IBM Watson Assistant: How to group conditions on dialog nodes?

Question

I have to maintain and expand a Watson Assistant chatbot developed by someone else. Sometimes we get conflicting intents, so I would like to have more control on the conditions that enter in a dialog node. See the attached image below as an example.

There is the more specific intent #ctpat_issues , but since sometimes user asks the more general #what_is followed by entity @ctpat_issues , I would like to resolve both in the same dialog node.

Is there possibility to group conditions with parenthesis? How are and/or conditions evaluated together?

Also, do you see any major problem in handling intents this way, with a mix of general and more specific intents? Any hints on how to learn the best practices regarding intents/entities/dialog organization?

Answer 1

First, see the documentation on how to define the conditions on dialog nodes . You can define them in the UI as shown in your screenshot or compose them with more options in the JSON editor. The conditions are based on SpEL and you could even apply regular expressions.

When looking into how to get the right answers from Watson Assistant, there is the way of being more precise with the conditions (as asked), but also to train Watson by correcting the recognized dialog node ( or here ).