Hello
Is it possible for the Rasa tokenizer, Count Vector Featureizer 1 and 2, Regex Entity extractor and generally all the Rasa components to read the features, tockens e.t.c from a knowledge bace i.e. TypeDB? If it is possible then how would I do that
TypeDB looks like a version of SparQL to create a knowledge base.
Why do you want to store nlu components in a knowledge base?
Just to have more control on the NLU and Natural Language Generation. Also NLU sees words as symbols. Some words, though, can be broken down to reveal deeper meanings. For example: automobile consists of the words auto, which means “self” and mobile, which means movement, implying that automobile is an object that has the property of being able to move by itself.
maybe this might be something more useful as a retrieval action. I think you can create an NLU component that connects to the knowledge for entity extractions.