Thank you so much for your response and I am really sorry for late reply.
Yes as of now we are planning to provide suggestion to user based on his/her input in chatbot widget, to get those suggestion we have created a vocabulary from our training data, lookup table and synonyms table. Is that approach is correct ?
In-case, If we need to do the same in RASA side, I am thinking to implement the same using fuzzywuzzy or levenshtein,
This will be based on some threshold value and might give very good result in some case or completely random result in some case, imagine a scenario if a entity value has multiple words in it’s values.