A vector space model (VSM) takes words and transforms them into a vector space. In this continuous space, similar words are represented in points near each other. This vectorial mapping process is called embedding.
Word2vec takes raw text, finds the words nearest to each other, and maps them into vectors. The system will take words, build them into a dictionary, and then input indexes.
A skip-gram model is applied to the sequences with a skip window. In that skip window, a number of words are taken into consideration on the left and the right. Then the model skips, and it will reuse and input several times to produce a label as implemented in embedding.py. The embedding size, skip window, and the number of skips are initialized ...
