The document describes Meta-Prod2Vec, a method for embedding products that leverages both co-occurrence information from user sessions as well as side information about products like categories and brands. It improves upon Prod2Vec, which learns embeddings from co-occurrence data alone, by incorporating side information to help address cold start problems when little co-occurrence data exists. Met
![Meta-Prod2Vec: Simple Product Embeddings with Side-Information](https://cdn-ak-scissors.b.st-hatena.com/image/square/72b73f37823781c9728ffa10419279621ca06f12/height=288;version=1;width=512/https%3A%2F%2Fcdn.slidesharecdn.com%2Fss_thumbnails%2Fmetaprod2vec-161007133531-thumbnail.jpg%3Fwidth%3D640%26height%3D640%26fit%3Dbounds)