Developing the knowledgebase of QnA Maker pairs
Your QnA pairs should be designed and developed based on the usage of the client application.
Each pair can contain:
Metadata: can be filtered when making queries to enable you to tag your QnA pairs with additional information about the source, content, format, and purpose of your data.
Follow-up messages: help determine a path through your knowledgebase for the user to get to the correct answer.
Alternative questions: important to allow the search to match your answer from different forms of the question. Active learning suggestions are converted into alternative questions.
Developing a knowledgebase for insertion into a DevOps pipeline requires that the knowledgebase be isolated during batch testing.
A knowledgebase shares the cognitive search index with all other knowledgebases of the QnA Maker resource. While the knowledgebase is isolated by partition, sharing the index can cause a difference in score compared to the published knowledgebase.
To get the same score on the test and production knowledgebases, isolate a QnA Maker resource.In this architecture, the resource only needs to live as long as the isolated batch test.