Unverified50% confidenceFactExact time
Automated evaluation metrics such as BLEU, ROUGE, and perplexity primarily measure surface-level statistical features and are difficult to capture issues of linguistic naturalness.
1
Sources
50%
Confidence
Long-term
Relevance
8/15/2026
First Seen
Sources
语言学家入局AI：LLM评估背后的人力真相
Related Claims
Unverified困惑度（Perplexity）、BLEU分数、ROUGE分数等自动化指标最初为机器翻译等任务设计，在衡量创意性方面存在天然缺陷73% similarUnverifiedBLEU is commonly used for machine translation evaluation, and ROUGE is widely applied to summarization tasks; both measure text quality by statistically calculating n-gram overlap65% similarUnverifiedBLEU is a natural language generation evaluation metric proposed in the 1990s that measures quality by calculating the n-gram overlap between generated text and reference translations, and remains the standard benchmark in the field of machine translation.61% similarUnverified学术界正在探索基于语义嵌入的相似度检测方案，将文本映射到高维向量空间后计算语义距离以识别AI改写后的内容同源性，但大规模实时部署的计算成本仍是挑战59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/751672API
curl https://kongchang.com/api/v1/knowledge/claims/751672MCP
get_claim(id=751672)