The json from ibm watson for conversation api does not triggering the negatives

Question

I am trying the watson conversation api on android phone. I had made a model on watson and completed the app. Everything works fine but when i am giving irrelevant inputs it always gives one of the intent instead of giving the negative response. I had created a separate conversation for triggering on anything else. But when i try the conversation on watson everything works perfect, but the out put json to my phone doesn't contains the 'anything else' responses

Answer 1

Every flow, Watson tries to recognize the intent, you can see the level confidence for every input from user in one payload window inside project from IBM Developers here. Or you can did this debugging the data return in your app.

One good idea is create one flow with conditions with confidence, entities, regex, context variable or true condition. This true condition flow will did Watson talk to user if one intent is less than confidence inside your condition, or according your flow, will return exactly what you want if you use conditions with Entities or regex.

Simon did one example for conditions with Entities, Context Variables and regex, check this link: Simon example.

And you can did conditions with confidence level for make sure if user type the correctly intents trainning inside your Conversation, check here.