AI-102 exam dumps

AI-102 practice question 309 of 493

Designing and Implementing a Microsoft Azure AI Solution. Professional level, Microsoft. Free question with the correct answer and a full explanation.

AI-102 Question 309

Select 3

You are designing a QnA Maker knowledge base for an FAQ bot that will assist customers with product support queries. During testing, you notice that the bot frequently provides incorrect answers to questions that are similar but require distinct responses. How can you improve the accuracy of the responses in the knowledge base?

  1. A

    Add alternate phrasing for similar questions and map them to the correct answers in the knowledge base.

  2. B

    Increase the confidence score threshold for matching user queries to answers.

  3. C

    Enable active learning to improve the knowledge base based on user interactions.

  4. D

    Train the knowledge base by providing additional context to ambiguous questions.

  5. E

    Manually adjust the confidence levels of incorrect answers in the Azure QnA Maker portal.

Show answer and explanation

Correct answers: A, C, D

Explanation

To improve accuracy in a QnA Maker knowledge base, it is essential to provide alternate phrasings for questions, enable active learning to refine responses based on user feedback, and add contextual information to handle ambiguity. These approaches directly enhance the quality and precision of the bot's responses, reducing the likelihood of incorrect answers.

  • A. Correct.

    Adding alternate phrasing helps the model learn to associate different ways of asking a similar question with the correct answer, improving accuracy.

  • B. Incorrect.

    Increasing the confidence threshold might reduce false positives but does not directly improve the knowledge base's ability to distinguish between similar questions.

  • C. Correct.

    Active learning allows the model to improve over time based on user feedback and interactions, making it a valuable tool for refining the knowledge base.

  • D. Correct.

    Providing additional context to ambiguous questions helps the model make more accurate predictions and reduces confusion between similar queries.

  • E. Incorrect.

    Manually adjusting confidence levels is not a feature available in the Azure QnA Maker portal, and confidence scores are calculated dynamically during runtime.

Timed practice exam

Take a AI-102 practice test under exam conditions

40 questions in 120 minutes, drawn from this bank, with a score report and a per-question review when you finish.

Start timed exam