AI-102 Question 421
Select 4You are designing a custom Azure AI Search solution to extract and process data from scanned invoices. You want to implement a document intelligence model as a custom skill in your Azure Cognitive Search pipeline. Which steps should you include to correctly configure and integrate the document intelligence model?
- A
Create and train a Form Recognizer model to extract relevant data fields from the invoices.
- B
Deploy the Form Recognizer model to an Azure Function and expose it as a REST API endpoint.
- C
Add the Form Recognizer REST API as a custom skill in the Azure Cognitive Search skillset configuration.
- D
Use the Azure AI Search built-in OCR skill to process the invoices directly without requiring any additional models.
- E
Define a skillset in Cognitive Search that includes the custom Form Recognizer skill and a knowledge store for storing the enriched data.
Show answer and explanation
Correct answers: A, B, C, E
Explanation
To implement a document intelligence model as a custom Azure AI Search skill, you need to train a Form Recognizer model to extract structured data from documents, deploy it as a REST API, and configure it as a custom skill in an Azure Cognitive Search skillset. This ensures the search pipeline can process the data effectively, extract relevant fields, and optionally store the output in a knowledge store for further analysis. Using only the built-in OCR skill would not meet the requirements for extracting detailed, structured information from invoices.
- A. Correct.
Correct: Creating and training a Form Recognizer model is required to extract specific data fields (e.g., invoice number, total amount) from the invoices.
- B. Correct.
Correct: Deploying the Form Recognizer model as a REST API is necessary to enable Cognitive Search to call the model as part of the pipeline.
- C. Correct.
Correct: Adding the Form Recognizer REST API as a custom skill in the skillset configuration allows Cognitive Search to leverage the model during indexing.
- D. Incorrect.
Incorrect: The built-in OCR skill is not sufficient for extracting structured data like invoice details; a custom model such as Form Recognizer is needed.
- E. Correct.
Correct: A skillset needs to be defined to include the custom skill (Form Recognizer) and optionally store enriched data in a knowledge store for additional analysis.