Does Rasa NLU support contextual or dialogue based intent prediction and slots filling?

How I encountered this problem:

  1. I want to use Rasa to build a inspector bot, rather than a chat-bot
  2. The inspector bot will analysis the text stream which is real talks between helpdesks and customer;
  3. Some bool slots are quite contextual sensitive. For example:
    • Help Desk: Is there a dealer near your apartment?
    • Customer: Yes!
    • Help Desk: Are you satisfied with our service?
    • Customer: Yes! The model will be confused by two YES and slots filling mission is impossible.

I am new to Rasa. I find all NLU pipelines and training data are based on human-chatbot. Is there a way to solve this problem?

Rasa is so great and I want to put this inspector bot into production.

For the example you gave, you can use Forms with from_intent slot mapping

Thanks, Hero! I will give it a shot. :star: @ChrisRahme

1 Like