Featurizader for ML models

Rasa Open Source
1

Hello everyone

I have a couple of questions about how RASA transforms user input into an object that can then be used in ML models, more specifically for dialog management. I know that the NLU is used to extract the intents, entities and slots, so that together with the previous action of the bot a binary vector can be created and the model trained. My questions are:

  1. But is it that simple? no extra information is added to the array of slots or entities.
  2. What happens when we have more than one action for the same intent in a story, how does the vector change for the model, only the previous action changes?

This information I have taken from the following video: https://www.youtube.com/watch?v=j90NvurJI4I&ab_channel=Rasa

thanks for your help!