Enhanced Unsupervised Graph Embedding via Hierarchical Graph Convolution Network

Table 3

Node classification results on Wikipedia.

	Baselines	Micro-F1	Macro-F1
Random walk	Deepwalk	0.669	0.560
	LINE(2nd)	0.576	0.387
	Node2vec	0.651	0.541
Matrix decomposition	GraRep	0.633	0.476
	Hope	0.601	0.438
Deep model	SDNE	0.643	0.498
Our model	HGCN	0.599	0.549
	HGCN	0.717	0.590

The result with is from the open-source OpenNE [39].