Partially Verified65% confidenceFactTime unknown
传统RAG采用固定的单向流程（一次检索、一次生成），无法自动重试或换一种方式检索
3
Sources
65%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTraditional RAG systems can only passively invoke pre-loaded theoretical knowledge and cannot actively perform real operations.77% similarUnverified标准RAG管道通常是单轮、单源的静态检索，缺乏迭代深化、跨源验证和批判性评估的能力72% similarUnverified传统RAG的固定流水线架构存在两个明显问题：无论是否需要额外上下文都会执行检索，以及只检索一次无法处理多跳推理71% similarUnverifiedTraditional RAG systems have three major limitations: returning empty results when the answer cannot be retrieved, inability to answer metadata questions, and lack of multi-turn iterative error correction.70% similarUnverifiedTraditional RAG (Retrieval-Augmented Generation) approaches suffer from indexing lag in large-scale active codebases, potentially retrieving deleted or outdated functions.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/26769API
curl https://kongchang.com/api/v1/knowledge/claims/26769MCP
get_claim(id=26769)