Text this: Managing imbalanced data sets in multi-label problems: A case study with the SMOTE algorithm