Word2Sense

AbstractWe present an unsupervised method to generate Word2Sense word embeddings that are interpretable — each dimension of the embedding space corresponds to a fine-grained sense, and the non-negative value of …...