AI-102 Question 422
Single answerYou are designing a custom Azure AI Search skill to extract structured data from invoices stored in an Azure Blob Storage container. You plan to use Azure Form Recognizer to analyze the invoice documents and integrate it as part of your custom skillset in Azure Cognitive Search. What step is essential to implement this document intelligence model as a custom skill in the search pipeline?
- A
Define and deploy an Azure Function that acts as the custom skill, calling Azure Form Recognizer to process the invoices.
- B
Configure Azure Cognitive Search to directly call Azure Form Recognizer without a custom skill.
- C
Set up a Key Vault to store the Azure Form Recognizer API keys and automatically link it to the search index.
- D
Write a custom REST API to preprocess the invoice data before indexing it in Azure Cognitive Search.
Show answer and explanation
Correct answer: A
Explanation
To use Azure Form Recognizer as part of an Azure Cognitive Search skillset, you need to implement a custom skill that calls the Form Recognizer API. The most common way to achieve this is by defining and deploying an Azure Function that acts as the custom skill. This function processes the input documents, calls the Form Recognizer API, and returns the extracted data for use in the search index. Native integration between Azure Cognitive Search and Form Recognizer is not available, so custom implementation is required.
- A. Correct.
This is correct. To integrate Azure Form Recognizer into Azure Cognitive Search as a custom skill, you must define an Azure Function (or equivalent custom skill) that calls the Form Recognizer API and processes the results.
- B. Incorrect.
This is incorrect. Azure Cognitive Search does not natively integrate with Azure Form Recognizer. A custom skill is required to bridge this functionality.
- C. Incorrect.
This is incorrect. While storing API keys securely in a Key Vault is a good practice, it is not the essential step to implement the document intelligence model as a custom skill.
- D. Incorrect.
This is incorrect. A custom REST API for preprocessing is not necessary as Azure Cognitive Search skillsets handle this functionality through defined skill pipelines.