Search code examples
pythonrasa-nlurasa-core

How to (edit or process ) the message from the user before being classified which intent by rasa-nlu model?

I am new to RASA.. I read the last updated documentation Rasa 1.0 but I don't know how to process the message of the user before nlu-model
I would like to do things like stemming spellchecking words etc ...

I appreciate any code example with spaCy library :)

Solution

  • Rasa already comes with a few pre-processing components like Tokenizers and you can include them in your configuration file of the NLU pipeline

    For tasks like stemming and spell-checker, they are currently out of scope of Rasa NLU. The developer has to pre-process the input and pass the processed text as a string to NLU pipeline for parsing.

    Nevertheless here are a few resources you can checkout -

    • Tutorial on common pre-processing tasks with python and spacy.
    • Hunspell spell checker and integration with Spacy