sense2vec refers to a system in a paper titled sense2vec - A Fast and Accurate Method for Word Sense Disambiguation In Neural Word Embeddings. It solves a problem with previous word embeddings like word2vec and GloVe where words of different senses (e.g. “duck” as an animal and “duck” as a verb) are represented by the same embedding.

sense2vec uses word sense information to train more accurate word embeddings.