Answer intent uses machine learning to understand a user question and return a single answer. The QnA framework used earlier used elastic search instead, which didn’t offer the same level of confidence as NLP. For this reason, the QnA framework always showed several possible question / answer pairs that would match a user message.
Although the answer to a question that is returned by answer intents has a high confidence, some customers have expressed their preference for the “old” user interface, which lists multiple question-answer pairs if the runner-up answers resolve in a configurable range of confidence.
This article explains how to use NLP and regular intents to resolve answers the “new way” while displaying it the “old way”. The target audience for this article are Oracle Digital Assistant developers who used the QnA feature available in 19.x version of Oracle Digital Assistant and who now move to using answer intents in Oracle Digital Assistant on Oracle Cloud Infrastructure (ODA 20.x and later). Read the complete article here.
For regular information become a member in the Developer Partner Community please register here.