Chain-of-Thought: Retrieval Quality Diagnosis

Diagnose retrieval quality issues in this RAG pipeline step by step.

User Query:
{{query}}

Retrieved Documents:
{{retrieved_docs}}

Generated Answer:
{{llm_response}}

Expected Behavior:
{{expected_outcome}}

Step 1: Analyze query understanding - was the query correctly interpreted?
Step 2: Examine embedding quality - do embeddings capture semantic meaning?
Step 3: Evaluate retrieval relevance - are top-k documents appropriate?
Step 4: Check context utilization - did the LLM use retrieved context?
Step 5: Assess answer quality - does response address the query?
Step 6: Identify failure point - where did the pipeline break down?

Provide specific recommendations for each identified issue.