How to add Fuzzy matching to the Entity Extraction?


Could some one please help me understand how we can add Fuzzy matching process while detecting the Entities? I read tutorials on NLU custom components, it says Fuzzy matching slows down the performance. But still could anyone explain how am I suppose to implement it? I tried creating my own custom NLU component but it didn’t work as expected also I’m not sure if it is correct in the first place.


If you want to add fuzzy matching to Rasa, custom components is the right way to got. Please, have a look at Custom NLU Components. It should help you getting started on custom components.

May I ask, what exactly your use case is?

I guess, you can find many tutorials online that explain how to do fuzzy matching. Maybe GitHub - seatgeek/fuzzywuzzy: Fuzzy String Matching in Python is something you can try out. If you need further help on how to set it up with Rasa, feel free to paste some code here and I try to help (e.g. your and your custom component code).

1 Like