Word2Vec

Two-layer neural network to convert words into higher-dimensional vectors

• Words that share common contexts in the corpus are located close to each other in the vector space, capturing semantic relations with vector operations
• Google has a pretrained Word2vec using the Google News corpus. We can also pretrain context-specific Word2vec on custom text

Another option is GloVE. Both are unsupervised/semi-supervised approaches to embedding, relying on the Distributional hypothesis of linguistics.