Please enable JavaScript.

Coggle requires JavaScript to display documents.

Elasticsearch - Coggle Diagram

- - - - used with exact match, sorting and aggregation
        ( for example email field )
      - tokenize whole phrase
      - analyze with keyword analyzer ( no-op )
      - unmodified string as single token
    - - there is no array type in elasticsearch
      - all fields can contain zero or more values
      - simple supply array of values
      - array of text field are simply concatenated
      - array values should be the same data type
    - - as specifically formatted strings ( ISO 8601 by default )
        
        milliseconds since epoch
        
        seconds since epoch ( with epoch_seconds format mapping parameter )
      - stored as long values internally ( converted to UTC first )
      - converted to long values when searching too
      - do not provide UNIX timestamp for the date field without
        providing format mapping parameter epoch_seconds
        ( unix seconds vs milliseconds internally )
  - - - uinverted inverted index
      - used for sorting, aggregation and scripting
      - additional data structure not replacement for inverted index
      - doc_values = false -> save disk space, increase indexing speed
    - - disable indexing for the field
      - values still stored in _source
      - if false -> save disk space and indexing speed, but still use for aggregation
      - useful if field will not be used for searching
      - useful for time series fields that you will not use for searching but for aggregation
    - - cannot be indexed or search
      - using this parameter to replace NULL value with another value
      - only works for explicit NULL value
      - replacement data type must be the same type as the field
      - does not affect value stored in _source
    - - used to copy multiple fields value in a group field
      - e.g first_name + last_name -> full_name
      - values are copied not terms / tokens
      - analyzer of the target field is used for the values
      - the target field is not part of the _source
- - - - fixed by the time when index was created
    - - copy of primary shard
      - provide redundant copy to protect over failure
      - can be added to cluster at any time
  - - - all primary shards and all replica shards are active
    - - all primary shards are active,
        but not all replica shards are active
    - - not all primary shards are active
- - - - split text at words boundaries and remove punctuation
- - - - query type
        
        match
        
        wildcard
      - key to search ( doc field )
        
        object with search value and options
        
        search value directly
      - groups
        
        leaf query
        
        compound query
        ( bool query combining leaf queries etc. )
  - - - TF/IDF ( old )
        term frequency / inverse document frequency
        
        Term Frequency
        сколько раз токен пристутствует в документе
        ( чем больше раз токен присутствует в документе тем больше релевантность у документа )
        
        INVERSE DOCUMENT FREQUENCY
        сколько раз токен присутствует во все индексе ( по всем документам ).
        Чем больше это значение тем меньше релевантность.
        
        FIELD LENGTH NORM
        чем больше длина значения поля тем меньше релевантность ( релевантность будет больше у поля состоящего из 50 символов чем из поля состоящего из 5000 символов )
      - OKAPI BM25 ( new )
        
        Лучше в обработке stop words. Теперь не нужно добавлять stop words в analizer в большинстве случаев ( по этому они и не входят в стандартный analizer )
        
        NON LINEAR TERM FREQUENCY SATURATION
        
        Дебаг с _explain api