The document summarizes a paper presented at NIPS 2016 titled "Supervised Word Mover's Distance" by Huang et al. that extends the Word Mover's Distance (WMD) to incorporate supervision. WMD measures document distance based on word embeddings, and the supervised version improves its ability to capture semantic relationships through supervision in training word embeddings. The document also references related work on word2vec, Earth Mover's Distance, Neighborhood Component Analysis, and Word Mover's Distance.