CertPrep Coach

AI-901 Information extraction practice questions

12 sample questions with full explanations. Try each one before you open the answer. New to this area? Read the Information extraction study guide first.

  1. Question 1

    An accounts team wants to pull supplier name, invoice number, due date and total from thousands of PDF invoices, with as little setup as possible. What should it use in Azure Content Understanding?

    • A The prebuilt-invoice analyzer
    • B A custom analyzer with a field schema designed from scratch
    • C A vision-enabled GPT model prompted to read each invoice
    • D Basic OCR only, then reading the values by hand
    Show answer and explanation

    Answer: A

    Content Understanding includes prebuilt analyzers for common scenarios. prebuilt-invoice already knows the usual invoice fields, so there's nothing to design.

    • B: A custom analyzer would work, but designing a schema from scratch is more effort than using the prebuilt one for a common document type.
    • C: A vision model could read invoices, but you'd have to write and test the prompts and output format yourself. The prebuilt analyzer already does this.
    • D: Basic OCR only reads the text. It doesn't know which value is the due date or the total.

    Source: Extract information from documents (Get started with AI-powered information extraction in Azure) (Microsoft Learn)

  2. Question 2

    An appliance maker wants to extract fields such as model number, purchase date and fault description from its own warranty-claim form. No prebuilt analyzer covers this form. What should it create in Azure Content Understanding?

    • A A deployment of the prebuilt-invoice analyzer
    • B A custom analyzer with a schema listing the fields to extract
    • C Basic OCR, which returns the text on each form
    • D A prebuilt-videoSearch analyzer
    Show answer and explanation

    Answer: B

    Custom analyzers are tailored to your needs. You define a schema that describes what information to extract and how it should be structured, and the analyzer maps values in the document to those fields.

    • A: The invoice analyzer looks for invoice fields, not warranty-claim fields such as fault description.
    • C: Basic OCR returns the text, but it doesn't know which value is the model number or the fault description.
    • D: prebuilt-videoSearch analyzes videos, not forms.

    Source: Extract information from documents (Get started with AI-powered information extraction in Azure) (Microsoft Learn)

  3. Question 3

    An insurer is comparing basic OCR with Azure Content Understanding for its claim forms. For each statement, choose Yes if it is true or No if it is false.

    • Basic OCR understands which value on the form is the policy number.
    • Content Understanding can map text values in a document to fields defined in a schema.
    • A Content Understanding schema can include nested fields, not just a flat list.
    Show answer and explanation
    Basic OCR understands which value on the form is the policy number.No
    Content Understanding can map text values in a document to fields defined in a schema.Yes
    A Content Understanding schema can include nested fields, not just a flat list.Yes

    Basic OCR recognizes printed text but doesn't understand meaning, context or relationships between words. Content Understanding goes further: its analyzers map values to the fields in your schema, and schemas support structured and nested fields.

    Source: Extract information from documents (Get started with AI-powered information extraction in Azure) (Microsoft Learn)

  4. Question 4

    You analyze a scanned contract with a custom Azure Content Understanding analyzer that has a field schema, and open the JSON result. For each statement, choose Yes if it is true or No if it is false.

    • The result includes the extracted content as markdown.
    • The result includes the fields and the values found for them.
    • The result includes a newly generated image of the contract.
    Show answer and explanation
    The result includes the extracted content as markdown.Yes
    The result includes the fields and the values found for them.Yes
    The result includes a newly generated image of the contract.No

    Content Understanding returns JSON with the extracted markdown, the fields, the data in each field and a confidence score. It analyzes content; it doesn't generate new images.

    Source: Extract information from documents (Get started with AI-powered information extraction in Azure) (Microsoft Learn)

  5. Question 5

    A property website wants to use Azure Content Understanding to extract descriptions and other insights from listing photos, starting with a prebuilt analyzer. Which analyzer should it use?

    • A prebuilt-invoice
    • B prebuilt-audioSearch
    • C prebuilt-videoSearch
    • D prebuilt-imageSearch
    Show answer and explanation

    Answer: D

    Content Understanding's prebuilt analyzers cover common content types. prebuilt-imageSearch is the one for images such as photos.

    • A: prebuilt-invoice extracts fields from invoices.
    • B: prebuilt-audioSearch analyzes audio recordings.
    • C: prebuilt-videoSearch analyzes video. The listings are still photos.

    Source: Extract information from documents (Get started with AI-powered information extraction in Azure) (Microsoft Learn)

  6. Question 6

    A clinic receives hundreds of voicemails a day. For each one, staff want the caller's name, a summary of the message, any requested actions and a callback number, without listening to the whole recording. Which option fits best?

    • A Azure Content Understanding, analyzing the audio with a schema for those fields
    • B Azure Speech real-time speech to text
    • C Azure Language PII detection
    • D The Voice Live API
    Show answer and explanation

    Answer: A

    Content Understanding supports audio. You can define a schema of insights to extract from each recording, such as caller, summary, requested actions and callback number, and get a transcript plus those fields as JSON.

    • B: Speech to text gives you a transcript, but someone would still have to read it and pick out the caller, actions and number.
    • C: PII detection works on text and finds personal details, but it doesn't summarize the message or list requested actions.
    • D: Voice Live is for live conversations with a voice agent, not for analyzing recorded voicemails.

    Source: Extract information from audio and video (Get started with AI-powered information extraction in Azure) (Microsoft Learn)

  7. Question 7

    A company records its team meetings on video. It wants a summary of each meeting, who spoke and what they said, and a list of assigned actions. Which option fits best?

