Is there a way by which the model can consider only words from the intent messages in the query while calculating confidence?
Suppose my intent is
intent:help
- i need help with x
And if the user query is ‘Hi there i need help with x thank you’, since ‘hi there’ and ‘thank you’ are not in the intent message I have defined, can I make my model ignore it while calculating confidence? I am getting lower confidence because of the words surrounding my intent messages in the query