I've recently started getting more familiar with the Expert Feedback Loop for VA but don't understand how to handle utterances that don't relate to an intent (or topic), but rather relates to an existing article.
I've been slowly creating intents and not associating them to topics to help the NLU trigger the right response (if no active topic exists, then it returns the search fallback), but how do I capture this in the Expert Feedback Loop?
Do I select Unsure, or Mismatch > Mark as not relevant to this model?
Topics may exist for these eventually, we're working on it!
