Algorithm 2 Model training

Require: Attributed networks $G^s$ and $G^t$, parameters ${\lambda }_1$ and ${\lambda }_2$, batch sizes $K_1$, $K_2$ and $K_3$;

1: for $it=1\to \mathit{Max}\_\mathit{It}$ do 2: Sample a group of node pairs of size $K_1$ in the anchor links set T; 3: Let ${\mathbb{R}}_1=-\frac{1}{K_1}{\sum }_{(x_l^s,x_n^t)}\log P(x_l^s,x_n^t)$, perform a gradient calculation on ${\mathbb{R}}_1$; 4: Sample a group of contexts $(x_i^s,x_c^s,\gamma )$ of size $K_2$ in the source network $G^s$; 5: Let ${\mathbb{R}}_2=-\frac{{\lambda }_1}{K_2}{\sum }_{(x_i^s,x_c^s,\gamma )}\log \sigma \left(\gamma {{\mathbf{w}}_c^s}^T{\mathbf{e}}_i^s\right)$, perform a gradient calculation on ${\mathbb{R}}_2$; 6: Sample a group of contexts $(x_i^t,x_c^t,\gamma )$ of size $K_3$ in the target network $G^t$; 7: Let ${\mathbb{R}}_3=-\frac{{\lambda }_2}{K_3}{\sum }_{(x_i^t,x_c^t,\gamma )}\log \sigma \left(\gamma {{\mathbf{w}}_c^t}^T{\mathbf{e}}_i^t\right)$, perform a gradient calculation on ${\mathbb{R}}_3$. 8: end for