I have occurrences where an intent is triggered when a knowledge article should be surfaced. I have gone through the intent utterances and verified that the utterances do not exist, however the NLU test still shows some high % probability.
Here is an example: somebody types "Who is the city manager?" into the digital bot, which is an exact match on a title of a knowledge article. The intent that is triggered is for "chat" where we connect the user to an agent, however there is no utterance for "city manager" or even "manager". I am dumbstruck on how this is occurring or how to prevent it for this and other articles/intents that cross over in probability. Any advice is much appreciated!!


#ConversationalAI(Bots,AgentAssist,etc.)------------------------------
Nicole VanWie
UC Engineer
------------------------------