Hi there,
this might be a question concerning the Rasa Core. I will try to keep my example as simple as possible.
User should be able to check a product’s weight, materials and dimensions. Hence, I have the three intents check_product_weight, check_product_materials and check_product_dimensions.
As the user can ask “What is the product made of?” and the bot would utter “What is the product’s item number?”, the user would just send the item number.
Some NLU examples for the three intents above just contain the entity item_number and look like the same, and that is where I am stuck. Should I better create a new intent like tell_product_number for that?
My bot recognizes the intent of the user’s utterance “What is the product made of?” correctly as product_materials. But when the user enters the item number, sometimes the bot wants to run the wrong action, e.g. action_product_weight, although the user asked for the materials.
Do I simply need more dialogue examples (in my stories.md) to face this issue or do I have to change something regarding my intents or my pipeline?
My pipeline is a pre-configured spacy pipeline and looks as follows:
pipeline:
- name: "SpacyNLP"
- name: "SpacyTokenizer"
- name: "SpacyFeaturizer"
- name: "RegexFeaturizer"
- name: "CRFEntityExtractor"
- name: "EntitySynonymMapper"
- name: "SklearnIntentClassifier"
Regards, Nick
Thanks for the advice having a look at the Forms, I might need that for something else now…