node2vec: Scalable Feature Learning for Networks

本文借鉴word2vec提出了node2vec,通过maximize the likelihood of preserving network neighborhoods of nodes in a d-dimensional feature space得到特征表示。利用二阶随机游走产生节点社区。