AI-102 Question 215
Single answerYou are building an AI-powered customer support chatbot for a retail company using Azure Language Understanding (LUIS). The chatbot should identify customer intents such as 'Track Order', 'Cancel Order', and 'Product Inquiry'. During development, you notice that the model struggles to differentiate between 'Track Order' and 'Cancel Order' intents due to overlapping phrases. What should you do to improve intent recognition?
- A
Add more example utterances for each intent that closely align with real-world customer queries.
- B
Reduce the total number of intents in the model to simplify recognition.
- C
Use Azure Cognitive Search instead of LUIS for handling these intents.
- D
Enable active learning in LUIS to identify confusing utterances and retrain the model with labeled data.
Show answer and explanation
Correct answer: A
Explanation
To improve intent recognition in LUIS, the most effective approach is to provide more example utterances for each intent. This helps the model better understand the subtle differences in language patterns between intents. While active learning and other techniques can complement this approach, adding diverse and representative training data is the foundational step for addressing intent confusion.
- A. Correct.
Adding more example utterances provides the model with a broader set of training data, improving its ability to differentiate between similar intents.
- B. Incorrect.
Reducing the total number of intents might simplify the model but does not address the issue of overlapping phrases in this scenario.
- C. Incorrect.
Azure Cognitive Search is not designed for intent recognition and would not be appropriate for this task.
- D. Incorrect.
Active learning can help improve the model, but it is not the primary solution to the problem of overlapping intent phrases.