AI-102 Question 393
Single answerYou are designing an Azure AI Document Intelligence solution for a company that processes thousands of invoices every month. The invoices come in different formats from various vendors. The company needs to extract key fields such as 'Invoice Number', 'Vendor Name', and 'Total Amount'. Additionally, they want to review and validate the extracted data before it is stored in their database. Which sequence of steps should you follow to implement the solution?
- A
Use the prebuilt Invoice model in Azure AI Document Intelligence to extract fields, then validate the data using a custom application.
- B
Train a custom model in Azure AI Document Intelligence using labeled data, then use a Logic App to validate and store the data.
- C
Use the prebuilt Invoice model in Azure AI Document Intelligence and configure the Review feature to validate the extracted data.
- D
Use Azure Cognitive Search to extract fields from invoices and use Azure Functions to validate and store the data.
Show answer and explanation
Correct answer: C
Explanation
The prebuilt Invoice model in Azure AI Document Intelligence is tailored for extracting data from invoices, making it the most efficient choice for this scenario. To meet the requirement of manual validation, the Review feature allows human reviewers to inspect and correct extracted data before it is stored or processed further. This approach minimizes development effort while ensuring accuracy and compliance with business requirements.
- A. Incorrect.
The prebuilt Invoice model is designed to extract structured data from invoices, but it does not inherently include a review mechanism. You would need to build a separate application for validation, which adds complexity.
- B. Incorrect.
Training a custom model requires significant labeled data and time, which is unnecessary for common invoice fields that the prebuilt Invoice model can already extract. Logic Apps can be used for workflows but aren’t ideal for manual validation.
- C. Correct.
The prebuilt Invoice model in Azure AI Document Intelligence is optimized for extracting common fields from invoices, and you can configure the Review feature in Azure Form Recognizer Studio to validate extracted data before further processing.
- D. Incorrect.
Azure Cognitive Search is not designed for structured data extraction from documents like invoices. It’s primarily used for indexing and searching unstructured or semi-structured text.