AI-102 Question 310
Select 3You are designing a knowledge base using Azure Cognitive Services QnA Maker. After importing a dataset, you notice that some questions in the knowledge base are returning irrelevant answers during testing. What steps should you take to improve the accuracy of responses?
- A
Add alternate phrasing for questions to the knowledge base
- B
Train the knowledge base by marking incorrect answers and providing correct ones
- C
Re-import the dataset without making any changes to the existing knowledge base
- D
Use the test tool in QnA Maker to simulate user queries and refine the knowledge base
- E
Increase the number of questions by adding unrelated FAQs to the dataset
Show answer and explanation
Correct answers: A, B, D
Explanation
To improve the accuracy of a QnA knowledge base, it is essential to enhance the dataset by adding alternate phrasing, train the model with feedback, and test the knowledge base using relevant queries. These steps ensure the knowledge base provides more accurate and contextually relevant responses while avoiding practices that could degrade overall quality.
- A. Correct.
Adding alternate phrasing improves the ability of the knowledge base to recognize similar user queries, which can enhance response accuracy.
- B. Correct.
Marking incorrect answers and providing correct ones during training helps the model learn and improve its response accuracy.
- C. Incorrect.
Re-importing the dataset without changes does not address the issue of irrelevant answers and will not improve accuracy.
- D. Correct.
Testing the knowledge base with simulated queries allows you to identify gaps or inaccuracies and refine the QnA model accordingly.
- E. Incorrect.
Adding unrelated FAQs can dilute the quality of the knowledge base and lead to more irrelevant answers.