E.g
In the above example, set_volume is a FormAction, and inform_volume is used to answer slot (volume). For this kind of form intent (set_volume) that can contain inform intent (inform_volume), when nlu classifies the intent, it will mistakenly classify the form intent as inform intent
Its not recommended.to have same utterances for different intents.
You can use rasa data validate to validate your training data.
How to solve this example
set_volume intent will trigger your form. one of the slot in your form is volume, and the response to utter_ask_volume will be inform_volume.
Mute, 40 and maximum are entity values, which will be extracted from the set_volume. If the slot and entity names are same then rasa fill automatically fill the slot for you. or you can specify them in the slot_mappings method of your FormAction.
I can understand what you mean, but there is a problem. Assuming that the value of volume entity is in the tens of thousands, will all entities be listed in the corpus of set_volume intent? If not, the intent classifier will misclassify entities that do not appear in the corpus.Does the data in the database serve as a training corpus for entity recognition?
In cases where you have lots of possible entity values you can make use of Lookup Tables for entities.
Can Lookup Tables use databases (such as mysql)?
I’m not sure. someone else might be able to help you in that regard @MseXing
Ok, thank you!