Class: Fuzzy

Bravey.Nlp.Fuzzy

new Fuzzy()

A version of the Natural Language Processing core. Doesn't follow entity order but tries to guess names by samples. Less precise but easier hits.
Source:

Methods

addDocument(text, intent) → {boolean}

Add a new document.
Parameters:
Name Type Attributes Default Description
text string The document content.
intent string The related intent.
guess.fromFullSentence boolean <optional>
false Indicates that the document is a full untagged sentence. (i.e. "Please call the 333-123456")
guess.fromTaggedSentence boolean <optional>
true Indicates that the document is tagged with braces. (i.e "Please call the {phone_number}")
guess.expandIntent boolean <optional>
false Extends or creates a new intent if exceeded entities are found. New entities id will be autogenerated and progressive.
guess.withNames Array.<string> <optional>
[] Uses the given names for auto-extending intents. Positions in array are matched with positions of entity into the specified sentence.
Source:
Returns:
True when found.
Type
boolean

addEntity(entity) → {boolean}

Adds an entity recognizer.
Parameters:
Name Type Description
entity EntityRecognizer The entity recognizer to be addded.
Source:
Returns:
True when successful.
Type
boolean

addIntent(intentName, entities) → {boolean}

Adds an intent.
Parameters:
Name Type Description
intentName string The name of the new intent.
entities Array.<IntentEntity> The produced entities.
Source:
Returns:
True when successful.
Type
boolean

getConfidence() → {number}

Returns the confidence ratio.
Source:
Returns:
The confidence ratio.
Type
number

hasEntity(entityName) → {boolean}

Check if an entity with a given name exists.
Parameters:
Name Type Description
entityName string The entity name.
Source:
Returns:
True when found.
Type
boolean

setConfidence(ratio)

Set confidence ratio, from 0.5 to 1. The higher the more strict.
Parameters:
Name Type Description
ratio number The confidence ratio to be set.
Source:

test(text, methodopt) → {NlpResult|false}

Check if a given sentence matches an intent and extracts its entities.
Parameters:
Name Type Attributes Default Description
text string The sentence to be processed.
method string <optional>
"default" The extraction method. "anyEntity" extracts all found entities and guess intent, regardless the intents structure.
Source:
Returns:
  • When an intent is found.
    Type
    NlpResult
  • When the sentence doesn't match any intent.
    Type
    false