I am working on a use case where I want to pass three-slot values in one go. To explain it better let’s just say a user enters, “Please send the data on xyz@gmail.com of Apple for 2020-02-20”. Here the email address, Apple and 2020-02-20 are all entities filled using slots. A user can also enter the above information with a different order of entities. In that case, rasa gets confused and messes up the values. How can I do this efficiently such that the user has the freedom to enter the values in any order and they get extracted correctly?
what entity extractor are you using? You should be able to use duckling for email and date, which is independent of training data (it is rule based/pretrained). That should extract dates and emails in most contexts. something like Apple would be a custom entity to extract with DIET, in which case how well it extracts the entity depends on your configuration and also on the amount and variety of your training data.
@mloubser Can I use 3 entity extractors like spacy nlp crfentityextractor and duckling together, all at once?
Yes you can! As long as you don’t use conflicting entity names, you can use any combination of the extractors. In fact, the financial-demo bot does exactly that - see the config and NLU data