Unverified50% confidenceSolutionExact time
RAG evaluation should cover four dimensions: the retrieval layer (Recall@k, Precision@k, MRR, nDCG), the generation layer (faithfulness, answer relevance, answer correctness), context utilization (context precision and recall), and the system layer (latency p50/p95, cost per query, abstention rate).
1
Sources
50%
Confidence
Long-term
Relevance
10/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCommonly used metrics for RAG retrieval quality include Recall@k and MRR (Mean Reciprocal Rank).78% similarUnverifiedSetting differentiated weights for documents from different sources in a RAG system can be achieved through two approaches: adjusting score weights during the retrieval stage (Reranking) and explicitly specifying conflict priorities via System Prompt during the generation stage, and these two approaches can be used in combination.71% similarUnverifiedRAG知识库质量的关键指标包括文本切片策略（Chunk Size与Overlap）、Embedding模型选型、检索Top-K设置以及重排序（Reranking）策略69% similarUnverifiedRAG evaluation is a multi-stage pipeline of "retrieval → ranking → generation": the retrieval stage is measured by recall and precision, while the generation stage is measured by metrics such as Faithfulness.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/998523API
curl https://kongchang.com/api/v1/knowledge/claims/998523MCP
get_claim(id=998523)