Unverified50% confidenceFactExact time
The study introduces a new query-side faithfulness metric, HETR (Hallucinated Expansion Term Rate), which counts the number of injected expansion terms that have no support in any relevant document.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedROUGE metrics exhibit length bias: models that generate shorter summaries have an advantage in precision, models that generate longer summaries have an advantage in recall, and ROUGE cannot measure factual consistency.65% similarUnverified当查询词较短而候选词较长时，重叠系数可能比 Jaccard 更合适，因为它不会因候选词的额外 N-gram 而过度惩罚相似度分数59% similarUnverifiedFor short messages or partially incomplete Enigma ciphertexts, the sample available for statistical analysis is too small, and a large number of spurious solutions appear in the search space, making it difficult to identify the original plaintext.59% similarUnverifiedResearch shows that hallucination rate is negatively correlated with the knowledge density of the prompt; the richer and more precise the context provided in the prompt, the lower the probability that the model will generate fabricated content.59% similarUnverifiedMention-level precision and recall of around 70% means that real-world large-scale applications will still produce a considerable proportion of missed detections and false positives, making it difficult to eliminate manual review in the short term.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1002087API
curl https://kongchang.com/api/v1/knowledge/claims/1002087MCP
get_claim(id=1002087)