BenchMark Quizzes by MDFT Pro

Handle Spurious Requests In Conversational Language Understanding

MDFT Pro, a well-known training agency, has deployed a Conversational Language Understanding model to help students navigate course enrollment, schedule inquiries, and academic support services.

Claire, the Student Services AI Manager, has noticed that students are asking questions about topics completely unrelated to the model’s capabilities, such as restaurant recommendations, weather forecasts, and personal advice. These spurious requests are receiving incorrect responses that confuse students and potentially provide misleading information about the training agency’s services and policies.

What should Claire do to ensure the model properly identifies and handles spurious requests that don’t relate to its intended capabilities?

Choose the correct answer from the options below.

Explanations for each answer:

Add examples to the custom intents is incorrect. Adding examples to custom intents would train the model to better recognize those specific intents, but wouldn't help identify requests that don't relate to any of the model's capabilities.
Enable active learning is correct. Active learning helps identify and handle spurious requests by suggesting new training examples based on user interactions, including requests that don't match existing intents.
Add entities is incorrect. Adding entities helps extract specific information from recognized intents but doesn't help identify requests that fall outside the model's intended scope.
Add examples to the None intent is incorrect. The None intent helps handle out-of-scope requests, but active learning is more effective at automatically identifying and managing spurious requests as they occur.

Learn more about improving CLU models:

CLU Model Improvement

Exam Preparation Quiz

Handle Spurious Requests In Conversational Language Understanding