Redressing Rasa interpreters for Other NLUs

ty
(Kevin Smith) #1

Good day all,

So I am looking at a conceptual sever in the Rasa stack to allow my app to utilize other NLUs to process intents while relying on Core to be the driver in a conversation. I’ve seen where google will let you use a custom action to use both NLU and Core. However, my project requires more than one channel of communication or the ability to field at least three types of requests (google, alexa json, and cortana). If this has been addressed somewhere else, I’d appreciate the link

Thanks all Kev

(Ty Dunn) #2

Hey @Kev! In order to use only rasa core, you will need to implement a subclass of Interpreter. This Interpreter subclass should have a parse method that takes the user message as input and outputs the intent and entities.

If you are referring to using rasa core with Google Dialogflow, check out this. Otherwise, if you are talking about Google Assistant, check out this tutorial here

(Kevin Smith) #3

Thanks Ty,

I’ve been working on getting my bot into a position to integrate with both Alexa and Google Assistant in order to reduce overhead while opening across both platforms. I’m struggling, probably unnecessarily with deployment of my bot. There seems to be a gap in the materials for those who are self taught.

Thanks again