I am trying to formalize some Natural language processing (NLP) model(s) more mathematically, such as Word2vec.