Sentence Embedding

通用型单句Embedding

• 词向量平均
• 词向量乘上词权重后进行累加
• 综合考虑词频、词向量矩阵分解
• http://www.hankcs.com/nlp/cs224n-sentence-embeddings.html
• 类似word2vec训练sentence2vec
• Distributed Representations of Sentences and Documents
• 实现： https://rare-technologies.com/doc2vec-tutorial/
• 如果句子中有数字
  • http://www.cse.lehigh.edu/~brian/pubs/2018/bigcomp/NumericSentenceEmbedding.pdf
• Sequential (Denoising) Autoencoders (SDAE)
  • 在句子中加入噪声后作为数据，原句子作为target，构建seq2seq模型
  • Learning Distributed Representations of Sentences from Unlabelled Data
  • An Overview of Sentence Embedding Methods

场景相关Embedding

• seq2seq， RNN  encoder-decoder，机器翻译的训练语料是sentence pair（Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation）
• 搜索引擎的结果点击，也能形成sentence-pair
  • Deep Sentence Embedding Using Long Short-Term Memory Networks
•  self-attention + bi-lstm
  • 形成一个sentence embedding matrix，矩阵的每一行是对句子的一个语义的representation
  • A structured self-attentive sentence embedding  https://github.com/kaushalshetty/Structured-Self-Attention  https://github.com/jx00109/structured-self-attentive-sentence-embedding