Table 2.

A summary of methods

Category	Method	Preserved proximity	Time complexity	S	Learning model
Common neighbor based	CN [92], Salton [25], JI [26], Sorens [27], HPI [28], HDI [29], LLHN [30], PA [31], LNBCN [32]	$1_{\mathit{st}}$ order	$O(d_m^{2}n)$ $\sim$ $O(d_m^{3}n)$	$✓$	Unsupervised
	AA [13], RA [29], LNBAA [32], LNBRA [32]	$2_{\mathit{nd}}$ order	$O(d_m^{3}n)$	$✓$	Unsupervised
	TSCN [33]	$k_{\mathit{th}}$ order	$O(n^3)$	$\times$	Unsupervised
Path Based	LPI [3]	$2_{\mathit{nd}}$ $\sim$ $3_{\mathit{rd}}$ order	$O(d_m{n}^2)$	$✓$	Unsupervised
	KI [35], GLHN [30], ACT [37], RWR [38], SR [39], MFI [40]	$k_{\mathit{th}}$ order	$O(n^3)$	$\times$	Unsupervised
	LRW [37], SRW [37]	$l_{\mathit{th}}$ order	$O(l{d}_m{n}^2)$	$✓$	Unsupervised
Probabilistic and statistical models based	SBM [41]	$k_{\mathit{th}}$ order	–	$\times$	Supervised
Classifier based	SVM [51], KNN [52], DT [53], Bayes [54], LR [55], MLP [56]	$1_{\mathit{st}}$ $\sim$ $2_{\mathit{nd}}$ order	$O(d_m^{3}n)$ $\sim$ $O(n^2)$	$✓$	Supervised
Network embedding based	MF [67]	$1_{\mathit{st}}$ $\sim$ $2_{\mathit{nd}}$ order	$O(n^3)$	$✓$	Supervised
	GraRep [69]	$2_{\mathit{nd}}$ $\sim$ $k_{\mathit{th}}$ order	$O(mn+d_i{n}^2)$	$✓$	Supervised
	DeepWalk [22]	$2_{\mathit{nd}}$ $\sim$ $k_{\mathit{th}}$ order	$O(d_{i}nlogn)$	$✓$	Unsupervised
	Node2vec [72]	$2_{\mathit{nd}}$ $\sim$ $k_{\mathit{th}}$ order	$O(d_{i}rn)$	$✓$	Semi-supervised
	Struc2vec [73]	Structural Identity	$O(n^3)$	$✓$	Unsupervised
	UniNet [74]	$1_{\mathit{st}}$ $\sim$ $k_{\mathit{th}}$ order	–	$✓$	Semi-supervised
	GCN [76]	$1_{\mathit{st}}$ $\sim$ $k_{\mathit{th}}$ order	$O(d_{i}m+d_i^{2}n)$	$\times$	Semi-supervised
	GraphSAGE [77]	$1_{\mathit{st}}$ $\sim$ $k_{\mathit{th}}$ order	$O(d_i^2{r}^{L_n}n)$	$✓$	Unsupervised
	WLNM [78]	$1_{\mathit{st}}$ $\sim$ $k_{\mathit{th}}$ order	–	$✓$	Supervised
	DGCNN [79]	$1_{\mathit{st}}$ $\sim$ $k_{\mathit{th}}$ order	–	$✓$	Semi-supervised
	SEAL [80]	$1_{\mathit{st}}$ $\sim$ $2_{\mathit{nd}}$ order	–	$✓$	Semi-supervised
	Cluster-GCN [81]	$1_{\mathit{st}}$ $\sim$ $k_{\mathit{th}}$ order	$O(d_{i}m+d_i^{2}n)$	$✓$	Semi-supervised
	LINE [86]	$1_{\mathit{st}}$ $\sim$ $2_{\mathit{nd}}$ order	$O(d_{i}m)$	$✓$	Supervised
	SDNE [89]	$1_{\mathit{st}}$ $\sim$ $2_{\mathit{nd}}$ order	O(mn)	$✓$	Semi-supervised
	VERSE [91]	$1_{\mathit{st}}$ $\sim$ $2_{\mathit{nd}}$ order	$O(d_{i}rn)$	$✓$	Semi-supervised

Let $d_m$ denotes the maximum degree of a network, l denotes the number of the random walk steps. For embedding approaches, $d_i$ denotes the dimensionality of embedding vector, $L_n$ is number of layers, r is the number of sampled neighbors per node