
| { | ||
| "token": "カタカナ", | ||
| "start_offset": 11, | ||
| "end_offset": 15, | ||
| "type": "<KATAKANA>", | ||
| "position": 2 | ||
| } | ||
| ] | ||
| } | ||
| ``` |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.