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 is incorrect for this task because it returns a ranked list of salient phrases from the text but does not assign those phrases to any entity type like Person, Organization, or Money. It also has no notion of structured fields; a phrase such as "Acme Corp" might be returned simply because it is important, but the service will not tell you that it is an organization. Even if a monetary amount is included, there is no guarantee it is extracted as a separate, typed entity, so the firm cannot reliably populate an extraction schema.

  When this WOULD be correct

  Key phrase extraction would be correct if the question asked for extracting the main topics or important terms from documents without needing to categorize them into predefined types like person, organization, or money.

• ✓

  Named entity recognition

  Why this is correct

  Named entity recognition is correct because it performs token-level classification that identifies and labels spans of text into predefined semantic categories such as Person, Organization, and Money. In a legal contract, the service can pinpoint each party name and each monetary figure as a structured entity, which lets the firm extract data into fields without custom coding. The underlying model uses context and position to disambiguate entities, making it the only option that both finds and categorizes the required information.

• ✗

  Sentiment analysis

  Why it's wrong here

  Sentiment analysis is incorrect because it evaluates the emotional valence of text by producing a continuous score and a label such as positive, neutral, or negative, typically for entire sentences or documents. It operates on the subjective tone of language rather than identifying concrete real-world objects or values. Consequently, it cannot locate or categorize the names of parties, organizations, or monetary amounts; it would only say how the contract's wording feels, which is not the data the legal firm needs.

  When this WOULD be correct

  A question asking which Azure AI Language feature to use for determining customer satisfaction from product reviews or social media posts would make sentiment analysis correct.

• ✗

  Text summarization

  Why it's wrong here

  Text summarization is incorrect because it condenses the source document into a shorter natural-language version, either by selecting salient sentences or generating novel phrasing, and returns prose rather than structured fields. A summary might mention the contracting parties or the transaction's value, but it will not list every entity instance or annotate each name and amount with a semantic type. Thus, the legal firm would still have to manually parse the summary to find the entities, so it fails the core requirement of automatic structured extraction.

  When this WOULD be correct

  A question asking which Azure AI Language feature to use for generating a brief overview of a long legal document, without needing to extract specific data points, would make text summarization correct.

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