Good day,
I want to create a chatbot that partially directs conversations over a graph. That is: the response generation should be created over the graph. Here is the plan step by step 0. user makes an input
-
the intent and the entities are captured via the NLU.
-
the intent and the entities are passed to the graph.
a. The graph captures the intent and the entities. b. The graph checks if entities are missing and asks queries (like a form, but the additional entities are not a must). c. The graph passes the appropriate answer to the given entities and the intent (dispatcher.utter_message)
-
the user receives an answer (possibly via queries)
Idea: To realize step 2, the Rasa process must be stopped until the graph has created the answer. The graph should take over the conversation during this time.
Is there any way to realize this? I mean that you can stop the Rasa process and then pass the conversation to an external program. Unfortunately I can’t find the source anymore.
Can you help me further?