This document provides an overview of a tutorial on graph representation learning for recommender systems. The tutorial covers embedding nodes in homogeneous graphs using random walk-based approaches like DeepWalk and node2vec. It also discusses higher-order embedding methods like LINE, which directly model graph properties, and GraRep, which represents the probability of k-step random walks. The graph embeddings can be used for tasks like entity retrieval and classification and as inputs to recommender system models.