Skip to content

Latest commit

 

History

History
100 lines (51 loc) · 5.89 KB

tokenizer-reference.md

File metadata and controls

100 lines (51 loc) · 5.89 KB
mapped_pages

Tokenizer reference [analysis-tokenizers]

::::{admonition} Difference between {{es}} tokenization and neural tokenization :class: note

{{es}}'s tokenization process produces linguistic tokens, optimized for search and retrieval. This differs from neural tokenization in the context of machine learning and natural language processing. Neural tokenizers translate strings into smaller, subword tokens, which are encoded into vectors for consumptions by neural networks. {{es}} does not have built-in neural tokenizers.

::::

A tokenizer receives a stream of characters, breaks it up into individual tokens (usually individual words), and outputs a stream of tokens. For instance, a whitespace tokenizer breaks text into tokens whenever it sees any whitespace. It would convert the text "Quick brown fox!" into the terms [Quick, brown, fox!].

The tokenizer is also responsible for recording the following:

  • Order or position of each term (used for phrase and word proximity queries)
  • Start and end character offsets of the original word which the term represents (used for highlighting search snippets).
  • Token type, a classification of each term produced, such as <ALPHANUM>, <HANGUL>, or <NUM>. Simpler analyzers only produce the word token type.

Elasticsearch has a number of built in tokenizers which can be used to build custom analyzers.

Word Oriented Tokenizers [_word_oriented_tokenizers]

The following tokenizers are usually used for tokenizing full text into individual words:

Standard Tokenizer : The standard tokenizer divides text into terms on word boundaries, as defined by the Unicode Text Segmentation algorithm. It removes most punctuation symbols. It is the best choice for most languages.

Letter Tokenizer : The letter tokenizer divides text into terms whenever it encounters a character which is not a letter.

Lowercase Tokenizer : The lowercase tokenizer, like the letter tokenizer, divides text into terms whenever it encounters a character which is not a letter, but it also lowercases all terms.

Whitespace Tokenizer : The whitespace tokenizer divides text into terms whenever it encounters any whitespace character.

UAX URL Email Tokenizer : The uax_url_email tokenizer is like the standard tokenizer except that it recognises URLs and email addresses as single tokens.

Classic Tokenizer : The classic tokenizer is a grammar based tokenizer for the English Language.

Thai Tokenizer : The thai tokenizer segments Thai text into words.

Partial Word Tokenizers [_partial_word_tokenizers]

These tokenizers break up text or words into small fragments, for partial word matching:

N-Gram Tokenizer : The ngram tokenizer can break up text into words when it encounters any of a list of specified characters (e.g. whitespace or punctuation), then it returns n-grams of each word: a sliding window of continuous letters, e.g. quick → [qu, ui, ic, ck].

Edge N-Gram Tokenizer : The edge_ngram tokenizer can break up text into words when it encounters any of a list of specified characters (e.g. whitespace or punctuation), then it returns n-grams of each word which are anchored to the start of the word, e.g. quick → [q, qu, qui, quic, quick].

Structured Text Tokenizers [_structured_text_tokenizers]

The following tokenizers are usually used with structured text like identifiers, email addresses, zip codes, and paths, rather than with full text:

Keyword Tokenizer : The keyword tokenizer is a noop tokenizer that accepts whatever text it is given and outputs the exact same text as a single term. It can be combined with token filters like lowercase to normalise the analysed terms.

Pattern Tokenizer : The pattern tokenizer uses a regular expression to either split text into terms whenever it matches a word separator, or to capture matching text as terms.

Simple Pattern Tokenizer : The simple_pattern tokenizer uses a regular expression to capture matching text as terms. It uses a restricted subset of regular expression features and is generally faster than the pattern tokenizer.

Char Group Tokenizer : The char_group tokenizer is configurable through sets of characters to split on, which is usually less expensive than running regular expressions.

Simple Pattern Split Tokenizer : The simple_pattern_split tokenizer uses the same restricted regular expression subset as the simple_pattern tokenizer, but splits the input at matches rather than returning the matches as terms.

Path Tokenizer : The path_hierarchy tokenizer takes a hierarchical value like a filesystem path, splits on the path separator, and emits a term for each component in the tree, e.g. /foo/bar/baz → [/foo, /foo/bar, /foo/bar/baz ].