BenchMark Quizzes by MDFT Pro

Improve Language Understanding Agent Response Accuracy

MDFT Pro, a well-known training agency, operates a student support chatbot powered by Language Understanding service to handle natural language queries from students and prospective learners. Claire, the Digital Experience Manager, has received numerous complaints from users that the chatbot frequently responds with generic messages like “Sorry, I don’t understand that” when students ask legitimate questions about course enrollment, scheduling, prerequisites, and certification requirements.

The current model seems unable to recognize many variations of common student inquiries, significantly impacting the user experience and requiring human intervention for basic support requests.

Which three actions should Claire perform in sequence to improve the chatbot’s ability to respond to student requests?

Choose all correct answers from the options below.

Explanations for each answer:

Enable active learning is correct. Active learning captures real user utterances that the model struggled with, providing valuable data for improving the model's understanding.
Validate the utterances logged for review and modify the model is correct. Reviewing and adding these utterances as training examples helps the model learn from actual user interactions and improve responses.
Train and republish the Language Understanding model is correct. After adding new utterances and examples, retraining and republishing applies the improvements to the live model.
Add prebuilt domain models as required is incorrect. Prebuilt domains provide general functionality but won't address the specific gaps in understanding that cause generic responses.
Migrate authoring to an Azure resource authoring key is incorrect. Changing authoring resources doesn't improve model performance or reduce generic responses to user queries.

Learn more about LUIS active learning:

LUIS Active Learning

Exam Preparation Quiz

Improve Language Understanding Agent Response Accuracy