Unverified50% confidenceFactExact time
原始word2vec需要在整个词汇表上计算softmax归一化，计算复杂度为O(V)，负采样将问题转化为二分类以降低计算量
1
Sources
50%
Confidence
Long-term
Relevance
8/13/2026
First Seen
Sources
Related Claims
UnverifiedWord2Vec compresses words into dense vectors of 100–300 dimensions via two training approaches: CBOW or Skip-gram.70% similarUnverifiedCollapse采用整词词表，无法泛化到未登录词（OOV），且每词单向量只能捕捉主导词义，多义词处理能力有限68% similarUnverifiedEmbedding models map text to high-dimensional dense vectors (typically 768 to 3072 dimensions), and ANN algorithms (such as HNSW and IVF-PQ) reduce retrieval complexity from O(n) to approximately O(log n).67% similarUnverifiedGlobal self-attention has a modeling complexity of O(n²), and subsequent research has developed relative positional encoding schemes such as RoPE and ALiBi to enhance long-text processing capability.67% similarUnverifiedword2vec是静态词向量方法，给每个词一个固定的向量，无论它出现在什么句子里66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/740926API
curl https://kongchang.com/api/v1/knowledge/claims/740926MCP
get_claim(id=740926)