Text this: Improving clustering-based and adaptive position-aware interpolation oversampling for imbalanced data classification