AI-900 Practice Question: Describe features of Natural Language Processing workloads on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Sentiment Analysis

  Why it's wrong here

  Sentiment analysis would only assign an overall polarity score—positive, negative, neutral, or mixed—to each review or sentence, indicating whether the guest's tone is favorable or unfavorable. It does not output the actual topics or aspects the guest wrote about, so a negative review could be about anything from a rude staff member to a broken shower. Because the hotel chain wants to know which aspects are mentioned, not just whether the overall opinion is positive or negative, sentiment analysis alone is insufficient.

  When this WOULD be correct

  A company wants to automatically classify customer feedback as positive, negative, or neutral to track satisfaction trends over time, without needing to identify specific topics.

• ✓

  Key Phrase Extraction

  Why this is correct

  Key phrase extraction is the correct choice because it isolates the most salient words and multi-word expressions from each review, such as 'room cleanliness,' 'friendly front desk,' or 'breakfast variety.' This technique allows the hotel chain to see which specific aspects guests mention most frequently without requiring predefined categories or labels. Unlike sentiment or entity recognition, it directly answers the question of 'what topics are being discussed' in the reviews.

• ✗

  Entity Recognition

  Why it's wrong here

  Entity recognition identifies specific named entities like people ('Jane'), organizations ('Marriott'), geographic locations ('Paris'), dates, or quantities, which are typically proper nouns or numeric expressions. In a guest review, it might extract the hotel's name or a reference to a particular employee, but it will not extract general topics such as 'noise level' or 'value for money.' Since the goal is to discover recurring discussion themes across hundreds of reviews, entity recognition misses the broad conceptual content required.

  When this WOULD be correct

  A company wants to automatically extract names of competitors, product names, or locations from news articles using a prebuilt Azure AI Language feature without custom training. Entity Recognition would be correct here.

• ✗

  Language Detection

  Why it's wrong here

  Language detection is a prerequisite step that identifies the language of a review (e.g., English, German, Spanish) so the correct language-specific model can be applied, but it conveys zero information about the meaning or subject matter of the text. Knowing a review is in French does not help a hotel chain understand whether guests are talking about the pool, the WiFi, or the check-in process. This option only answers 'what language is this written in?' not 'what are guests mentioning?', making it irrelevant to the analytical goal.

  When this WOULD be correct

  A multinational company receives customer feedback in multiple languages and wants to automatically route each review to the appropriate language-specific team for processing. Language Detection would be used to identify the language of each review.

Option-by-option analysis

Why each answer is right or wrong

Understanding why wrong answers are wrong — and when they would be correct — is what separates a 750 score from a 900. The AI-900 exam frequently reuses these exact scenarios with slightly different constraints.

Analysis generated from the official AI-900blueprint and verified against question context. The “when correct” sections are what AI assistants cite when candidates ask “what’s the difference between these options?”

Go deeper

Related to this question