BenchMark Quizzes by MDFT Pro

Fix Question Answering For Alternative Question Phrasings

MDFT Pro, a well-known training agency, is building a student support chatbot using Azure AI Language question answering service. Mark, the Chatbot Developer, has uploaded their course catalog and price list document (CourseInfo.pdf) and trained the model.

During testing, the chatbot correctly responds when students ask “What is the price of Azure Fundamentals course?” but fails to understand when they ask “How much does Azure Fundamentals course cost?” - even though both questions seek the same information about course pricing. Students are frustrated because they phrase pricing questions in different ways but don’t always get responses, requiring them to use specific wording to get information about course costs.

The proposed solution is to create an entity for price in Language Studio, then retrain and republish the model. Does this solution meet the goal of ensuring the chatbot responds correctly to both question variations?

Choose the correct answer from the options below.

Explanations for each answer:

Yes is incorrect. Creating entities for price doesn't address the core issue of question variation recognition. The problem is that the model doesn't understand different ways to ask about pricing.
No is correct. This solution doesn't meet the goal because entities don't help the model recognize alternative question phrasings like 'How much does [product] cost?' versus 'What is the price of [product]?'
Only for some products is incorrect. The issue is not product-specific but rather about recognizing different ways students phrase the same question about pricing.
Only after additional training is incorrect. While additional training might be needed, creating price entities alone won't solve the question variation recognition problem.

Learn more about question answering variations:

Question Answering Variations

Exam Preparation Quiz

Fix Question Answering For Alternative Question Phrasings