Bot led information extraction (using just simple Bot decision tree approach) - is it possible?

Rasa Open Source
1

The benefit of rasa is clearly that it can deal with somewhat unstructured responses and dynamic conversations. I’d like to develop a chatbot like this, but would prefer to use a standard questionnaire type format to begin with (and later add NLU etc). The bot would primarily be extracting information from the user in the first case (however I would like the user to be able to ask the bot for clarification etc in the future).

Is it possible to configure rasa to lead the conversation asking questions using a decision tree type approach - or am I looking at the wrong tool for my initial requirement?

If yes, any examples of this approach being employed would be welcomed

My use case is along the lines of…

  • (Bot speaks first) - what is your username?

    • (Human response) - (action check against database)

    • if authorised:

      • ​ ask Q1

    • if rejected

      • ask username again

    • Ask Q1

      • store response - if yes - ask Q1a , if no - ask Q2

      …

    • Goodbye

2

yes! you can do this with forms, see this tutorial Building contextual assistants with Rasa Forms and the docs Forms

I’d love to make this information easier to find. Were you searching the docs for a keyword that you didn’t find?

3

Thanks, seems obvious now! I was looking for questionnaire etc.

4

Hi @jaravo, @amn41

I have the similar use case like jaravo discussed above. I am a new RASA framework so I want to know from jaravo that did rasa worked for you? If yes then what are resources like blogs, articles you followed? any help will be really appreciated.

Thanks