Entity handling by intentizer

With Replace variable values with entity types during intent classification enabled, the original phrases are subjected to entity extraction and expand the set of training phrases

Entity are not used to train the intention model
If the potential phrase is in the set of training phrases and the phrase subjected to intention classification matches this set, the result of the intention classification is a given intention class with a score of 1.0

Handling of training set phrases with Replace variable values with entity types during intent classification enabled:

Phrase	Entities	Result
“What will the weather be like tomorrow?”	tomorrow -> sys.Time:time	“what will the weather be like sys.Time:time ?”
“Cancellation of an appointment with an internist”	appointment -> appointments:appointments internist -> doctor:doctor	“cancellation of an appointments:appointments with an doctor:doctor”
“07 February 2023”	07 February 2023 -> date:date	“date::date”

⚠️
Entities needs to marked on phrases in training set for this feature to work

Classification of intents

Action scheme:

If the incoming phrase is in the training set, return the result for the sample with the specified class with a score of 1.0. When Replace variable values with entity types during intent classification is enabled 1.0 is returned also with phrases that have same entity type and do not match exactly phrase in training set.
- Example: If phrase “What will the weather be like tomorrow?” is present in training set then user utterance “What will the weather be like today?” will also return score 1.0.
if the incoming phrase is in the cache, return the result stored in the cache
if the incoming phrase is not in the training set perform classification, return the result from the model and save the result in the cache