    • A Sora 2
    • B A vision-enabled GPT model, given one still frame from each meeting
    • C Azure Content Understanding video analysis
    • D Azure Speech text to speech
    Show answer and explanation

    Answer: C

    Content Understanding supports video analysis. From a meeting recording it can extract details such as who spoke and what they said, a summary of the discussion and assigned actions.

    • A: Sora generates new videos. It doesn't analyze existing ones.
    • B: A single still frame shows who was in the room, not what anyone said or decided.
    • D: Text to speech creates audio from text. It doesn't analyze recordings.

    Source: Extract information from audio and video (Get started with AI-powered information extraction in Azure) (Microsoft Learn)

  8. Question 8

    A call center plans to analyze recorded calls with Azure Content Understanding. For each statement, choose Yes if it is true or No if it is false.

    • Content Understanding can produce a written transcript of each call.
    • The analysis results are returned in JSON format for further processing.
    • Content Understanding can analyze documents and images, but not audio.
    Show answer and explanation
    Content Understanding can produce a written transcript of each call.Yes
    The analysis results are returned in JSON format for further processing.Yes
    Content Understanding can analyze documents and images, but not audio.No

    Content Understanding supports audio and video as well as documents and images. For a call, it can return a transcript and other insights such as a summary, and, as with its other analyzers, the results come back as JSON.

    Source: Extract information from audio and video (Get started with AI-powered information extraction in Azure) (Microsoft Learn)

  9. Question 9

    You are defining fields for a custom Azure Content Understanding document analyzer. For each statement, choose Yes if it is true or No if it is false.

    • An InvoiceNumber field that copies the number exactly as printed should use the extract method.
    • A field that assigns each document one of the values Invoice, Receipt or Other should use the generate method.
    • A field that writes a one-sentence summary of a contract's termination terms should use the generate method.
    Show answer and explanation
    An InvoiceNumber field that copies the number exactly as printed should use the extract method.Yes
    A field that assigns each document one of the values Invoice, Receipt or Other should use the generate method.No
    A field that writes a one-sentence summary of a contract's termination terms should use the generate method.Yes

    Use extract for values that appear directly in the document, such as an invoice number. Use classify when the value must be one of a fixed set of options, such as a document type. Use generate for values the model has to infer or write, such as a summary.

    Source: What is Azure Content Understanding in Foundry Tools? (Microsoft Learn)

  10. Question 10

    A support center wants topics, customer sentiment and key insights from each recorded call, using a prebuilt Content Understanding analyzer without defining its own schema. Which analyzer should it use?

    • A prebuilt-audio
    • B prebuilt-callCenter
    • C prebuilt-videoSearch
    • D prebuilt-audioSearch
    Show answer and explanation

    Answer: B

    prebuilt-callCenter is a domain-specific analyzer that extracts topics, sentiment and key insights from call recordings, so no custom schema is needed.

    • A: prebuilt-audio is a base analyzer that you build a custom analyzer on, with your own schema.
    • C: prebuilt-videoSearch produces transcripts and descriptions for video segments. The calls are audio.
    • D: prebuilt-audioSearch produces transcripts and summaries for search. The domain analyzer for topics and sentiment in calls is prebuilt-callCenter.

    Source: Azure Content Understanding in Foundry Tools prebuilt analyzers (Microsoft Learn)

  11. Question 11

    A team photographs whiteboards after each meeting. They want Azure Content Understanding to pull out the handwritten notes as text so they can search them later. The pictures themselves don't matter, only the words on them. Which approach does Microsoft recommend?

    • A A custom analyzer based on prebuilt-image, with a field for the notes
    • B A document analyzer, such as a document field extraction schema
    • C prebuilt-imageSearch, which writes a one-paragraph description
    • D prebuilt-videoSearch, with each photo treated as a short clip
    Show answer and explanation

    Answer: B

    Image analyzers aren't optimized for scenarios that depend mainly on text extracted from an image. When the goal is to read the text, Microsoft suggests a document field extraction schema instead. Document analysis also captures handwritten notes.

    • A: It could return something, but image analyzers aren't optimized for analysis that's based mainly on extracted text. They suit visual details, such as defects or products on a shelf.
    • C: prebuilt-imageSearch describes what an image shows. A summary paragraph isn't the full text of the notes.
    • D: prebuilt-videoSearch is for video files, not still photos.

    Source: Azure Content Understanding in Foundry Tools image overview (Microsoft Learn)

  12. Question 12

    An insurance company wants one service that pulls structured fields from claim forms, photos of vehicle damage, and recorded customer calls, and returns them in a format the team defines. Which service fits best?

    • A Azure Speech in Foundry Tools
    • B Azure Content Understanding in Foundry Tools
    • C Azure Language in Foundry Tools
    • D Azure Translator
    Show answer and explanation

    Answer: B

    Content Understanding uses generative AI to process documents, images, video, and audio into an output format you define. That covers claim forms, damage photos, and call recordings with one service.

    • A: Azure Speech handles audio, such as transcribing calls, but not forms or photos.
    • C: Azure Language analyzes text. It can't read photos or audio directly.
    • D: Translator converts text between languages; it doesn't extract structured fields.

    Source: What is Azure Content Understanding in Foundry Tools? (Microsoft Learn)

Want more practice on Information extraction?

Practice the full question bank for this area, in random order, with your progress saved in your browser.

Practice Information extraction