NLU: look-up tables how to handle aliases/spelling mistakes

Rasa Open Source
2

either add spelling mistakes to your training data, or write a component to recognize them. Also you could try char level count vectorizer