Abstract—This study focuses on advancing the classification of software requirements, particularly within the subclasses of Non-Functional requirements. Four machine learning algorithms—Support Vector Machine (SVM), Logistic Regression (LR), K-Nearest Neighbors (KNN), and Naive Bayes (NB)—were initially implemented, with SVM exhibiting superior performance. To enhance overall accuracy, a voting classifier ensemble method was employed, resulting in significant improvement. In the realm of deep learning, a standalone Bidirectional Long Short-Term Memory (Bi-LSTM) network faced challenges in fine-tuning. To harness the strengths of both machine learning and deep learning, we proposed a hybrid model by integrating SVM with Bi-LSTM. This hybrid model surpassed all prior experiments, highlighting the synergistic potential between traditional machine learning and deep learning. Our findings showcase the effectiveness of combining SVM’s discriminative power with Bi-LSTM’s sequential understanding, yielding a robust classification model for software requirements. This research contributes to the advancement of requirement analysis, providing a practical solution for accurately identifying diverse requirement types within the nuanced domain of Non-Functional requirements.
 
Keywords—Software requirements, Bidirectional Long Short-Term Memory (Bi-LSTM), Support Vector Machine (SVM)


Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.