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.

• ✗

  Key Phrase Extraction

  Why it's wrong here

  Key Phrase Extraction performs semantic and syntactic analysis to surface the most salient topics or concepts in a document, such as 'contract breach' or 'due diligence'—it does not classify entity types like people or addresses. While a personal name could theoretically appear as part of a key phrase, the service neither marks it as PII nor provides any redaction or masking mechanism. Therefore, relying on it would leave sensitive data intact and fail the firm's compliance requirement.

  When this WOULD be correct

  A company needs to automatically extract the main topics or keywords from customer feedback surveys to identify common themes. They want a prebuilt Azure AI Language feature without custom training. Key Phrase Extraction would be the correct choice.

• ✗

  Sentiment Analysis

  Why it's wrong here

  Sentiment Analysis assigns a polarity score (e.g., positive, negative, neutral) to each sentence or document based on word embeddings and contextual cues, and is often used for customer feedback or legal case risk assessment. Although a strongly negative sentence might be an indicator that sensitive information is discussed, the service does not extract the specific entities or offer any masking output. Thus it falls far short of the systematic detection and redaction that the law firm needs.

  When this WOULD be correct

  A company wants to automatically gauge customer satisfaction from support tickets by analyzing the emotional tone of the text. Sentiment Analysis would be the correct prebuilt feature to classify each ticket as positive, negative, or neutral.

• ✓

  Personally Identifiable Information (PII) Detection

  Why this is correct

  Azure AI Language's PII Detection service uses pre-trained natural language processing models to identify sensitive entities—such as names, email addresses, phone numbers, and government ID numbers—within unstructured text. For legal documents, this allows the law firm to automatically locate and then redact that content by replacing it with a placeholder, category label, or null value, while preserving the rest of the document. Because it is a dedicated entity recognition and redaction pipeline, it directly satisfies the firm's need to strip personal data from records before sharing them.

• ✗

  Language Detection

  Why it's wrong here

  Language Detection analyzes character patterns and n-gram statistics to identify the dominant language of a document, returning a language code and a confidence score, but it makes no claim about the content of the text. It is typically used as a routing step before other text analytics APIs, not to inspect for personal data. Because it cannot tag or remove names, phone numbers, or other identifiers, it is an unsuitable tool for this redaction task.

  When this WOULD be correct

  A multinational company needs to automatically route customer support tickets to language-specific teams based on the language of the message. Language Detection would be the correct choice to identify the language of each ticket.

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