Elasticsearch tokenizer filters
Elasticsearch provides various tokenizer filters that can be used to modify the tokenization process of text during indexing and searching. These filters are applied to individual tokens generated by tokenizers and can transform or filter the tokens based on specific criteria. Here are some commonly used Elasticsearch tokenizer filters:
Lowercase Filter: Converts tokens to lowercase. Useful for case-insensitive searches.
Uppercase Filter: Converts tokens to uppercase.
ASCII Folding Filter: Replaces non-ASCII characters with their ASCII equivalents. For example, "é" becomes "e".
Stop Filter: Removes common words (stop words) from the token stream. Stop words are typically frequently occurring words that add little meaning to the search, such as "a," "an," "the," etc.
Stemmer Filter: Applies stemming algorithms to reduce words to their root form. For example, "running," "runs," and "ran" would all be stemmed to "run."
Synonym Filter: Replaces tokens with their synonyms based on a configured synonym dictionary. Useful for expanding the search scope to include similar terms.
Word Delimiter Filter: Splits tokens into subwords and applies various rules like splitting on camel case, punctuation, or numeric changes. For example, "HelloWorld" can be split into "Hello" and "World."
Length Filter: Filters out tokens based on their length. Tokens shorter or longer than specified thresholds can be excluded.
Pattern Replace Filter: Replaces tokens that match a specified regular expression pattern with a replacement string.
Phonetic Token Filters: Provide phonetic algorithms like Soundex or Metaphone to generate tokens based on their phonetic representation. This allows for approximate matching based on pronunciation.
These are just a few examples, and Elasticsearch provides many more token filters that can be customized and combined to suit specific requirements. Token filters are typically used in conjunction with tokenizers to control the indexing and search behavior of text in Elasticsearch.
댓글
댓글 쓰기