Unverified50% confidenceFactExact time
DRACO's evaluation dimensions typically include factual accuracy, citation completeness, information coverage and the soundness of the reasoning chain.
1
Sources
50%
Confidence
Long-term
Relevance
11/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA useful decision benchmark should simultaneously report both accuracy and calibration, ensure that evaluation questions match deployment questions, and be able to distinguish strong from weak systems.71% similarUnverifiedA modular fact-checking pipeline can simultaneously improve both prediction accuracy and confidence reliability in open-domain verification.70% similarUnverifiedResult validation requires establishing objective quantitative standards—such as classification accuracy, reply compliance, and manual correction ratio—and calculating statistics over a sufficient sample size.69% similarUnverifiedRecall@k and MRR are commonly used to evaluate retrieval quality, while evaluating generation typically looks at the consistency between answers and sources, i.e., faithfulness.69% similarUnverifiedThe document requires preserving the distinction between "learned information, retrieved evidence and generated inference" wherever it affects the reliability of an answer.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1126768API
curl https://kongchang.com/api/v1/knowledge/claims/1126768MCP
get_claim(id=1126768)