Table 5 Analysis of ablation experiment results.
From: Information flow optimization for adaptive neighbor selection graph embedding
Dataset | Metric | \(GF_R\) | \(GF_L\) | \(GF_{NOL}\) | \(GF_{NOH}\) | \(\text{GF}_{\text{AVT}}\) | \(GF_{NONH}\) | GF |
|---|---|---|---|---|---|---|---|---|
AMiner | Macro-F1 | 0.863 | 0.885 | 0.872 | 0.857 | 0.854 | 0.866 | 0.903 |
Micro-F1 | 0.864 | 0.883 | 0.874 | 0.853 | 0.876 | 0.865 | 0.908 | |
ROC-AUC | 0.785 | 0.793 | 0.793 | 0.772 | 0.792 | 0.790 | 0.811 | |
F1 | 0.784 | 0.782 | 0.784 | 0.745 | 0.775 | 0.782 | 0.786 | |
DBLP | Macro-F1 | 0.883 | 0.894 | 0.913 | 0.892 | 0.903 | 0.915 | 0.945 |
Micro-F1 | 0.882 | 0.892 | 0.926 | 0.897 | 0.915 | 0.928 | 0.952 | |
ROC-AUC | 0.854 | 0.881 | 0.870 | 0.844 | 0.875 | 0.874 | 0.899 | |
F1 | 0.856 | 0.875 | 0.877 | 0.836 | 0.882 | 0.861 | 0.899 | |
IMDB | Macro-F1 | 0.762 | 0.773 | 0.765 | 0.753 | 0.762 | 0.771 | 0.784 |
Micro-F1 | 0.748 | 0.787 | 0.774 | 0.746 | 0.781 | 0.775 | 0.796 | |
ROC-AUC | 0.924 | 0.932 | 0.922 | 0.914 | 0.925 | 0.932 | 0.960 | |
F1 | 0.935 | 0.937 | 0.923 | 0.919 | 0.938 | 0.935 | 0.954 |