Algorithm 1 Sampling context algorithm

Input: Network G, node attributes A, parameters $r_1$, $r_2$, q, e and d; Output: $(x_i,x_c,\gamma )$; 1:if$rando{m}_1<r_1$then 2: $\gamma \leftarrow +1$; 3:else 4: $\gamma \leftarrow -1$; 5:end if 6:if$rando{m}_2<r_2$then 7: Uniformly sample a random-walk sequence S of length q; 8: if $\gamma =+1$ then 9:  Under the condition $|i-c|<d$, sample $(x_i,x_c)$ in S; 10: else 11:  Sample $x_c$ in $\mathbb{C}$; 12: end if 13:else 14: if $\gamma =+1$ then 15:  Uniformly sample $(x_i,x_c)$ which satisfies $a_i=a_c$; 16: else 17:  Uniformly sample $(x_i,x_c)$ which satisfies $a_i\ne a_c$; 18: end if 19:end if