Efficient Estimation of Word Representations in Vector Space .

Abstract

Word2Vec learns dense vector representations of words where semantic similarity is captured by vector distance.

Models

• CBOW: Predicts word from context
• Skip-gram: Predicts context from word

Magic

Vector arithmetic works:

king - man + woman ≈ queen
Paris - France + Italy ≈ Rome

Impact

Made word embeddings practical and ubiquitous. Foundation for modern NLP before Transformers.