Describe Artificial Intelligence workloads and considerations

The primary purpose of the prebuilt-invoice model in Azure AI Document Intelligence is to extract vendor details, invoice dates, line items, tax amounts, and total charges from billing documents.

This is the correct answer because the prebuilt-invoice model is specifically designed to process invoices and extract key financial and transactional information commonly found in billing documents. The model focuses on the core elements that make up an invoice: vendor information (who is billing), invoice dates (when the transaction occurred), line items (what is being billed), tax amounts (applicable taxes), and total charges (the final amount due).

Why the other answers are incorrect:

The second option incorrectly includes 'payment confirmation numbers' as part of the standard extraction. The prebuilt-invoice model is designed for invoices, which are requests for payment, not payment confirmations. Payment confirmation numbers would typically appear on payment receipts or remittance documents, not on invoices themselves.

The third option describes document classification functionality rather than data extraction. While Azure AI Document Intelligence does offer classification capabilities, the prebuilt-invoice model's primary purpose is extraction of structured data from invoices, not classification of different document types.

The fourth option describes functionality more aligned with vendor onboarding or registration processes. The prebuilt-invoice model is specifically designed for processing billing invoices, not supplier registration documents, business licenses, or tax identification documentation, which would require different extraction patterns and fields.

Named Entity Recognition (NER) is a natural language processing technique that identifies and classifies specific data elements within text into predefined categories. These categories typically include:

• People (names of individuals)
• Places (locations, cities, countries)
• Organizations (companies, institutions, agencies)
• Dates and times
• Monetary values
• Percentages
• Other domain-specific entities

The correct answer accurately describes this functionality. NER extracts these structured pieces of information from unstructured text, making it valuable for information extraction, content categorization, and data analysis tasks.

The other answers describe different NLP capabilities:

The second answer describes Sentiment Analysis, which focuses on determining the emotional tone (positive, negative, neutral) and subjective opinions in text rather than identifying specific entities.

The third answer describes Key Phrase Extraction, which identifies the main topics and important phrases that capture the essence of a document, but doesn't classify specific entity types.

The fourth answer describes Intent Recognition (or Intent Classification), which is used in conversational AI to understand what action a user wants to perform, such as booking a flight or checking weather, rather than extracting named entities from text.

The primary function of the business card model in Azure AI Document Intelligence is extracting contact information such as names, phone numbers, email addresses, and company details from business card images.

This is the correct answer because Azure AI Document Intelligence's business card model is specifically designed as a pre-built model that uses optical character recognition (OCR) and machine learning to identify and extract key contact information fields from business card images. The model focuses on the core data elements typically found on business cards: personal names, job titles, company names, phone numbers, email addresses, physical addresses, and websites.

The other answers are incorrect for the following reasons:

The first incorrect option suggests the model analyzes contact hierarchies and organizational relationships. This goes beyond the actual capability of the business card model, which is focused on data extraction rather than relationship analysis or organizational mapping.

The second incorrect option mentions validating authenticity by comparing against corporate identity databases. The business card model does not perform authentication or validation functions - it is purely an extraction tool that reads and structures the information present on the card.

The fourth incorrect option adds extraction of social media profiles, professional certifications, and networking preferences. While the model can extract information that is physically present on a business card, it does not specifically target these extended fields as core functionality, and many business cards don't contain such detailed information. The standard business card model focuses on the traditional contact information fields.