AI-103: Developing AI Apps and Agents on AzureDemo

By Microsoft · Question Mode

QuestionQ57

Implement information extraction solutions

You have a Microsoft Foundry project that contains an agent.

The agent uses Azure AI Search as its retriever. You plan to ingest PDFs into an Azure AI Search index so the agent can ground its responses in text from both the documents and their embedded images. Users require citations that link back to the source files.

You need to ensure that, during indexing, the images are extracted into a structure that the built-in optical character recognition (OCR) skill can use as input.

Which indexing approach should you use?

  • A an indexer to extract image data into a normalized_images collection
  • B a Shaper skill to restructure the OCR input
  • C a skillset to run the OCR skill directly against the content field of the index
  • D the outputFieldMappings parameter to write image data to a searchable field
Explanation

In Azure AI Search, setting an indexer's imageAction to generateNormalizedImages during document cracking extracts embedded images from source files (such as PDFs) into a normalized_images field/collection. Skills that operate on images, including the built-in OCR skill, expect this normalized_images collection as their input context ("/document/normalized_images/*"); images cannot be fed to OCR without first being normalized this way. A Shaper skill only restructures existing enrichment output, running OCR directly against the content field is invalid because OCR requires image input rather than the document's text content, and outputFieldMappings maps enrichment results to searchable index fields but does not extract images in the first place.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!