Unverified50% confidenceBenchmarkExact time
Using only the subject metadata field while discarding the article body, the model achieves F1=1.000, because the subject sets of real and fake news articles are completely disjoint.
1
Sources
50%
Confidence
Long-term
Relevance
9/23/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe F2 score is used instead of F1 in literature screening scenarios because the cost of missing a relevant paper is far greater than the cost of including an irrelevant one.67% similarUnverified低质量的知识库会导致幻觉式引用，即模型看似使用知识库内容实则引用了错误或无关片段61% similarUnverifiedShort content, overly narrow vertical-domain titles, and single-company case studies are content formats that have nearly zero citation rates across the four major platforms.60% similarUnverifiedGrounding prompts that instruct the model to answer only based on provided documents and cite the relevant passages can greatly reduce the probability of fabrication, but cannot eliminate it entirely.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/942801API
curl https://kongchang.com/api/v1/knowledge/claims/942801MCP
get_claim(id=942801)