Efficient Estimation of Word Representations in Vector Space (word2vec, 2013): words become vectors, and king − man + woman = queen

Mikolov, Chen, Corrado and Dean (Google) introduce the skip-gram and CBOW models that learn dense word embeddings from billions of words in hours, capturing semantic and syntactic analogies. Embeddings became the substrate for search, recommendation and every language model that followed. 12 pages.

Companies in this report

Google