Word2Vec is a 2013 method for learning word embeddings: lists of numbers for each word, so that words used in similar ways get similar numbers.
In plain words
Word2Vec learns from plain text by guessing words from their neighbours. The continuous bag-of-words version predicts a missing word from the words around it; the skip-gram version does the reverse. After training on billions of words, the vectors capture relationships, so similar words end up close together. It was fast: good vectors from 1.6 billion words took under a day.
A quick example
Imagine training Word2Vec on lots of news text, then looking up the vector for "Paris". Its nearest neighbours will tend to be other cities and words about France, because they appear in similar sentences.
Where you'll meet it
- Early NLP systems for search, recommendations and similarity.
- Teaching how embeddings capture meaning.
Related
Sources
- PaperEfficient Estimation of Word Representations in Vector Space, Mikolov et al. (arXiv) · read 29 Sept 2026
- PaperDistributed Representations of Words and Phrases and their Compositionality, Mikolov et al. (arXiv) · read 29 Sept 2026
- Docsword2vec, TensorFlow tutorials · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.