Unverified50% confidenceFactExact time
The paper designed an experiment with 9,000 calls to compare representational consensus against paraphrase consensus.
1
Sources
50%
Confidence
Long-term
Relevance
9/23/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe experiment used balanced accuracy as the evaluation metric, covering 8 classification datasets.69% similarUnverifiedThe experiment scale reached 440 runs, with the evaluation criterion being the answer + evidence contract, meaning results must be correct and accompanied by complete, verifiable evidence.68% similarUnverified作者将语义正确性与输出格式合规性分离开来评分，并认为这是整个实验中最有价值的发现66% similarUnverifiedThe article tested three Skills based on a simulated paper project in the knowledge graph domain.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/942852API
curl https://kongchang.com/api/v1/knowledge/claims/942852MCP
get_claim(id=942852)