Researchers have proposed a new method for improving requirements classification in software engineering, specifically addressing class imbalance issues in datasets. The SMOTE-Tomek preprocessing technique is combined with stratified K-fold cross-validation to enhance the representation of minority classes and improve classification accuracy. In experiments using the PROMISE dataset, logistic regression achieved an accuracy rate of 76.16%, significantly outperforming a baseline model. --- Why it matters: This work matters to researchers in AI because it demonstrates the potential for machine learning models to address class imbalance issues in datasets, leading to improved classification accuracy and more interpretable results. Source: https://arxiv.org/abs/2501.06491