AI-103: Developing AI Apps and Agents on Azure

By Microsoft · Question Mode

QuestionQ81

VerifiedImplement generative AI and agentic solutions

You have a Microsoft Foundry project containing an agent. The agent uses Azure AI Search for Retrieval Augmented Generation (RAG).

You plan to ingest and index PDF product manuals.

You need to design a solution that supports semantic similarity matching, ensuring the agent retrieves relevant data even when user questions are phrased differently than the wording used in the product manuals.

Which indexing approach should you use?

Explanation

Vector search in Azure AI Search indexes content as high-dimensional embeddings that capture semantic meaning rather than exact text. This allows queries phrased differently from the source documents to still match relevant content based on conceptual/semantic similarity, which is essential for RAG scenarios where user questions may not use the same terminology as the underlying product manuals. Semantic ranking only re-ranks results from an initial keyword or hybrid search and does not itself generate the embeddings needed for similarity matching; suggesters provide autocomplete functionality; and analyzers process text for lexical (keyword) search, not semantic understanding.

Learn more

Community Discussion

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