Question 840
An Azure AI Search semantic ranker runs after a vector query, but the index stores only numeric vectors and no searchable descriptive text. The ranker cannot build meaningful semantic summaries. Which index design should be reviewed?
Explanation
Correct Answer
Azure AI Search semantic ranking consumes text fields from the candidate result set. Numeric embeddings without suitable readable content cannot supply useful extractive summaries.
Incorrect Answers
Vector dimensionality does not supply the missing textual content that the semantic reranker analyzes.
Vector similarity is a different retrieval stage and does not itself produce the semantic captions the application requires.
The semantic ranker operates inside Search before downstream generation and cannot rely on fabricated captions.