Unverified50% confidenceFactExact time
Sentence-BERT通过孪生网络结构将句子映射为固定维度的稠密向量，专为句子级语义相似度优化
1
Sources
50%
Confidence
Long-term
Relevance
7/6/2026
First Seen
Sources
Content Mismatch Detected — Article Cannot Be Generated
bilibiliyasoven7/3/2026
Related Claims
Unverifiedsentence-transformers库在BERT基础上引入了孪生网络训练范式，使输出向量在余弦相似度空间中具备良好的语义对齐性79% similarUnverifiedSynonym normalization can be achieved by using sentence embedding models such as Sentence-BERT to map words into dense vectors, and then computing semantic proximity via cosine similarity.78% similarUnverified语义分块基于句子嵌入的相似度变化检测段落边界，可避免固定长度分块导致的语义断裂74% similarUnverified神经网络架构中，每个词语、句子乃至概念都被映射为高维向量空间中的一个点或区域69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/114030API
curl https://kongchang.com/api/v1/knowledge/claims/114030MCP
get_claim(id=114030)