ibm-watsonwatson-dialogwatson-conversation

How to use #out_of_scope intent with Watson's Conversation service?


I have defined #out_of_scope intent in the conversation workspace and added several user examples. Now when I ask a question that is not specified as an example, the JSON returned has #out_of_scope intent with a very low confidence score.

Why is that ? Does this mean, that if a user asks a question which is not in the list of specified examples, it will not be mapped to #out_of_scope intent ?

Thank you, Sandhya


Solution

  • Technically #out_of_scope is just a label. If the user input is not similar to your sample data, it will not be highly confident. The real benefit of the out of scope class is that it will go a long way in making so your other classes are not highly confident against any "noise"