Word Embedding

w# Word Embedding

A Word Embedding is a mapping from words in context into a real valued vector space ${\mathbb{R}}^n$ used for NLP.

Also see One-Hot-Encoding.

Uses the Tokenization of the text.

• TF-IDF

• POS Tagging example

• Recurrent Neural Network

• Bidirectional Recurrent Neural Network

• LSTMs can better learn grammatical dependencies

Semantic Word Embedding

Word embeddings that take context into account by placing vectors in the vectorspace in such a way that they show relations among words. That way you can even encode word relations through vector differences

• Word2vec
  • Trained with Common Bag Of Words
• GloVe
• fastText (multilingual)