I want to make a custom component which I want to run at the very start of my pipeline which checks the spelling of the incoming message and corrects it. I am confused as to which DefaultV1Recipe.ComponentType to use. I saw in the code there is a NLUmessage convertor but I cannot use it. Can someone please guide me ?
In general, we don’t recommend doing this. I would consider including some common mis-spellings and type-o’s in the training data but the NLU should handle this anyway.
If you still want to do this, I would use one of the tokenizers to start.