Today you can export the trained model to NLU and the Discovery Service
We want the Watson Assistant to collect entities (and intents) during the conversation and use this information to generate a better query for the Discovery Service
Today we do this by naming convention, we name the entities in the Assistant the same as the entities in WKS (WDS) and construct a query. For a detailed explanation see: https://www.ibm.com/developerworks/community/blogs/ibmandgoogle/entry/Build_a_document_based_chatbot_using_Watson_Assistant_and_Discovery_Service?lang=en
Why is it useful?
|Who would benefit from this IDEA?||Customers have a single place to maintain their domain model, this increases the speed of iterations/improvements|
How should it work?