Word2vec

Word2vec is a method to embed items (originally words) in a vector spaces so that similar ones will end up close to each other. The definition of "similar" is a mixture of syntagmatic and paradigmatics, i.e. a similar item is one that co-occurs with another or one that can replace another without changing meaning of the sequence (e.g. sentence) in which it appears.

Applications
Pinterest reports usage of neural networks "inspired by word2vec" that works better than "board co-occurence" ("Pin X is saved to N boards by many people. Related Pins for Pin X are generated from all the other Pins saved to the N boards").