AI-901 ยท Information extraction with Content Understanding
How does a document analyzer turn an invoice into data?
Pulling text, layout, tables and defined fields from invoices, receipts and forms, field types and confidence, and when a document analyzer fits.
Key points
- Content extraction first reads text (printed and handwritten), selection marks, barcodes, paragraphs, sections and tables.
- Field extraction then fills your schema (customer, dates, totals, line items) with typed, normalized values.
- Each field can carry a confidence score and a source location, so low-confidence values can go to human review.
Source: https://learn.microsoft.com/en-us/azure/ai-services/content-understanding/document/overview
Practice questions
Which prebuilt analyzer is designed for retail and restaurant sales receipts?
- prebuilt-receipt
- prebuilt-idDocument
- prebuilt-contract
- prebuilt-callCenter
Show answer
Answer: A. prebuilt-receipt
prebuilt-receipt targets sales receipts. prebuilt-invoice is for invoices and similar billing documents, and prebuilt-idDocument is for identity documents.
Which items does document content extraction capture?
- Tables, including cells that span rows or columns
- Selection marks such as checkboxes
- Barcodes
- Speaker turns in a recording
- Camera shot boundaries
Show answer
Answer: A. Tables, including cells that span rows or columns, B. Selection marks such as checkboxes, C. Barcodes
Document content extraction captures tables (including spanning cells), selection marks such as checkboxes, and barcodes. Speaker diarization belongs to audio and video, and shot detection to video.
True or false: in a custom document analyzer, confidence scores and source grounding are always returned without any setting.
Show answer
Answer: False
False. Confidence and grounding are opt-in: set estimateFieldSourceAndConfidence to true in the analyzer config (or estimateSourceAndConfidence on a single field).
Practice 8 questions on this topic in Bitzu
Short lessons, instant explanations and spaced review of what you get wrong. Free on iOS and Android.
