README.md

GraphEmbedding

Implement of several Graph Embedding methods，including Homogeneous and Heterougeneous Graph.

Methods

Method	Paper
Deepwalk	DeepWalk: Online Learning of Social Representations
LINE	LINE: Large-scale Information Network Embedding
Node2vec	node2vec: Scalable Feature Learning for Networks
GCN	Semi-Supervised Classification with Graph Convolutional Networks
GAT	Graph Attention Networks
Metapath2vec	Metapath2vec: Scalable Representation Learning for Heterogeneous Networks
HAN	Heterogeneous Graph Attention Network
GATNE	Representation Learning for Attributed Multiplex Heterogeneous Network

Environment

python
networkx
dgl
pytorch
tensorflow

Input

Supervised learning

We use DBLP dataset and set CrossEntropy loss.

We get node embeddings from model and use LR as classifier.

Un-Supervised learning

We use Amazon dataset to realize link prediction task, using NCE loss.