Question 18
Domain 2: Implement Knowledge Mining and Azure AI Search SolutionsYour Azure AI Search index has 5 million documents. Users need to search using natural language queries like "recent announcements about product recalls" and get the most relevant results — not just keyword matches. The search must be fast and cost-effective. Which configuration achieves this?
Correct answer: B
Explanation
Azure AI Search semantic ranking is designed to “understand the intent and context” of natural language queries and return the “most relevant results” beyond keyword matching. A Standard S1 tier supports semantic ranker at a lower cost than higher tiers, and a semantic configuration lets you define key fields so the ranker can prioritize titles, content, and other important fields.
Why each option is right or wrong
A. Full-text Lucene query with field boosting
B. Semantic ranker on Standard S1 tier with a semantic configuration defining key fields
Azure AI Search semantic ranking is available only on supported Standard tiers, and the S1 tier is the lowest-cost option that still supports it; the service applies semantic ranking to the top 50 results from the initial lexical query, which is why it fits a large 5 million-document index without requiring a higher SKU. A semantic configuration is required to specify the prioritized fields (for example, title, content, and keywords) that the ranker uses when reordering results, so the query can surface the most relevant matches for natural-language intent rather than simple term frequency.
C. Vector search with embeddings for all 5 million documents
D. OData filter with date ranges