*4.2. Analyzed Methods*

To validate the performance of our approach, we compared it with several state-of-theart and representative methods for multilabel graph learning and imbalanced graph learning, including **GCN** [12], **ML-GCN** [5], **SMOTE** [14], **GraphSMOTE** [17], and **RECT** [49]. The analyzed baseline methods are briefly introduced as follows:


In addition, three variants of the proposed method were implemented:


It is necessary to mention that all the baselines above, except **ML-GCN** (which is intrinsically designed as a multi-label classifier), are manually set to conduct multilabel node classification by modifying the last layer of their network structure. The implementation of the baseline approach relies on publicly released code from relevant

sources (**SMOTE**: https://github.com/analyticalmindsltd/smote\_variants (accessed on 25 June 2022), **GraphSmote**: https://github.com/TianxiangZhao/GraphSmote (accessed on 25 June 2022), **RECT**: https://github.com/zhengwang100/RECT (accessed on 25 June 2022), **GCN**: https://github.com/tkipf/pygcn) (accessed on 25 June 2022).
