*5.1. Imbalanced Multi-Label Classification Performance*

Tables 2 and 3 show the performance of all the methods in terms of Micro-F1 and Macro-F1. The results are presented as the mean of ten repeated experiments. Based on these results, we reached the following conclusions:


of the two strategies works the best. As supporting evidence, **SORAG***F* is the best performer in 5/6 tasks and the second-best performer in the remaining task.

**Table 2.** Imbalanced multi-label classification comparison in terms of Micro-F1. The first and second best results are boldfaced and underscored, respectively.


**Table 3.** Imbalanced multi-label classification comparison in terms of Macro-F1. The first and second best results are boldfaced and underscored, respectively.

